# Understanding the Digital Ecosystem
Findings from the 2019 Federal Election
# About this Report
This report is the result of a collaborative effort that brought together academics from across Canada and abroad to research the impact of digital media on Canadian elections. Here you will find the initial results of 18 different research projects, which range in topic from online political advertising, to online harassment, to disinformation, and more. This is a product of the Digital Ecosystem Research Challenge.
# About the Digital Ecosystem Research Challenge
Dr. Elizabeth Dubois at the University of Ottawa and Dr. Taylor Owen at McGill University launched the Digital Ecosystem Research Challenge in April, 2019, in order to examine the impacts of digital media on Canadian elections. Through the Challenge, 18 winning projects collaborated to analyze a large scale dataset collected from social media, online news, web, and survey data. In creating this collaboration, the goal was to map the digital ecosystem in order to support increased civic and digital literacy.
# About the Editors
### Dr. Elizabeth Dubois
Dr. Elizabeth Dubois (PhD, University of Oxford) is an Assistant Professor in the Department of Communication and member of the Centre for Law, Technology and Society at the University of Ottawa. She is also a Board
Member for CIVIX, a Fellow at the Public Policy Forum of Canada, and a former member of Assembly based at Harvard's Berkman-Klein Center. Her work examines political uses of digital media including media manipulation, citizen engagement, and artificial intelligence. She leads a multi-disciplinary team, which includes political scientists, computer scientists, and communication students. Her work can be found at **www.elizabethdubois.ca** and on Twitter: **@lizdubois**.
#### Dr. Taylor Owen
Taylor Owen is the Beaverbrook Chair in Media, Ethics and Communications and an Associate Professor in the Max Bell School of Public Policy at McGill University. He was previously an Assistant Professor of Digital Media
and Global Affairs at the University of British Columbia and the Research Director of the Tow Center for Digital Journalism at Columbia University. He is currently a Senior Fellow at the Center for International Governance Innovation (CIGI), and a Fellow at the Public Policy Forum. He also sits on the Governing Council of the Social Sciences and Humanities Research Council (SSHRC). His work can be found at **www.taylorowen. com** and on Twitter: **@taylor\_owen**.
# Acknowledgments
We would like to thank each of the 18 research teams that contributed to the Digital Ecosystem Research Challenge collaboration and to this report. Our timeline was extremely tight and we are very grateful for their commitment. Their efforts and hard work are evident in the studies they have conducted, and we are indebted to them for their diligence to this collaboration. We would also like to thank our dedicated research assistants and research fellows, Ève Grébert, Catherine Ouellet, Dan Malone, Pascal Dangoisse, Charlee Heath, Aengus Bridgman, Oleg Zhilin, Maya Fernandez, and Guillermo Renna, for their work and commitment to this project. Lastly, we would like to thank the Centre for Law, Society and Technology for their support and important contributions.
This work was funded in part by the Government of Canada.
# CONTENTS
| Executive Summary | 06 | |
|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|----|--|
| Introduction | 10 | |
| Methods and Data | 12 | |
| Political Advertisement
and Personal Data | 14 | |
| Facebook Advertising During
the Canadian Federal Election
Laura Edelson, Divam Jain, Damon McCoy | 16 | |
| Who Targets Me? Political Advertising on
Facebook in the 2019 Election
Sam Andrey, Karim Bardeesy, Sam Jeffers,
Justin Savoie, Clifton van der Linden | 20 | |
| Political Parties and Data Privacy
Sara Bannerman, Nicole Goodman, Julia Kalinina, Jenny Zhan | 24 | |
| Online Harassment and Hate Speech | 28 | |
| Toxic Interactions and Political Engagement on Twitter
Anatoliy Gruzd, Philip Mai, Raquel Recuero, Felipe Soares | 30 | |
| Trolling on the Campaign Trail: How Candidates
Experience and Respond to Online Abuse
Heidi Tworek, Chris Tenove, Trevor Deley, Grace Lore | 34 | |
| Rise of Islamophobic Emotional Rhetoric during the
2019 Canadian Federal Election
Zeinab Farokhi | 38 | |
| Public Opinion and Political Participation | 42 | |
| Memes in the 2019 Canadian Elections. That's it.
That's the Headline.
Fenwick McKelvey, Jeremy Blackburn, John F. Gray,
Scott DeJong, Janna Frenzel, Quinn Valencourt | 44 | |
| Say it on the 'gram: Political Engagement on Instagram
during the 2019 CanadianFederal Election Campaign
Sabrina Wilkinson | 48 | |
| The Quality of Online News and
Citizens' Perceptions of Pre-Election Polls | 52 | |
Jean-François Daoust, Frédérick Bastien
# [Flow of Misinformation](#page-55-0) 56 and Disinformation
58 **Fringe Politics: The Deep Web's Impact on the 2019 Canadian Election**
> [Greg Elmer, Marc Tuters, Anthony Burton, Melody Devries,](#page-57-0) Ganaele Langlois, Stephen J. Neville, Sabrina Ward-Kimola
- 62 **Affective Media, Social Movements, and [Digital Dissent: Emotions and Democratic](#page-61-0) Participation in the 'Post-Truth' Era** Megan Boler
- 66 **[Media Manipulation, Memetic Campaigns,](#page-65-0) and the 2019 Canadian Elections**
Joan Donovan, Brian Friedberg, Nicole Leaver, Jon Penney
# 72 [Issue Framing and Agenda Setting](#page-71-0)
74 **[Climate Change, Social Media, and Political Parties](#page-73-0)**
Shelley Boulianne, Stephanie Belland, Anders Olof Larsson
78 **[Digital Media, Migration, and the](#page-77-0) Canadian Federal Election**
James Walsh, Aziz Douai, Amin Ibrahim
82 **[Moral Rhetoric and the Digital Environment:](#page-81-0) The 2019 Federal Election Campaign**
Mireille Lalancette, Tania Gosselin
86 **[Federal Election Candidates' Interest in Indigenous Issues](#page-85-0)**
Jean-François Savard, Mathieu Landriault, Isabelle Caron, Christian Rock
90 **A Study of Indigenous Voices Online [in the 2019 Canadian Federal Election](#page-89-0)**
Kelsey Leonard, Chelsea Gabel, Marrissa Mathews
# Executive Summary
he rise of social media platforms has transformed how we produce, share and consume information. While this re-engineering of our public sphere has enabled unpresented freedom of speech and new forms of political expression and organization, it has also proven to be vulnerable to manipulation and abuse - particularly during elections. In every democratic election since the 2015 Brexit referendum and the 2016 US presidential election, we have seen an evolution in the tools and tactics of those seeking to manipulate public discourse, as well as in to the responses from governments, activists, journalists, and scholars.
An important element of these efforts has been the mobilization of research communities seeking to better understand how the spread of information through our rapidly evolving digital public sphere shapes the character and integrity of our elections. This community, the methods it deploys and the data it uses are in a period of rapid iteration and learning. The objective of the Digital Ecosystem Research Challenge was to support this community in its broad efforts to better understand both the Canadian election and this wider global phenomenon.
The Digital Ecosystem Research Challenge had three core components.
First, we held an open competition to select research projects to study the 2019 election. 18 projects were chosen by an international adjudication committee and grants were awarded in July 2019.
2 Second, working with the Digital Democracy Project at the Max Bell School of Public Policy at McGill University, we provided all research teams with access to online and survey data about the Capadian election
Third, we hosted a symposium of challenge winners, policy makers and civil society to discuss their findings, summarized in this report.
This was a unique experiment in research collaboration in real time during an election. Conducting research on large-scale data sets using emerging research methods is challenging. But as we hope this report illuminates, the results are powerful. Below we summarize the initial findings of the 18 projects. Each will be publishing more detailed reports, based on our growing data set and the benefit of longer research timelines.
Ultimately, a more complete understanding of this complex digital ecosystem is required to build resilience to threats posed by increased integration of digital tools into democratic processes. We hope this report provides an initial review of this space.
# Core findings
### On Political Advertisement and Personal Data:
Damon McCoy's team assessed compliance with Canadian laws regarding transparency in political advertising by analyzing every advertiser listed in Facebook's Ad Library and linking thirdparty advertisers to their legal name. **They found:**
While all advertisers complied with the rules set out by Elections Canada, several still obfuscated their identity or end goals by using "faux fact check" advertisers, which presented themselves as nonpartisan while advancing a clearly partisan agenda.
Sara Bannerman's team studied the extent to which voters were aware of how much data political parties collect, and whether that data collection made respondents less likely to speak with party personnel. **They found:**
Most respondents were unaware that parties collected data on Canadians' religion, ethnicity, and political preferences.
Once made aware of that fact, a majority of respondents said the knowledge made them less likely to talk to party personnel.
Sam Andrey's team assessed the degree to which political parties used Facebook to target particular voters, and whether Facebook was complying with Canadian laws regarding transparency in political advertising. **They found:**
While all parties targeted voters to some extent, such as by age and location, the Liberal Party appeared to make greater use of Facebook's targeting tools than its competitors.
Based on the study's results, Facebook appeared to be complying with the new transparency laws.
# On Online Harassment and Hate Speech:
Anatoliy Gruzd's team studied toxic messages directed at political candidates with a particular view to determining whether female candidates received more negative comments online than male candidates. **They found:**
Less than 10% of all tweets studied directed at candidates were toxic or insulting.
The study also found no statistical correlation between the volume of toxic messages and the candidate's gender.
Heidi Tworek's team studied how candidates and their staff assessed and responded to negative messages on Twitter. **They found:**
About 40 percent of the messages received by candidates on Twitter were categorized as expressing at least "low negativity."
While no candidates or staffers said negative online messaging had derailed their campaign, there was a consensus that the volume of negativity online had deleterious effects on Canadian democracy.
Zeinab Farokhi looked at the degree to which anti-Muslim and anti-Liberal Party rhetoric were correlated during the 2019 election. **She found:**
> The number of tweets linking anti-Muslim sentiment with anti-Liberal messages dramatically escalated as the election approached, before receding again following the election.
# Core findings
# On Public Opinion and Political Participation:
# Fenwick McKelvey**'**s
team tracked the effect of memes on the 2019 election. **They found:**
Two distinct canons of memes: anti-Trudeau memes and pro-Liberal/ NDP memes.
The most common themes in memes shared during the election involved attacks on Justin Trudeau's character, but these memes had limited spread, and especially since Trudeau remained Prime Minister after the election, they did not make a major difference in the election result.
Sabrina Wilkinson studied the political use of Instagram during the election. **She found:**
> While Canadians engaged with politics on Instagram for a wide variety of reasons, their interactions tended to be with others inside a pre-existing social circle, rather than with professional accounts, such as those run by political parties and candidates or third-party political organizations.
Jean-François Daoust and Frédérick Bastien studied how source and presentation affected voters' perception of the reliability of data they saw online. **They found:**
The source of the data (whether a trusted news organization or an ordinary citizen) made little difference to how reliable respondents considered a pre-election poll.
Additionally, when included, methodological information and the graphical presentation of the poll's margin of error did little to change respondents' perceptions of the poll's reliability.
### On The Flow of Misinformation and Disinformation:
Greg Elmer's team explored how content and themes that originated on obscure, far-right corners of the internet became insinuated into mainstream political discourse. **They found:**
The alt-right used site-specific strategies, such as Rebel Media's
practice of using YouTube's narrative vlog style to create "microcelebrities," to help recruit new members and reinforce the beliefs of those already engaged with these fringe communities.
Megan Boler's team examined the "deep stories" and feelings animating voters' online interactions during the 2019 election. **They found:**
Anger and perceptions of hypocrisy in relation to revelation of Scheer's American citizenship were shared across political spectrum regardless of party or partisan differences.
Deep stories animating political perspectives regarding racial and national belonging in each sector of the Canadian political spectrum were less visible within election debates per se than they were within the context of debates regarding Wexit and the firing of Don Cherry, shortly following the 2019 election.
Joan Donovan's team tracked the spread of disinformation and media manipulation campaigns. **They found:**
Several such campaigns were identified and their tactics analyzed. Some of these campaigns, such as a website that promoted false stories about Prime Minister Justin Trudeau's time as a teacher, provoked responses from mainstream media.
Ultimately, the study did not find any disinformation or media manipulation campaigns that succeeded in affecting the integrity of the 2019 election.
# Core findings
### On Issue Framing and Agenda Setting
Shelley Boulianne's team studied the ways in which political partisans discussed climate change during the election. **They found:**
- There were significant differences in concern for climate change and environmental issues across party lines.
- While a majority of Canadians showed significant levels of concern for the environment, such issues were of low priority for Conservative Party supporters.
Jean-François Savard**'**s team looked at the frequency with which the five major parties tweeted about Indigenous issues. **They found:**
- NDP candidates tweeted the most about these issues, followed by the Green Party and the Liberals.
- Candidates running in urban areas tweeted more frequently about Indigenous issues, and the most common topics were reconciliation and environmental issues, including access to clean drinking water.
Kelsey Leonard's team analysed issues discussed by Indigenous peoples in the lead-up to the election. **They found:**
Prominent in these discussions were issues like access to clean drinking water in Indigenous communities and accusations of "ethnic fraud" perpetrated by candidates who claimed Indigenous ancestry without any evidence.
James Walsh's team examined discussions of migration. **They found:**
- In comparison with headlines in traditional, mainstream media outlets, Twitter posts were significantly more negative. The most common theme present in Twitter posts was of the risk supposedly posed by immigrants coming to Canada.
- Posts portraying immigration in a positive light were about as likely as negative posts to receive likes, but less likely to be retweeted.
Mireille Lalancette and Tania Gosselin explored the moral rhetoric advanced by political parties and the degree to which that was reflected in citizens' party preferences. **They found:**
While ordinary citizens were less likely to describe their voting decisions in moral terms, politicians frequently used the rhetoric of moral values, including a strong focus on loyalty and authority from right-wing leaders Andrew Scheer and Maxime Bernier.
# Introduction
n October 21, 2019, Canadians went to the polls to elect the 338 Members of Parliament that represent them in the House of Commons. This election cycle was shaped by digitally mediated transformations in the way citizens, political candidates, advocacy groups and other political actors relate to one another. In the context of elections, digital media can present valuable opportunities to increase communication and participation as well as to shine light on topics that might otherwise be overlooked by traditional media or major political parties. At the same time, citizens, journalism organizations, political parties, and governments must be equipped to protect themselves against potential threats posed by digital media including, for example, the proliferation of false information online and the misuse of citizens' personal data.
The digital ecosystem is the collection of various digital tools and systems, as well as the actors which create, interact with, use, and remake those tools and systems. This complex ecosystem involves a diversity of actors and networks interacting across a variety of channels and platforms, each with their own interests. The notion of an ecosystem underscores the importance of considering these interactions. Ultimately, a more complete understanding of this complex ecosystem is required to build digital literacy and resilience to threats posed by increased integration of digital tools into democratic processes.
While the digital plays an increasingly important role in how we communicate, learn, organize and campaign during election, challenges to studying the impacts of digital media on elections have led to gaps in our understanding. As individuals navigate the Internet they leave traces of their behaviours and preferences - called trace data. This data is collected by the digital platforms and tools used in everyday life such as social media and search engines. But accessing this data is expensive and requires technical skill. Since the data as well as the tools for collecting and analyzing the data are often proprietary, large-scale research is difficult to undertake and replicate. A high level of technical skill is required to make sense of the extremely opaque information systems in which this data is created and maintained. Furthermore, it is technically
difficult to keep up with rapid innovations in digital media. There is also a lack of coordination among civil society and academia which leads to duplication and inefficiency in the work that has been done. This in turn limits the extent to which new research can be quickly applied in order to increase civic and digital literacy among citizens. The Digital Ecosystem Research Challenge addresses these challenges by providing researchers and civil society the opportunity to communicate and collaborate on research to break down silos.
#### Regulatory context
The Canadian Government took steps in advance of the 2019 election to promote electoral integrity. These measures included the Plan to Safeguard Canada's 2019 Election, which rests on four pillars: foreign interference, institutional capacity, responsibility of social media companies and equipping citizens (Government of Canada, 2019). The Government also increased resources to the RCMP Foreign Actor Interference Investigation team, established the Security Intelligence Threats to Elections (SITE) Taskforce, committed \$7 million to increase digital literacy among Canadians and set up a Critical Election Incident Public Protocol (Government of Canada, 2019). Legislative changes to the Canada Elections Act (CEA), specifically on spending by third parties, voter registration and access, political financing, and the protection of personal information were also key measures adopted by the Government (Government of Canada, 2019). These government initiatives were adopted in a political climate in which foreign interference in the 2019 election was judged "very likely" by Canada's Communication Security Establishment (Communication Security Establishment, 2018).
While the Government's 2019 plan and its updates to the *Elections Act* demonstrate a commitment to tackling digital threats to election integrity, challenges remain. The law lacks necessary clarity in certain areas and leaves open the possibility that actors could exploit technologies to skirt the law (*Reepschlager & Dubois*, 2019).
# Increasing digital literacy through research
The first step to digital literacy and to data-driven policy making is understanding the nature of our political information ecosystem. Continued research in this space is critical to address profound gaps in knowledge and provide nuance to current generalised statements in the discourse. Research specific to the Canadian context is essential because digital ecosystems differ by country. Mapping the Canadian ecosystem requires trained researchers to ensure reliability and validity and to avoid bias. To develop effective tools for tackling threats posed by the transforming digital ecosystem, civic literacy initiatives and citizens require an understanding of the information ecosystem in which they are expected to collect and evaluate political information.
This report highlights projects which build from academic work on digital ecosystems in the electoral context to identify current successes and barriers, opportunities for public policy and civil society, and areas for additional research and collaboration. These projects collaborated to analyze a large scale dataset collected from social media, online news, web, and survey data. The data and methods are further described in the next section and over the coming months we will be making available a series of online appendices which further detail methodological approaches.
### Reading the report
This report looks at 181 research projects from across Canada focused on five themes to develop a broader understanding of Canada's digital media ecosystem:
Online political advertisements;
Online harassment and hate speech;
Public opinion and political participation;
The flow of disinformation and misinformation; and
Issue framing and agenda-setting.
The report is designed so you can easily read any of these five sections on its own or in any order you like. For each section you will find an introductory page which reviews some of the most relevant academic literature in the area in order to set the scene for the subsequent project reports. Notably, each area could easily be its own detailed and lengthy report. We endeavour here to offer highlights and encourage readers to contact corresponding authors for each project in order to delve deeper.
All 18 projects outline the key questions asked, the way the study was conducted, initial findings, impacts on the election, and suggestions for next steps both for researchers and policy makers.
**1** Note, one project will be available in an updated online version of the report only. It will be available in early spring 2020.
# Online Data Collected
During the elections, our focus was on collecting data from four platforms: Twitter, Crowdtangle (Facebook and Instagram), Reddit, and traditional news.
# Twitter
Twitter data was collected from August 1 to November 30, 2019. The objective of the Twitter data collection effort was to capture all major party candidates, major news organizations, as many Canadian journalists as possible, and a broad swath of the public conversation. Journalists and media organizations were identified using an iterative approach that began with a core seed list of all Members of Parliament, the Twitter handles of all declared candidates, and approximately 300 journalists, academics and news outlets validated by a Canadian politics media expert. Additional accounts were added throughout the election based on frequency of posting on Canadianelection related themes, and mentions to and from existing tracked accounts. This algorithm yielded approximately 5000 likely accounts of interest, each of was reviewed manually for Canadian politics relevance and then tracked. This yielded a total of 3889 explicitly tracked accounts, of which 830 were journalists, 272 were official accounts of news outlets, 1280 were candidates of major parties or official party accounts, and 1507 were third-parties (a broad category including provincial politicians of note, registered third-party advocacy groups, academics, and public users of Twitter with many followers).
To supplement this elite-oriented collection effort, we gathered a list of Canadian politics hashtags. We used a similarly iterative strategy and began with a list of 33 well known Canadian politics seed hashtags. This list was expanded in an iterative manner based on hashtag co-occurrence and frequency of use by our existing tracked users on a weekly basis throughout the entire campaign. 1854 hashtags were identified in this manner, and then manually reviewed to identify their Canadian politics relevance. This yielded a list of 552 hashtags that were used for Canadian politics discussion in the pre-election or writ periods. In total, these methods collected about 19 million tweets.
Additionally, we were planning on using PowerTrack to obtain a more complete Twitter dataset but this process was delayed
due to the complicated nature of the vetting procedure. We hope to receive this dataset in the coming month. We are also in the process of collecting the 3000 most recent tweets of all accounts that posted during the elections, but this will take many months. In addition, we have a time series network dataset of followers for each candidate and politics journalist tracked, except Justin Trudeau who was only scraped once. Finally, we have collected a time series of Twitter's trending hashtags.
We have been sharing Twitter data with grantees during and after the elections by responding to requests for tweets matching specific criterias such as hashtags, authors, or textual content. Per Twitter's terms and conditions, we have been limited to sharing data by means of lists of tweet IDs, which can then be used to fetch the data from Twitter (we provide the code for this).
# CrowdTangle
Facebook and Instagram data was sourced from CrowdTangle, a social media analytics tool owned by Facebook. CrowdTangle tracks posts on Facebook and Instagram, made on public accounts, pages or groups. The tool does not track every public account and does not track private profiles or groups, so this data is not perfectly representative of performance across the entire platform. CrowdTangle also does not track posts made visible only to specific groups of followers.
Throughout the election we iteratively added Facebook and Instagram pages, groups and accounts to follow based on keyword and manual searches by a paid research assistant. In total, we collected detailed information on 1,575 Canadian politics pages and groups on Facebook and 719 accounts on Instagram. For these groups, we scanned them at regular intervals so as to track the growth of posts and associated reactions. As an additional supplement these data, we searched for all political links shared on Twitter and in the general Facebook and Instagram collection effort and then used the CrowdTangle API and added these posts and their associated groups to our dataset. This dataset can be shared without restrictions.
# Reddit
We collected 128,412 submissions from 181 Canadian subreddits. This collection has likely missed some comments that were deleted before our scripts managed to find them. This data can be augmented using PushShift as appropriate. We have also collected posting history for submission authors found in our dataset. This dataset can be shared without restrictions.
# News
We scraped all news articles we found in our Twitter dataset and extracted their content using a combination of Newspaper3k and custom code. We also collected RSS feeds of as many news outlets as we were able to and used the same extraction mechanism to get the contents. As a result, we successfully extracted 334,447 articles using the first method and 81,703 articles with the second. This dataset can be shared without restrictions.
# Press Releases
We collected a dataset of 475 articles on party websites (title, date, and body). The per-party article counts are shown below. This dataset can be shared without restrictions.
# Survey and metered data
We collected survey data over nine waves from 14,554 Canadian citizens 18 years and older using the sample provider Qualtrics. The fielding dates are found below:
Wave 1: July 24 - August 7
Wave 2: August 17 - August 23
Wave 3: August 28 - September 5
Wave 4: September 11 - September 16
Wave 5: September 19 - September 24
Wave 6: September 27 - October 3
Wave 7: October 4 - October 13
Wave 8: October 14 - October 20
Wave 9: October 24 - November 4
Quotas were set for sample collection so that each wave was nationally representative by region, gender, language and age according to the 2016 Canadian census. Data was weighted within reach region of Canada by age and gender, as well. We used an iterative proportional fitting algorithm for our weighting procedure with a minimum weight of 0.25 (N=21) and a maximum weight of 2.98 (N=1).
Survey respondents were asked questions related to basic demographics, as well as their partisan, ideological and issue preferences. They were also asked to report their exposure to the news media in the previous week. Occasionally they were exposed to randomized experiments - the nature of which varied wave to wave. The median completion time was 25 minutes. The survey instruments are available upon request. We present 90% confidence intervals with our figures.
We supplemented our survey data with an unrepresentative sample of 754 respondents provided by Qualtrics for whom we tracked online activity for a period of four weeks through the campaign period (Sept. 19 - Oct. 20). This allowed us to observe online news consumption for this set of respondents.
# Political Advertisement and Personal Data
# What you need to know
As the Internet has become embedded in daily life, political actors have turned to social media, search engines, and other online platforms for political advertisement. These newer media are seen by many voters as primary sources of information (*Owen, 2014*). Making use of social media platforms, candidates can reach very large audiences (*Spenkuch & Toniatti, 2018*) and can communicate directly with electors (*Iyengar, 2011*).
Microtargeting is a specific marketing technique that political parties are increasingly using. Companies like Facebook and Twitter can offer advertisers the opportunity to select very specific and narrow audiences for their advertisements. This can be based on location, language, age, gender, interests, and more. When using these platforms people offer up much of this personal information which is stored as data about them. Political parties also keep track of personal data, including the issues a person cares about and their likelihood of voting for or donating to a political party. Keeping track of voter information is nothing new, of course, but the scale and speed is increasing as digital tools are integrated into campaigns.
Political campaigns are using that information to target voters with specific messages based on their profile (*Owen, 2014*). Microtargeting could help ensure people receive messages that matter to them. But it can also mean that different people receive contradictory information. Furthermore, the personal data required for microtargeting is often sensitive and it is unclear how safely it is stored, whether it is correct, and how individuals might learn about or change the data their political representatives hold about them.
# What we still don't know
- How much, and what, personal information is collected by political parties? How do they make use of it? This is opaque because political parties are not subject to the same privacy laws as companies and because political parties will want to keep their strategies secret from their competitors.
- Although several researchers agree that online political advertisements influence voters' perception of candidates during an election (*Lee Kaid et al., 2007*), more needs to be done to understand how these negative and positive ads influence political participation.
- New online advertising rules were introduced ahead of the 2019 federal election to require the creation of advertisement repositories on large enough websites and platforms. How are platforms, advertisers, and citizens responding?
# What you'll learn from the projects
- On Facebook, even though advertisers complied with the letter of new rules, they still tried to hide their activities, using tactics like identity obfuscation, microtargeting, and faux-fact-checking (*McCoy, et al.*). **[PAGE 16](#page-15-0)**
- Using personal information to microtarget on Facebook was a common tool that all major political parties were using but the frequency varied by riding (*Andrey, et al.*). **[PAGE 20](#page-19-0)**
- Canadians are largely unaware that their personal information is collected and used by political entities. But once aware, collection of personal information by political parties affects the will of voters to be involved with party personnel during campaign period (*Bannerman, et al.*). **[PAGE 24](#page-23-0)**
# Facebook Advertising During the Canadian Federal Election
#### **Laura Edelson**,
New York University, PhD Candidate lj992@nyu.edu
#### **Divam Jain**,
Data Scientist, Independent Researcher
#### **Damon McCoy**,
New York University, Assistant Professor
# **What we wanted to know**
New rules were introduced ahead of the 2019 Federal Election to help make online political advertising more transparent. But some actors may try to subvert those rules. Focusing on social media and advertising during the 2019 Federal Election, this study identified new types of inauthentic political advertising activity on Facebook and documented them for further research into mitigation strategies.
We focused our attention on political advertising on Facebook to **answer the following research questions:**
How well did Facebook political advertisers' comply with rules set forth by Elections Canada (*"Third Party Database", n.d.*), which require ads and the money spent on them be traceable to Canadian legal entities registered with Elections Canada?
What techniques were used by advertisers to obfuscate their advertising activity? We attempted to identify different types of activity which might be intended to mislead users or auditors, in particular:
Organizations using a variety of identities to reinforce a single message or distance themselves from a certain message.
Advertisers microtargeting different messages to different audiences, particularly based on gender, age, geolocation and language.
Partisan advertising masquerading as unbiased factual reporting.
# **How we did it**
Retrieved all Canadian political ads and Facebook pages active between Sept 11th and Oct 21st 2019.
The Facebook-provided Ad Library included **2,681 pages**, which ran **45,633 ads** which were active during the election period. Facebook also provided a range of spending for each ad along with impressions per age-group, gender and province for the ads.
Manually categorized Facebook pages into political party/candidate or third-party.
Mapped third-party Facebook pages to their legal names as registered with Elections Canada (*"Third Party Database", n.d.)*.
Manually reviewed ads for activity which might have violated electoral law or the spirit of the law (*Canada Elections Act, 2019; "New Registry Requirements for Political Ads on Online Platforms", n.d*.).
Assessed the prevalence of such activity across advertisers and their ads.
### Key terms and and what they mean
#### Advertiser Facebook Page
All advertisers have a Facebook page, which is their identity on the platform. It comes along with a web page to allow users to interact with the advertiser directly. Although uncommon, advertisers may have more than one Facebook page.
#### Ad Sponsor
Simple text supplied by the advertiser which discloses the entity paying for the ad. This is in addition to information about the Facebook page an ad is run from.
#### Registered Third Party
A legal entity which has registered with Elections Canada to run partisan ads during the election or pre-election period, or issues-based ads during the election period.
#### Inauthentic advertising activity
For political ads specifically, ads which attempt to misrepresent the source of the ad, use demonstrably false content, conflicting messaging based on microtargeting, or hide the advertiser's partisan lean. This definition of inauthentic activity is based on Facebook's definition of inauthentic behavior ("Community Standards.", n.d.).
#### **Fact Check Ad**
An ad which claims to be providing impartial factual rebuttals to political speech or ads.
#### **Microtargeting**
An online advertising technique of showing an ad exclusively to a certain group. In this case, we focus on demographic groups based on age, gender, and location. Other factors like ethnicity, personal interests, or partisan lean were not available in the data.
### What we found
Overall Compliance with rules: All major political advertisers listed in the Facebook Ad Library complied with the rules set forth by Elections Canada. Figure 1 outlines national spend estimates by parties and candidates for those parties alongside the total third party spend in the election period. Of third party advertisers, Elections Canada was the largest, while the top three partisan advertisers—those directly campaigning for or against a candidate or party—were "Canada Proud," "Proud To Be Canadian," and "Fire Fighters for Canada." Provincially, the top advertisers were often different except for 'Canada Proud,' which outspent most partisan advertisers in every province.
**Identity Obfuscation:** There were groups of third parties, such as "Canada Strong and Proud," which have affiliates and whose level of collaboration and coordination are unclear (*Paas-Lang, 2019*). This leaves open the possibility that advertisers amplified one message by giving it multiple supporting voices with smaller spend footprints, which may not be separate in reality. "Proud To Be Canadian," our second-largest partisan advertiser, was part of this network.
Figure 1. Party & Advertiser Type: Estimated spend by Parties and Candidates.
### **What we found**
**Microtargeting**: Based on the impression data provided by Facebook, it was clear that microtargeting was used for practically every ad. While this practice is not inherently inauthentic, it can be abused, as has been the case in other elections (Ribeiro et al., 2019). An example of undesirable inauthentic microtargeting, which was reported on by the Canadian press, was a Chinese language attack ad paid for by the Conservative Party of Canada (Boynton, 2019; Slaughter, 2019). **Figure 2** contains an image of this ad and its targeting. The ad was notable in that it was only run in Chinese, unlike other ads, which were multilingual, and that it was not published using the party's primary Facebook page alongside other Chinese ads.
A positive benefit of microtargeting was that Elections Canada focused spending on raising voter awareness in demographics with historically low voter turnout. Most partisan advertisers also used microtargeting to concentrate different messages on different demographics or geographic regions.
Figure 2. Conservative Party Chinese language page ad. Translation of ad text: "Trudeau previously legalized marijuana, now he intends to legalize hard drugs!" This ad was removed for violating Facebook's policies on images of drug usage.
Figure 3. Faux Fact Check ad with deceptive disclosure string text.
**Faux Fact Checks**: Another notable strategy we saw deployed for the election was what we refer to as "Faux Fact Check" ads. We found two different Facebook pages with the phrase "fact check" in the page title. However, these ads appeared to be from partisan sources and promoted negative messages about their political opponents. The first page, "Election Fact Check," clearly disclosed that it was sponsored by the Conservative Party of Canada. The other, which attacked the Conservatives, stated that it was sponsored by "Canada Fact Check," providing a link to an associated website and the political consultant running the campaign, obfuscating the source of funding. An ad from this page is pictured in Figure 3. We also found ads from North99 that solicited donations for a "fact check" website. That site, conservativeplan.ca, also appeared to be partisan. We believe this is the first study to uncover the systematic use of 'Faux Fact Check' ads on Facebook.
# What this means for the election
- As expected in online advertising, ads appeared to be targeted based on user data. We documented targeting based on age, gender, and location, but the extent of user data is far greater, meaning political advertisers could target messages to small groups of people without the need for dog whistles or awareness of those messages outside of targeted communities.
- "Fact checks" which were meant to combat misinformation and disinformation online were incorporated into the misinformation they were meant to address. The co-opting of this term to obfuscate partisan political messaging was a newly documented pattern of inauthentic activity during the election.
- Advertisers used different identities to amplify or distance themselves from certain messages they
- funded. Advertisers were able to project that an opinion was more widely believed than it was or run ads they didn't want to be directly affiliated with by using multiple Facebook pages with different names or languages.
- The above findings are limited by the scope of data provided by Facebook about political ads. We did not receive targeting criteria with ads, and could only infer some coarse grained factors based on impression information. Further, funding entities for Facebook ads are advertiser-provided with no obvious verification, which could lead to potentially incomplete data in our analyses. The data also do not provide the universe of ads Facebook deemed non-political, which possibly contained inauthentic activity that could not be identified.
# What's next
- Elections Canada should explore the feasibility of deeming the term 'fact check' journalistic and bound by the same ethical rules that news organizations must follow when used in the context of political ads.
- Require all ads on each platform along with the advertiser name to be catalogued by ad sponsors registered with Elections Canada. The catalogue should be accessible publicly with real time data.
- Require more detailed targeting and impression information for political ads. This would include which ridings were targeted (or equivalent) and all targeting information which is not personally identifying (for example: Facebook identified interests/income level/marital status, etc.).
- **Future Work:** Detailed assessment of paid and unpaid political content by political actors on other platforms, particularly Youtube, WeChat, TikTok, Twitter, Reddit, Snapchat, and Instagram, for the same patterns of activity we saw in the Facebook dataset.
Analysis of fine-grained geo-targeting. Our current dataset was limited to provincial-level data, which prevented us from understanding the different ways parties targeted voters in different ridings.
# Who Targets Me? Political Advertising on Facebook in the 2019 Election
#### **Sam Andrey**,
Ryerson Leadership Lab, Director of Policy and Research sam.andrey@ryerson.ca
#### **Karim Bardeesy**,
Ryerson Leadership Lab, Executive Director and Co-Founder
#### **Sam Jeffers**,
Who Targets Me, Co-Founder
#### **Justin Savoie**,
Vox Pop Labs, Data Scientist
#### **Clifton van der Linden**,
Vox Pop Labs, Chief Executive Officer
# **What we wanted to know**
Political advertising in federal election campaigns has increasingly shifted from television and radio to online platforms, from 86% of advertising spending by political parties in 2011 to 76% in 2015 (*Elections Canada, 2011/2015*). After Google withdrew from political advertising for the 2019 election campaign, Facebook served as the election's primary online advertising platform (*Delacourt, 2019; Maher, 2019*). Six in ten Canadian adults report using Facebook News Feed with 36% checking it every day — more than any non-Google platform (*Ryerson Leadership Lab, 2019*). **This study sought to investigate:**
The effectiveness and compliance of Facebook with Canada's new political advertising law. Specifically, it sought to determine the demographics of voters being targeted, as well as the methods and messages used to target them.
How Canadian political and third parties used Facebook advertising during the 2019 campaign.
### **How we did it**
305 Canadians were recruited without incentive from an online research panel to install the Who Targets Me web browser extension.
Participants provided demographic information (postal code, age, gender, political spectrum, etc.).
The extension collected 8,142 advertisements from participants' Facebook News Feed from October 14-21, 2019 along with targeting method where available (see next section).
# **Key terms and what they mean**
Paid posts of text, images and/or videos that appear in a Facebook News Feed alongside other non-paid posts.
#### Political advertisement
Advertisements from political entities (parties, party leaders, candidates and third parties) that promote or oppose a party or a candidate, or that take a position on political or electoral issues.
#### Political advertisement law
Amendments to the *Canada Elections Act* made in 2018 added new requirements for online platforms selling political advertisements to publish a registry of all political advertisements with information about who authorized the advertisement.
# **Key terms and what they mean**
#### Targeting method
Facebook provides four broad methods (which can also be combined) to target advertisements to specific audiences:
> **Demographics** such as location, age, gender, language and/or education level provided by Facebook users.
**Interest** through the pages and/or advertisements that users interact with (e.g., leader/party pages, the environment, business, etc.).
**Custom audience** of specific users created by matching uploaded personal information (e.g., name, email, phone number, postal code) with Facebook's user data.
**Lookalike audience** through which Facebook identifies the attributes and/or interests of a custom audience and delivers the advertisement to a new audience that is similar.
### **What we found**
Figure 1. Political advertisements as a percentage of all Facebook advertisements by date
In the week leading up to the election, approximately 13% (n=1,085) of the 8,142 Facebook advertisements observed were political (see **Figure 1**).
This overall picture masks considerable variation in political advertising by riding. A minimum of 20 advertisements were observed in 71 of 338 ridings. **Figure 2** shows the percentage of total advertisements that were political in these 71 ridings. In eight (11%) of those 71 ridings, there were no political advertisements observed. In comparison, in the five ridings with the highest percentage of political advertisements, between 50% and 96% of advertisements were political (see **Figure 3**). All five ridings were closely contested by multiple political parties. For example, in the swing riding of Drummond in the province of Québec, the 29 political advertisements were from five different political parties.
Figure 2. Distribution of federal ridings by percentage of total Facebook advertisements that were political
Figure 3. Advertisement breakdown of top five targeted ridings
# **What we found**
The political parties deployed a variety of advertisement targeting methods to reach voters (see **Figure 4**). Just under half of the Liberal Party's advertisements (20 out of 41 adveritsements) targeted specific voters, either those that visited the party's websites or lookalike audiences. The remaining half were targeted only by location. In comparison, the Conservative Party (4 of 29 advertisements) and the NDP (1 of 4 advertisements) targeted specific voters less often. Most of these advertisements targeted location or a combination of location and age range. None of
the other political parties or major third parties were observed using custom or lookalike audiences, instead they were focusing primarily on location, age and interest.
This analysis is the first in Canada to review how political entities are using Facebook's targeting tools. This study is also the first attempt to better understand the impact of advertising in targeted ridings has on the frequency of political advertisements displayed on Facebook.
Figure 4. Targeting methods used by political parties and their leaders
# Total Advertisiments: Liberal Party: 103 Conservatives: 75 NDP: 17 Bloc Québécois: 6 Green: 0
#### Advertisement Targeting Methods:
No targeting (only > age of 18 in Canada)
Location targeting
Age range and location targeting
Interest targeting
Website visitor targeting
Custom audience
Lookalike audience
Targeting info not available
# What this means for the election
- Political entities use targeting options on Facebook differently from one riding to the next. For example, in this study 62% of the advertisements from the riding of Victoria (BC) were political compared to none in the neighbouring riding of Chilliwack—Hope (Figure 3).
- The Liberal, Conservative and NDP parties all used the personal information of voters to target Facebook advertisements. The Liberal Party deployed more precise targeted advertisements than the other parties. Further investigation is needed to determine whether this provides an explanation for the much smaller average advertisement cost that the Liberal Party ran compared to the other parties (Andrey & Bardeesy, 2019; Owen et al., 2019).
- Based on the data collected, Facebook complied with Canada's political advertisement law contributing to greater transparency. The study did not collect any advertisements from registered political entities that were not included in Facebook's registry, nor were any advertisements from political parties identified during the advertising blackout period required by the Canada Elections Act.
- The data presented is derived from participants using Facebook News Feed on Google Chrome or Mozilla Firefox. The data collected certainly did not encompass all the election's political advertisements, and political entities may have deployed different advertising methods on mobile devices or on Facebook's other platforms such as Instagram.
# What's next
- Expand legal transparency requirements: Facebook voluntarily posted more details in its political advertising registry than required by law, providing each advertisement's approximate cost and total number of times the advertisement was viewed. To facilitate greater transparency, these steps should be legal requirements. In addition, each advertisement's targeting method(s) should be included in the advertising registry, including reporting location targeting criteria at a riding or postal code level.
- Ban the use of personal information without consent to target online advertisements: Federal privacy laws do not apply to Canada's political parties, and Elections Canada is required to provide registered parties and candidates with voters' names and addresses to facilitate communication. This enables parties and candidates to microtarget online advertisements to specific lists of voters without their consent. This practice not only risks undermining public trust in voter registration but is inconsistent with Canadians' expectations — in fact, 63% of Canadians support banning targeted online advertisements during elections altogether (Ryerson Leadership Lab, 2019).
- Continue research and engagement on the impact of online advertisements: This project did not seek to understand the impact of advertisements on voting behaviour. It remains an open question how impactful they are in doing so and the role of public or platform policy in preventing disinformation through paid advertisements.
# Political Parties and Data Privacy
#### **Sara Bannerman**,
McMaster University, Department of Communication Studies and Multimedia, Canada Research Chair in Communication Policy and Governance and Associate Professor, banners@mcmaster.ca
#### **Nicole Goodman**,
Brock University, Department of Political Science, Assistant Professor,
#### **Julia Kalinina**,
York University, Osgoode Hall Law School, Research Coordinator and JD Candidate
#### **Jenny Zhan**,
University of Waterloo, Research Coordinator and MPH Candidate
# **What we wanted to know**
The political power of data has attracted a lot of attention in the news media recently. Questions have been raised regarding how political parties are collecting and using personal information. To date, however, little research has been done to explore federal political parties' collection and use of electors' personal information, and the effects on political trust and willingness to interact with political parties during elections. To fill this lacuna, our project sought to answer **three questions:**
Are Canadians aware that federal political parties collect their personal information?
Do partyies' privacy policies affect electors' willingness to speak with party campaign personnel?
Does awareness of parties' collection of personal information make a difference in electors' willingness to speak with party campaign personnel?
# **How we did it**
Survey data of 1,594 Canadians age 18 and over was collected from September 19 to 24, 2019.
Questions were asked about their knowledge of federal political parties' collection of personal information, and about whether collection of this information or a party's stance on privacy affected their willingness to engage with party campaign personnel.
Data analysis: To analyze the data presented in this report we use descriptive statistics. In other analyses of the data we also employ regression analysis.
# **Key terms and what they mean**
#### Personal information/data
In Canada, under the *Personal Information Protection and Electronic Documents Act*, "personal information" includes any factual or subjective information, recorded or not, about an identifiable individual. This includes information in any form, such as: age; name; income; identification numbers; ethnicity; opinions; evaluations; disciplinary actions; credit records; loan records; and other information.
#### Personal data
The predominant term for "personal information" in the European Union. It is defined broadly in the *General Data Protection Regulation* as "any information relating to an identified or identifiable natural person" (*ch. 1, s. 4(1)*).
#### Privacy
"The quality or state of being apart from company or observation" (Merriam Webster, *n.d.*).
#### Privacy law
The legal regulation of the collection, use, and storage of personal data which can be used to identify individuals by governments or private parties.
#### PIPEDA
*Personal Information Protection and Electronic Documents Act* is the Canadian federal law governing how private sector organizations may collect, use, and disclose personal data in the course of commercial activity; it does not currently apply to federal political parties.
Many respondents were not aware of the extent to which political parties collect personal data. Respondent knowledge varied by type of information. Notably, respondents were most likely to report knowing that parties collect names (n=999; 63%), contact information (n=914; 57%), and gender (n=822; 52%). Most respondents did not know, however, that political parties may also collect and use information about political views (only n=526; 33% knew), ethnicity (only n=522; 33% knew), and religion (only n=393; 25% knew).
After being informed that federal parties may collect such personal data, 51% of respondents indicated that they were "somewhat" (n=430; 27%) or "very" (n=386; 24%) unlikely to speak with party campaign personnel. Thirty-two percent of those surveyed said they would be "somewhat" (n=366; 23%) or "very" (n=150; 9%) likely to speak to campaign personnel. Sixteen percent did not know or preferred not to say.
**What we found** Awareness of political party data collection practices
# the ways in which political parties collect and use data may have important implications for privacy governance and democratic engagement in Canada.
# **What we found**
Finally, we informed respondents that "[f]ederal privacy law does NOT currently apply to federal political parties. Parties have different privacy policies. Some parties believe that federal privacy laws should apply to federal political parties while others think they should be exempt." The survey then asked whether "a party's stance on privacy make[s] a difference" in willingness to speak with party personnel. A majority of respondents (55%) indicated that a party's stance on privacy did alter their willingness to engage with campaign personnel: 16% (n=258) said it made a "major difference;" 25% (n=400) said it made a "moderate difference;" and 13% (n=214) reported that it made a "minor difference." In contrast, 27% (n=435) of respondents indicated that privacy policies made "no difference" in their willingness to engage with campaign personnel. Eighteen percent (n=287) said they did not know or preferred not to say.
Awareness of political party data collection practices
# What this means for the election
- Many Canadians are not aware that federal political parties collect their personal data;
- Awareness of federal parties' collection of personal information may discourage some electors from engaging with party personnel during elections;
- Parties' stances on privacy are an important consideration in Canadians' willingness to engage with party personnel;
- Appropriate data practices and privacy policies are needed in Canada to support and reinforce Canadians' confidence in party institutions and democratic engagement.
While we did not directly ask whether extending privacy law to political parties would improve voters' trust in, and opinions of, parties, many respondents reported that their willingness to engage with party personnel was affected after learning about the extent to which parties may collect electors' personal data. Knowing that parties may collect their personal information, a majority of respondents report being unlikely ("very" or "somewhat") to engage with campaign personnel. Consequently, the ways in which political parties collect and use data may have important implications for privacy governance and democratic engagement in Canada.
# **What's next**
- Federal privacy laws should be extended to apply to federal political parties. The question of whether the *Personal Information Protection and Electronic Documents Act* (PIPEDA) should be amended to apply to federal parties, or whether legislation should be enacted that is specific to political parties should be subject to further research. 1
- The regulation of parties' collection and use of data is increasingly important to support the democratic process in the digital age. Collection and use of electors' personal data could not only play a decisive role in election campaigns, but such practices could also affect interactions between electors and party personnel. Voters should be assured that parties are following established standards in their practices with electors' data. 3
- In addition to the federal extension of privacy laws, we recommend that all provinces (except for BC, which has enacted such laws) should enact privacy laws that regulate provincial political parties' use of electors' personal information. 2
- **Future research agenda:** Future studies could more fully examine the extent to which knowledge of parties' data collection affects electors' levels of political trust, and whether awareness of how parties are using personal data affects other political attitudes and electors' willingness to participate politically. 4
# Online Harassment and Hate Speech
# What you need to know
Online harassment, bullying, and hate speech can threaten democratic structures. From voters, to candidates, to political reporters, there are people pushed out of the democratic system because they fear these types of harmful speech, silencing individuals who would otherwise participate in the democratic process (*Tenove et al., 2018*). But responding to, and quelling, this harmful speech is not easy. If rules are too strict then legitimate political speech could also be removed or otherwise limited.
The problem is, there is no one clear definition (*Faris et al., 2016*) — different countries and companies have different understandings of what constitutes harassment, bullying, or hate speech, and what speech should be protected. Even once these ideas are defined, detection and removal of content or accounts is typically inconsistent.
Some groups use harmful speech as a tactic to build and mobilize community and to push other voices out of the community and offline (*Marwick & Lewis, 2017*). These groups can use social media to engage in member recruitment, fundraising, and disseminate their message to large audiences (*O'Callaghan et al., 2014*).
What's more, the very design of many social media companies' content suggestion and prioritization algorithms can amplify harmful speech. Studies have shown that right-wing extremist content taps into people's emotions (*Matthes & Schmuck, 2017*), and that emotionally charged messages are shared more broadly on social media (*Brady et al., 2017*).
# What we still don't know
Who are the targets of online harassment, bullying and hate speech (*Silva et al., 2016*)? Some research shows that individuals from marginalized communities face the worst kinds of abuses (*Powell et al., 2018*).
How do public figures respond? Although research has been conducted on the harassment of public figures (*Stahel & Schoen, 2019*), we do not have a thorough understanding on its impact, what strategies public figures employ to protect themselves, which makes it difficult to create data driven public policy decisions (*Faris et al., 2016*).
Are efforts to slow harmful speech online working? This question is hard to answer because social media platforms do not share specific information about what they are doing or data about their success rate.
# What you'll learn from the projects
Two projects analyzed the content of tweets mentioning candidates during the election.
The most intense negative messages are a small proportion of all mentions of candidates on Twitter and do not appear to be related to gender (*Gruzd*). **[PAGE 30](#page-29-0)**
But, when you consider less intense forms of negativity, the proportions are higher and gender-based differences may be more prominent (*Tworek*). **[PAGE 34](#page-33-0)**
Candidates respond differently to negative messages depending on who sends them (*Tworek*). **[PAGE 34](#page-33-0)**
Farokhi examined Twitter content mentioning Muslims and Trudeau. In her case study she finds Islamophobic emotional rhetoric increased rapidly prior to and during the 2019 Canadian election. **[PAGE 38](#page-37-0)**
# Toxic Interactions and Political Engagement on Twitter
#### **Anatoliy Gruzd,**
Ryerson University, Associate Professor, gruzd@ryerson.ca
#### **Philip Mai**,
Ryerson University Co-Director, Social Media Lab,
#### **Raquel Recuero**,
Universidade Federal de Pelotas (UFPEL/Brazil) Associate Professor
#### **Felipe Soares**,
Universidade Federal do Rio Grande do Sul (UFRGS/Brazil) PhD Candidate
# **What we wanted to know**
The goal of this study was to better understand the online dynamics of violence on Twitter against candidates running for political offices. Online violence on online platforms is a pressing problem. This study will provide research-based evidence for policymakers, governing stakeholders, researchers, and social media intermediaries working to address current knowledge gaps and challenges associated with toxic interactions on platforms like Twitter. It will also help examine the capabilities and overall effectiveness of Twitter's platform-based guardianship (i.e., automated and human-led content moderation).
#### **Our research was guided by the following exploratory questions:**
What is the prevalence of toxic and/or insulting messages targeting political candidates?
Is there a difference in frequency of toxic and/or insulting messages directed at women versus men candidates on Twitter?
# **How we did it**
Candidate list: Compiled a list of 2,144 candidates running for office and recorded their Twitter handles, when available. In total, 1,344 candidates had a public profile on Twitter at the time of data collection during the fall of 2019.
Data collection: Used an open source program called Social Feed Manager to collect public tweets directed at each of the 1,344 candidates, from September 28 to October 27, 2019. We focused on 363,706 public tweets posted in English and directed at 1,116 candidates1 .
Content analysis: Hand coded a random sample of 3,637 tweets (1% sample) to categorize them as toxic or insulting (as defined in the next section) by using three coders (one undergraduate racialized, female student, one graduate female student, and one graduate male student). To ensure the trustworthiness of the manual coding procedure, only tweets that were flagged by all three coders were considered as either toxic or insulting for the subsequent analysis.
Statistical test: Used a statistical test (chi-square) to determine if there is an association between a candidate's gender and the likelihood of them receiving either a toxic or insulting tweet.
**1** 1,344 minus 228 candidates who have not received any tweets directed at them during the studied time period.
# **Key terms and what they mean**
#### Online violence
Threats and expressions of intent to perpetrate violence or abuse online, trolling, cyberbullying, hate speech directed at members of an identifiable group, actions aimed at damaging reputations, or non-consensual distribution and doxing of intimate images (also known as revenge-porn).
#### Trolling
"Inflammatory, extraneous, or off-topic messages" with the goal to gain fame by provoking other users and disrupting on-topic conversations (*Phillips, 2015*).
#### Cyberbullying
Is a form of bullying that happens online (*Kowalski et al., 2014*) where the main goal is usually an intentional attack and intimidation of a victim via computer-mediated means directly or indirectly (*Lohmann, 2014*).
#### Expressions of hate speech
On social media are typically directed towards a wider collective of people based on their religious affiliation, ethnicity, race, nationality, sexual orientation, gender, gender identity, disability, or other types of shared group characteristic (*Costello et al., 2017*).
#### Toxic & Insulting messages
For the purposes of this research, we focused on two common characteristics of online violence: the presence of toxic and insulting messages. A message is **toxic** when it is rude, disrespectful, or unreasonable. A message is **insulting** when it is inflammatory or negative toward a particular person or a group of people.
### **What we found**
What is the prevalence of toxic and/or insulting messages targeting political candidates?
In total, there were **307 (8.4%) toxic and 101 (2.8%) insulting tweets**. While most toxic tweets expressed partisan opinions using strong language, it is the insulting tweets directed at candidates that had an especially negative undertone directed at a specific individual(s); tweets such as:
This result is in line with related studies that examined toxic-type messages targeting the members of parliament in the UK and showed that the overall percentage of toxic messages on Twitter was under 10%. For example, one study found that 9.8% of tweets targeting British MPs were uncivil (*Southern and Harmer, 2019*), while another found that less than 4% of tweets directed at British MPs were abusive (*Gorrel et al., 2019*). Similarly, other studies found **swearing, dismissive insults, and abusive words to make up around 3% of online communications** more broadly (*Mead, 2014; Subrahmanyam et al., 2006*).
# **What we found**
Is there a difference in frequency of toxic messages directed at women versus men candidates on Twitter?
The chi-square test confirmed that there was no significant association between gender and receiving a toxic or insulting tweet. The test was performed on the sample of **3,635 tweets** (3637-2), containing **868 tweets targeting women candidates** and **2,767 targeting men candidates**. Note: Prior to running the test, we excluded two candidates who self-identified as non-binary due to the insufficient number of observations for statistical testing (a qualitative approach will be used instead to examine these two cases in our future work).
On the one hand, this is in line with a related study by Gorrel et al. (*2019*) who found that the "burstiness" of abuse on Twitter does not depend on gender in the context of Twitter use by the UK's MPs. On the other hand, in a different study, Southern & Harmer (*2019*) found that **women MPs in the UK were more likely to receive uncivil, stereotypical or tweets questioning their position as politicians** than their male counterparts.
Yet, in an earlier study about the UK's MPs, Ward & Mcloughlin (2017) found that **men were more likely to receive "abusive" messages on Twitter than women MPs**. These somewhat contradictory results are likely due to the differences in the study context (different countries), study population (candidates vs members of parliament), different time frames, and different approaches to defining toxic, uncivil and abusive messages.
# What this means for the election
Even though the overall percentage of toxic and insulting tweets was relatively low (under 10%), it is not necessarily their quantity, but also their severity which may negatively impact one's well-being, potentially leading to reduced online participation and withdrawal. This is particularly important considering the growing expectation for politicians to maintain an active online presence and engage the public on sites like Twitter, even if faced with toxic and insulting messages daily.
Some candidates tend to receive more extreme cases of online violence, such as the case of a hate campaign targeting a former environment minister and 2019 candidate Catherine McKenna, who received numerous online (and offline) threats (*Mia Rabson, 2019*); however, that variation is not necessarily explained by the candidate's gender. This study found that women candidates did not necessarily receive more tweets that are toxic or insulting. Since we only tested a sample of tweets (1%), future work will examine the full dataset to see whether our
findings apply to the broad set of tweets that were aimed at candidates in the 2019 Federal Election.
And while platforms like Twitter are starting to offer new features to report, block or hide hostile individuals or tweets, it may be viewed as undemocratic by the public if politicians or elected officials rely on such features to block their critics. In this context, social media platforms ought to take a more proactive role in preventing online harassment campaigns against their users. A case in point is the fact that out of 307 (8.4%) toxic and 101 (2.8%) insulting tweets flagged by our coders, the majority of these posts (255 toxic and 85 insulting tweets) are still publicly available as of January 2, 2020. Our future work will examine the trade-offs between individual's vs social media platforms' responsibilities to deal with online violence, because if left unchecked, the proliferation of toxicity and insults would undermine the capacity of platforms like Twitter to support deliberative and networked democracy (Loader & Mercea, 2011).
# What's next
- Even as more and more academic studies are starting to examine this area of research, there is still a need for standardization of definitions and approaches. For instance, because different researchers measure different things, it hampers our ability to conduct comparative studies and develop a theoretical foundation. To this end, the current study offers a "gold standard" dataset (*Gruzd et.al.*, 2020), methodological framework, and a codebook (see our online Appendix) that other researchers can rely on when identifying and studying toxic and insulting messages during election campaigns on Twitter.
- Going forward, we plan to apply a machine learning approach to analyze the full dataset of tweets in a systematic and automated way. We will also apply topic modelling techniques to examine differences and similarities between the types of toxic and insulting messages directed at women versus men candidates on Twitter. Finally, we will expand our work to other popular social media platforms like Facebook and reddit.
# Trolling on the Campaign Trail: How Candidates Experience and Respond to Online Abuse
#### **Heidi Tworek**, 1
University of British Columbia, Assistant Professor of History, heidi.tworek@ubc.ca
#### **Chris Tenove**,
University of British Columbia, Postdoctoral Fellow, cjtenove@mail.ubc.ca
#### **Trevor Deley**,
University of Ottawa, PhD Candidate
#### **Grace Lore**,
University of Victoria, Sessional Instructor
# **What we wanted to know**
Candidates in the 2019 federal election used social media to find, persuade and mobilize potential voters. Candidates also faced insults, threats, and other forms of online abuse. But little is known about the extent and impact of online abuse in Canadian politics. This study helps fill that gap by investigating the negative messages that candidates face, from mild insults to threats and hate speech. It also examines how candidates experience and respond to different forms of negative messaging and their strategies for responding to it. Such knowledge can inform policies to promote fairer political participation and healthier political discourse.
#### **The study addresses these research questions:**
How do candidates assess different types of negative messaging on social media platforms, and what strategies do they use to respond to them?
What quantity and intensity of negative messages do candidates receive on Twitter, according to manual coding of messages?
By developing a machine learning model, can we measure the quantity and intensity of negative messages directed at all candidates in the election, and do these vary according to candidates' party affiliation or perceived gender or race?
# **How we did it**
Interviews: over 25 candidates and campaign staff, to learn how they assess and respond to different types of negative messages on social media.
Tweet collection: Twitter Search API, collected all tweets at candidates that were announced by their party by August 9th, between August 9th and October 31.
Content analysis: hand-coded close to 2,000 tweets, including 1200 randomly-selected tweets, to create a data set of different types of negative messages, and to gain an approximate sense of their frequency.
Machine Learning: Using the coded tweets, we are training an algorithm to identify and categorize all negative messages directed at candidates on Twitter. We then analyzed variations in the volume and intensity of negative tweets directed at candidates.
Tworek team members: Jordan Buffie, UBC, MA student in Political Science. Jaskiran Gakhal, UBC, BA Student in Political Science, Sonya Manuel, UBC, BA Student in Political Science and Psychology. Gregory Eady, University of Copenhagen, Assistant Professor of Political Science. Maite Taboada, Simon Fraser University, Professor of Linguistics. Erin Tolley, University of Toronto, Associate Professor of Political Science **1**
## **Key terms and what they mean**
#### Negative messages
Various forms have been studied including abuse (*Amnesty International 2018; Greenwood et al 2019*), harassment (*Nadim & Fladmoe 2019*), hate speech (*Pintak et al 2019*), and incivility (*Rheault et al 2019; Southern et al 2019*). There are significant differences in how these terms are understood and measured (*Mishra et al 2019*). Furthermore, studies tend to focus on extremely negative messages, but moderately negative messages (possibly at high volume) may also affect public figures and political discourse. Rather than focus on one form of negative messaging, we addressed this methodological and definitional challenge by examining how candidates perceive different levels of negativity.
#### Intensity of negativity
We categorized messages as low, medium, or high negativity. We developed this scale by asking candidates and campaign staff to evaluate sets of sample tweets that we shared with them.
# **What we found**
How do candidates assess different types of negative messaging on social media platforms, and what strategies do they use to respond to them?
Our interviews revealed that many candidates and their teams grappled with significant levels of negative or abusive online messages. In general, they identified low negativity messages as dismissive, uncivil, or mildly insulting; medium negativity messages as highly insulting (often incorporating swear words); and high intensity as threatening, hateful, abusive, or making serious unproven allegations of misconduct. Candidates did not consider all criticisms of their views or actions as inappropriate, and many argued that democracy requires elected officials to be open to criticism. Some reported that abuse was frequently related to their identity, including sexist, racist, or Islamophobic content. Our future research will investigate these claims quantitatively, using our machine learning model to assess tweets directed at candidates.
Candidates and their staff took different approaches to addressing negative messaging, drawing on extremely varied levels of knowledge, training, and resources. Most developed these strategies reactively as threats or abuse arose.
Candidates and their teams described the psychological toll of negative messages on candidates themselves, on staff members who help manage social media accounts, and on members of the public who follow the candidates' social media. These concerns justified their decisions to hide content or block accounts on Twitter and Facebook.
What quantity and intensity of negative messages do candidates receive on Twitter, according to manual coding of messages?
Our analysis of 1200 randomly-selected tweets found that over 40% of these were negative (see **Figure 1**). While less than 1% of these issued threats or made hateful remarks (high negativity), approximately 13% were insulting (medium negativity) and about 27% were dismissive, uncivil, or mildly insulting (low negativity).
Importantly, negative messages were not always negative towards a candidate. Messages could be directed at a candidate on Twitter but actually attack other individuals or groups. Doing so enables users to reach larger audiences for their negative messages. Research and policy responses should also address this indirect targeting of negative content.
Figure 1. Distribution of sentiment labels by percentage for 1200 randomly-sampled tweets directed at candidates during the 2019 Canadian election.
# **What we found**
What quantity and intensity of negative messages were directed at all candidates in the election, and do these vary according to candidates' party affiliation or perceived gender or race?
We have developed a machine learning model to assess the negativity of tweets. The model was "trained" on the 1200 randomly-selected tweets we coded. Our preliminary findings using this model suggest that similar proportions of negative tweets – as shown in Figure 1 – may apply to our full set of almost 3 million tweets directed at candidates during the election campaign.
Please see the Online Appendix for more detailed findings using this machine learning set, including the different quantities and intensities of tweets directed at candidates of different party affiliation or perceived gender or race.
A major limitation for the quantitative study was lack of access to Facebook comments. Many interviewees identified Facebook as the most important platform for campaigning, but also the source of more localized threats and abuse. A second limitation is that, while we interviewed a diverse range of candidates from the Conservative, Liberal, Green, and New Democratic parties, we did not conduct interviews in French and have not yet interviewed Bloc Québécois or People's Party candidates.
At the end of the day, my concern is not to avoid criticism. My concern is that people may effectively ruin the community and the engagement that I've worked hard to build, and that they could push other people off the platform because it's such a polluted platform."
### What this means for the election
- All interviewees saw online harassment—and online incivility more broadly—as negatively affecting Canadian democracy. Candidates and their teams face very different levels of negative messaging, both in quantity and intensity. However, no interviewees stated that negative messaging derailed their campaign.
- Negative messaging is often seen as a barrier to entry for people from marginalized groups, particularly when deciding whether to participate in politics. Candidates reported that abuse was often related to their identity, including sexist, racist, and homophobic content. In the
- context of underrepresentation of women, Indigenous, LGBTQ\*, and racialized Canadians, ongoing attention to equity of representation, participation, and safety of candidates is important.
- Knowing who is behind negative messages content could better equip candidates to deal with them. For instance, some respondents said that they would like to block bots and fake accounts for problematic behavior, but not accounts of constituents. Many interviewees claimed that campaign teams could have used more training on effective responses to highly negative messages.
### What's next
- Political parties should develop best practices for responding to incivility and threats (e.g. when to block, hide, delete, involve police). Parties should provide support and resources to candidates and their staff, including training on how to best use social media tools. This assistance is particularly important for under-resourced campaigns and for minority candidates, women, and others whose identity makes them more likely to be at risk of abuse online and offline. Finally, parties may wish to develop codes of conduct to prevent members from contributing to online harassment.
- Platforms should do more to help candidates identify bots and inauthentic accounts, and continue to develop features for users to proactively reduce harassment and improve productive discussion.
Facebook should increase data availability to researchers. We will have further recommendations for Twitter once we have used our machine learning model to analyze the full set of tweets at candidates.
- Policy-makers should consider whether current legislation (including on harassment, libel, and hate speech) sufficiently addresses social media platforms. If the platforms respond insufficiently to demands for increased transparency of data and users, this may need to be mandated by the federal government.
- Future research: We will continue to investigate the different quantities and intensities of negative messages directed at candidates, to understand the factors that make people more or less likely to receive abuse. Building on these findings and our interviews, we will examine how negative messaging affects candidates as public communicators and electoral contestants.
# Rise of Islamophobic Emotional Rhetoric During the 2019 Canadian Federal Election
#### **Zeinab Farokhi**,
University of Toronto, PhD candidate zeinab.farokhi@utoronto.ca
# **What we wanted to know**
This study examined the rise of hate-centered politics pertaining to Muslims and the Liberal Party of Canada before and during the 2019 federal elections. In the wake of federal elections, the Liberal Party and the incumbent Prime Minister, Justin Trudeau, have been repeatedly associated with Muslims, mainly by right-wing extremist groups and media personalities on Twitter (*Patriquin, 2019*). The study presented here reveals that Islamophobic discourse was heavily employed as a political tactic to denounce Trudeau and the Liberal Party. Islamophobic discourse was not associated with messages critical of the Conservative Party but specifically used in anti-Liberal Party messaging. Further, the Conservative Party and its leaders were presented as the alternative in order to end the "Muslim menace" allegedly being brought into Canada by Trudeau. Hence, this study focused on the nexus between online anti-Muslim rhetoric and the Liberal Party of Canada. **The following research questions were the point of departure:**
**RQ1a:** Did the frequency of tweets connecting Trudeau with Muslims or perceived "Muslimness" increase over time during the run-up to the federal election?
**RQ1b:** What was the sentiment of those tweet?
**RQ1c:** Did the sentiment of those tweets change over time?
**RQ2a:** Which discourses were most commonly used in tweets mentioning Muslims and Trudeau?
**RQ2b:** Which discourses were most commonly used in tweets mentioning the Muslim Brotherhood and Trudeau?
### **How we did it**
Approximately **40,000 English language tweets** generated in Canada were collected using Crimson Hexagon from May 1 to December 1, 2019.
Search parameters: I used the following keywords: "**Muslims**" AND "**Trudeau**"; and "**Muslim Brotherhood**" AND "**Trudeau**."
**Digital ethnography** (*Postill & Pink, 2012*) was used to capture the general conversation about Muslims in Canada between May 1 and December 1, 2019.
The tweets were analyzed using Critical Discourse Analysis (*Van Dijk, 1998*) to identify **salient narratives concerning Muslims, Islam, and Muslim immigrants** in Canada.
Sentiments related to Muslims were captured using Crimson Hexagon's Sentiment Analysis tool, although it is worth noting that software lacks accuracy in capturing satirical content.
# **Key terms and what we found**
#### Sentiment
Any tweet is generally an expression of opinions or emotions. It can be positive, negative or neutral.
#### Sentiment Analysis
The process of analyzing and categorizing a portion of text or tweet to extract the context,-e.g., identifying hate speech in tweets. Crimson Hexagon was used to measure the sentiments.
#### Time periods
Tweets collected before, during, and after the election.
#### Salient narratives on Muslims
That Muslims are the propagators of terrorism and terror-related activities, owing in large part to their religion (Islam) which is thought to be steeped in "barbarism." As such, Muslims are constructed as both "regressive" and "unaccommodating," and therefore at odds with the core principles of 21st-century Western democracy (*Awan, 2016*).
#### Islamophobia
According to Zine (*2008*), Islamophobia refers to "a fear or hatred of Islam and its adherents that translates into individual, ideological and systemic forms of oppression and discrimination" (*p.6*).
#### Posters
Twitter accounts which posted tweets using the keywords during the sampling period.
#### Rightwing Extremism (RWE)
Perry & Scrivens (*2019*) suggest that "RWE in Canada is a loose movement, characterized by a racially, ethnically, and sexually defined nationalism. This nationalism is often framed in terms of white power and is grounded in xenophobic and exclusionary understandings of the perceived threats posed by such groups as non-whites, Jews, [Muslims], immigrants, homosexuals, and feminists" (*p.821*). For this study, Twitter accounts like "Act for Canada" and "Patriot North" and individuals who indulge in disseminating Islamophobic rhetoric on their timelines were selected.
# **What we found**
In response to **RQ1a**, the study results show that while the volume of conversation around "Muslims and Trudeau" (see **Figure 1**) and "Muslim Brotherhood and Trudeau" was mostly stable from May to August, these figures doubled in September, reaching their peak in October and then dramatically decreasing in the post-election period. It appears anti-Muslim content on Twitter that was directed at the Liberal Party and Trudeau gradually rose during the run-up to the 2019 election and decreased dramatically after the election. This may indicate that anti-Muslim sentiment was used as a tool to influence the voters during elections.
Indeed, in response to **RQ1b**, the sentiment of tweets connecting Trudeau to Muslims from May through November was mostly neutral followed by negative. However, most of the satirical tweets were interpreted as neutral by the software used for analysis which may underestimate the level of negativity. Ultimately, the overall sentiment of the tweets analyzed for this study can be stated as largely neutral to negative, with positive tweets a small minority.
In response to **RQ1c**, the negative sentiment (shown in red in **Figure 1**) related to "Muslims and Trudeau" and "Muslim Brotherhood and Trudeau" increased during the two-month period prior to the 2019 elections and rapidly decreased thereafter, indicating that negative emotions were prominent in the months preceding and during the elections.
Figure 1. Volume and sentiment of tweets related to Muslims and Trudeau Sentiment Analysis for keyword: Muslim
# **What we found**
In response to **RQ2a**, much of discourse used by the posters using the keywords "Muslims and Trudeau" was explicitly racist, anti-Muslim, and anti-immigrant, painting Trudeau as a "Muslim sympathizer" who intended to turn Canada into a Muslim country.
**Here are examples of such tweets:**
Trudeau keeps company of rapists, murderers, Muslim radicals, terrorists, pedophiles, in fact any disgusting criminal he can find.
"The brother of the 1st female Palestinian member of Congress is bragging about Islamic dominance over Canada." Rashida Talib's brother says "Canada will soon be a Muslim majority country." This is Trudeau's Canada #TrudeaMustGo #Scheer4PM
ATTENTION!!!! ANDREW SCHEER !!!! STOP!!! STOP !!!! STOP!!! ILLEGAL BORDER CROSSINGS!!!! AS CANADA IS SICK OF BEING HANDED OVER TO TERRORIST!!! MUSLIMS!!!! JIHADEST TRUDEAU!!!! IS TURNING CANADA INTO A MUSLIM COUNTRY!!
In response to **RQ2b**, much of discourse seen in tweets using the keywords "Muslim Brotherhood and Trudeau" centred on a narrative that Trudeau was a "pawn" of the terrorist organization Muslim Brotherhood who must be deposed, making way for Conservative Party leader Andrew Scheer or far-right People's Party leader Maxime Bernier.
**Here are the examples of some tweets:**
Trudeau wants to breakup Canada and make it a muslim state and be ruled under muslim law! Trudeau supports terrorists and the muslim brotherhood over Canadians! ANDREW SCHEER IS OUR ONLY HOPE TO SAVE CANADA'S DEMOCRACY!
Under the guidance of prominent MP's in Liberal Party foreign influence by Qatar, Muslim Brotherhood are influencing MSM and politics! Beware! Trudeau is bought and paid for!
Any vote for Justin Trudeau is a vote for the Muslim Brotherhood gangs. Vote wisely! #TrudeauMustGo #terrorists #muslimbrotherhood #ShariaLaw #M103
These findings suggest that anti-Muslim rhetoric is weaponised as a tactic against the Liberal government, which is perceived as an ally to Muslims. The discourse pertaining to both sets of keywords reveals that the Liberal Party in general, and Trudeau more specifically, are portrayed as "Muslim," "Islamist," and/ or "terrorist" sympathizers. Hashtags like #TrudeauMustGo, #TrudeauIsDone, and #JihadistTrudeau worked to create a narrative that anything Muslim is "vile" and "vicious." As such, the narrative goes, anyone who is determined to be associated
with Muslims should therefore be renounced from Canadian politics for the sake of "saving" Canadian democracy from the imagined Muslim "threat." Most of the tweets portray the Liberals as "pro-Muslim," and claim that Canadian multiculturalism has led to a Muslim "takeover" of Canada. According to these users, the solution to this ongoing "Muslim problem" is not only continuing to establish anti-Muslim and anti-immigrant policy, but also eventually electing a populist party into power.
# What this means for the election
- The study shows that Islamophobic emotional rhetoric increased rapidly prior to and during the 2019 Canadian election. Closer to the election, particular rhetoric targeting Muslims and Muslim immigrants in Canada was employed. Such discourse may give rise to irrational fears and foster prejudices against Muslims in Canada.
- These findings on the 2019 election are crucial as they map the increase in Islamophobic rhetoric on Twitter prior to elections. Indeed, anti-Trudeau political activists
use narratives to target and "otherize" Muslims, thereby exploiting anti-immigrant sentiments and appealing to fears that have been stoked in the current political climate, in order to mobilize voters against the then-Liberal government. Although this research did not focus on the mobilization of voters on the ground, it shed light on the rise of hate-centered politics during the election within Canadian online spaces. Further research would help policymakers and think tanks focus on addressing this issue.
### What's next
- Further research into who is tweeting and sharing this content online is needed. Initial investigations suggest right-wing extremist groups often share this content but more detailed analysis is needed. By examining Canadian online discourse around Muslims, policymakers and public officials can understand how right-wing extremist propaganda utilizes elections to capitalise on Islamophobic messages and sentiments so as to normalize hatred and undermine the integration of immigrants, especially Muslims, within Canadian society.
- Future research should focus on how extremists (whether right-wing or left-wing) use social media as a weapon to propagate and amplify extremist discourse, hate rhetoric, and widespread disinformation. Such future projects can inform Canadian educators and policymakers as to how Islamophobia is being shaped via cyberspace, while also allowing them to gauge the security threat posed by domestic right-wing extremist groups. Given the rise of populist governments in the U.S., Europe, and India—whose political narratives find a stronghold in otherizing, demonizing, and dehumanizing minorities (e.g. Muslims)—it is important to understand, monitor, and prevent such narratives from being amplified across and through various transnational and cyber platforms.
# Public Opinion and Political Participation
# What you need to know
As the Internet has become a fixture of modern life, it has contributed to "new ways of knowing, new strategies for gathering, storing, retrieving, and utilizing information" (*Dahlgren, 2004, p. xv*). In recent years, social media has grown exponentially, and people have correspondingly shared more information, interests and opinions (*Khan et al., 2014*). This has created a wealth of data that can be analyzed and used to identify people's opinions on a variety of topics, including their political attitudes and preferences.
Originally, public opinion research—research conducted to help understand and represent the public's views (*Perrin & McFarland, 2011*) —could only be done using in-person, face-to-face methods (*Steenkamp & Hyde-Clarke, 2014*). But now, the rise of the Internet and social media have created new and significant sources of information on public opinion (*Bachner & Hill, 2014*). Social media as a source of public opinion is more affordable and sometimes more readily accessible, but it is limited in generalizability and requires new skills to collect and interpret (*Dubois et al., 2018*).
The evolution of modern technology has also changed public discourse and the way people participate in it. There has been optimism about the Internet and related tools increasing public discourse and political participation. Brants, for example, explains that "because of its horizontal, open, and user-friendly nature, the Internet allows for easy access to, and thus greater participating, in the public sphere" (*2005, p. 144*), and others identify social media platforms as a main way to engage citizens (*Steenkamp & Hyde-Clarke, 2014*).
Yet, there is reason for caution. Online political communication tends to be more targeted and more private, and therefore can lead to fragmentation and can be less transparent (*Bennett & Lyon, 2019*). Social media use may also serve to reinforce existing views and habits by only attracting people who are already engaged in public discourse, rather than people who would otherwise be disengaged - the evidence is mixed on this point (*Dubois & Blank, 2018; Guess et al., 2018*). Further, some researchers have found that social media tools, rather than empowering the broader public, tend to benefit already-established actors like incumbent politicians and journalists (*Hawthorne et al., 2013*).
# What we still don't know
How digitally literate is the Canadian population? Do internet users intuitively distinguish between trustworthy and non-trustworthy sources? What factors play into whether Canadians trust the information they see online?
What is the content and style of the online public discourse? These forms of online participation are constantly evolving and so this question remains crucial.
How do Canadians use newer social media platforms to engage with political content? Much research focuses on Twitter and Facebook because they are the most established and most popular social media in Canada and because it can be easier to get data from and about them. But many other tools are used by voters and other political actors.
# What you'll learn from the projects
Memes play a key role in online discussion of politics, but their spread was limited, and their content was not generally reflective of the final results of the election. (*McKelvey, et al.*). **[PAGE 44](#page-43-0)**
Political discourse on Instagram most often involves people engaging with their already-existing networks, rather than with content created and promoted by political actors (*Wilkinson*). **[PAGE 48](#page-47-0)**
The presence or absence of information about source, methodology, and margin of error has little effect on subjects' interpretation of polling data. (*Daoust and Bastien*). **[PAGE 52](#page-51-0)**
# Memes in the 2019 Canadian Elections. That's it. That's the Headline.
#### **Fenwick McKelvey**,
Concordia University, Associate Professor fenwick.mckelvey@concordia.ca
#### **Jeremy Blackburn**,
Binghamton University, Assistant Professor
#### **John F. Gray**,
Mentionmapp Analytics Inc.,
#### **Scott DeJong,**
Concordia University, MA student
#### **Janna Frenzel**,
Concordia University, PhD student
#### **Quinn Valencourt**,
Concordia University, PhD student
### **What we wanted to know**
When Canadians communicate online, they share pictures. Some of these pictures are of friends, families, or maybe food. Other pictures are memes. Memes refer to "evolving tapestries of self-referential texts collectively created, circulated, and transformed by participants online" (*Phillips & Milner, 2017, p.30*). Memes help people express their emotions, share jokes, be fans, and, notably for our study, discuss politics. Canadians make memes about politics, but there are limited studies about Canadian political memes (*Lalancette & Raynauld, 2019; Piebiak, 2014*). Internationally, we know that memes are an important part of online mobilization and help to build political affinities (*Milner, 2016; Mina, 2019; Nissenbaum & Shifman, 2018*), yet there have been no large-scale research projects studying memes during a Canadian election.
#### **We sought to explain:**
If Canadians created and shared political memes during an election; and, if so,
How did memes encode Canadian political matters including issues, party loyalty and leader's brands?
# **How we did it**
Manual data collection: Student teams identified and monitored 15 meme sharing or known partisan Facebook pages and groups from 28 September 2019 to 28 October 2019.
Automated data collection: Facebook analytics from Crowdtangle (supplemental).
Content analysis: Preliminary coding during data collection also identified tone toward featured leaders as well as issues mentioned in memes.
# **Key terms and what they mean**
#### Memes
Graphics that have an 'internet ugly' aesthetic and use common references and fonts associated with internet culture rather than graphic design conventions (*Douglas, 2014*). Memes are "(a) a group of digital items sharing common characteristics of content, form, and/ or stance, which (b) were created with awareness of each other, and (c) were circulated, imitated, and/or transformed via the Internet by many users" (*Shifman, 2014, p. 41*). There is no one agreed upon definition.
#### Constitutive humour
Creates collective identities as an in-group phenomenon where people come to share references and values to be in on the joke. By laughing at others, constitutive humor can establish a collective us and an othered, laughed at, them *(Milner & Phillips, 2017*).
#### Astro-turf
The presentation of advertising and public relations as usergenerated content, the opposition of grass roots support.
#### Identity play
The multitude of ways people perform their identity to others, where the identities and lives of others modulate how an individual presents themselves in different contexts (*Philips & Milner, 2017*).
#### Political parallelism
The reflection of political orientations through media systems, where political division is reflected in the content that users see, and further propagated by highly political journalism (*Fletcher et al., 2019; Hallin et al., 2004*).
# **What we found**
Canadians shared memes during the election, or at least partisans did. Memes represented 27.7% of the total content in our sample from third-party advertisers and seemingly user-generated Facebook pages. The memes' style and sample differed by group. We identified three main types of memetic styles:
Astroturfers or third-party advertisers shared memes to appear as user-generated content as shown in **Figure 1**;
Hang-outs where partisans shared memes referential to popular culture and internet aesthetics as shown in F**igure 2**; and,
Enclaves that shared boutique, political memes not referring to Internet culture as shown in **Figure 3**.
People seem to share memes as constitutive humour and collective identity play. Some third-party advertisers engage in similar practices while trying to be more intentionally persuasive.
# **What we found**
Trudeau's character, from the study's preliminary results, was a source of constant criticism as seen in **Figure 4**. Memes mentioning Trudeau were mostly negative (672 out of 697 memes coded).
Figure 4. Frequent issue in sample
Notably, Scheer and Singh did receive more positive coverage, as seen in the **Figure 5**. The number of pro-Singh memes is greater than that of negative memes, another signal that press coverage of Singh's popularity late in the campaign did not translate into votes (*Hui & Dickson, 2020*).
Memes did not circulate widely. Preliminary image cluster analysis found few memes that spread between multiple groups. The study showed a split between two entirely different canons of memes, one created and shared by anti-Trudeau groups and the other created and shared by Liberal and New Democratic groups.
# What this means for the election
- Canadians made memes during the election and those memes matter for our political culture, but not for winning elections.
- Memes did not predict the outcome of the election. The overwhelming unpopularity of Trudeau within our sample did not signal his electoral loss. Nor did memes seem an effective form of persuasion. Conservative Canada Proud, where memes made up 20% of its top posts, did not lead to massive growth of membership larger than its Ontario base. It grew by only 13,000 followers over the course of the campaign to 162,346 as opposed to the roughly 400,000 followers of Ontario Proud. By contrast, progressive North 99 hardly used memes at all in its communication strategy.
- Memes matter a lot for partisan politics. Partisans joked and commiserated through memes throughout the campaign. Indeed, memes offer a unique window into partisan culture and indicate partisanship is an important factor in political engagement. New Democratic and Liberal memes joked about policy, their leader, strategic voting, and, for Liberals after Trudeau's Brownface scandal, a chance to process party loyalty and disappointment.
- Memes, to a lesser degree, indicate Canada's political parallelism with partisans having different interpretations of recent history. Some versions are playful, conspiratorial, and sometimes both. The 'baggage' plane in **Figure 1**, for example, adds a debunked conspiracy involving the late Fidel Castro (*Brown, 2016*). Fidel's face is seemingly equivalent to Trudeau's other 'baggage'. Similarly, the Conservative Party's press release questioning Trudeau's record at West Point Grey Academy and the deceptive Buffalo Chronicle story drew on long-standing questions in anti-Trudeau meme groups about Trudeau's past sexual conduct that predate these stories (*Lytvynenko et al., 2019*) and were similar in style to other conspiracy theories (*Hofstadter, 1964*).
- We faced major challenges around access to data that limited comparison and analysis. First, we had no direct access to Facebook analytics for most of our data collection. Instead, all data collection was manual, labour-intensive and error-prone. Second, methods to analyze visual political communication at scale are still in development and did not reach production during our study.
# What's next
- Improve definitions of political advertising to protect free expression through memes. Memes are an outlet of political culture and free expression that should be protected from being designated as advertising. The risk is that astro-turfing groups mimicking internet culture may use memes as part of their campaigns, confusing free expression with advertising. Elections Canada and the Elections Commissioner should clarify the definition of political advertising to protect user-generated content while improving reporting and transparency obligations for thirdparties using social media to appear as grassroots movements.
- Ensure platform content moderation respects fair dealing and the Non-commercial Usergenerated Content clauses of the Copyright Act to protect that political expression through memes. Without these protections, memes could be seen as a form of copyright infringement since they re-purpose copyright materials like television shows, movies and other cultural goods.
- Improve researchers' access to social media data and invest in on-going archiving. We do not know how memes were used in past elections as the data is non-existent and our data has no long-term solution for archiving. Tenove, Tworek and McKelvey (2018) have called for "the creation and use of social media companies' research repositories" that aggregate data from multiple social media platforms. Similar to medical records, these repositories should have secure and controlled access and should be managed by government or a responsible, neutral third-party on an on-going basis.
# Say it on the 'gram: Political Engagement on Instagram during the 2019 Canadian Federal Election Campaign
#### **Sabrina Wilkinson**,
Goldsmiths, University of London Doctoral Researcher sabrinawilkinson@live.com
### **What we wanted to know**
Social media has become a valuable tool for politicians and political parties in recent years. Many voters use these platforms to access information about candidates and their policy positions and engage in political dialogue. Numerous studies have examined how and why these phenomena take place on Facebook and Twitter (*Goodman et al., 2011; Raynauld & Greenberg, 2014; Stier et al., 2018*). But there is less research that investigates how voters use Instagram to engage with political information, especially in Canada (*Lalancette & Raynauld, 2017; Munoz & Towner, 2017*). This gap in research exists even though most major Canadian federal political parties and leaders have substantial followings on Instagram, and younger voters especially are regular and active users (*Pew Research Center, 2019*).
#### **To address this gap, this study asks:**
During the 2019 Canadian federal election campaign, how and why did Canadians engage with political information on Instagram?
# **How we did it**
Online panel survey of Canadian citizens 18 years and older, who expected to vote and access political information related to the election on Instagram, using the web-based survey tool Qualtrics.
The survey was run in French and English. Responses were collected from a subset of 208 respondents from October 4th to 13th, from a full sample of 1545.
In the full sample, there were national quotas for language, region, gender, and age to adequately match Canadian census data. There were also within-region "flex" quotas for age and gender.
The small size of the subset means the data included in this study was not nationally representative but results were useful initial indicators.
Descriptive statistics were used to analyse the data.
## Key terms and what they mean
#### (i) Political information
Includes Instagram posts made in support of a candidate by a family member, friend, public figure, or political party, or any other type of content that relates to the 2019 federal election found on the platform.
#### Access to political information
Instances where respondents passively received political information about the election on Instagram, including seeing related posts made by friends and family members, or political advertisements.
#### ☐ Engagement with political information
Respondents' participation in the dialogue about the election on Instagram.
### What we found
Ultimately, there were two key findings from this study. First, this research showed that respondents' engagement on Instagram related to the 2019 federal election more often entailed users' interaction with others in an existing social community, rather than with professional accounts designed to foster political participation. Second, despite a primary motivation being a desire to become a more informed voter, respondents generally outlined more than one rationale behind their political engagement on the platform.
#### Methods of engagement
The specific ways that respondents engaged with political information related to the election on Instagram, such as posting election-related content or liking, sharing, or commenting on posts made by political parties involved in the election.
#### Motivations for engagement
The rationales behind respondents' use of Instagram to engage with information related to the election. Examples include the desire to engage with others on Instagram who share the same political views, or as a way to pass time.
To the first key finding, this study showed many respondents used Instagram to engage politically on issues related to the vote (**Figure 1**). Just over three quarters of respondents used at least one method of engagement. But while these Canadians engaged with political information in a variety of ways, they were more likely to interact with people with whom they had existing relationships (27%, N = 208) than organizational accounts that made posts about the election, such as political parties (13%, N = 208) and non-partisan organizations (10%, N = 208). This divergence signals the important role that social ties played in respondents' communications around the 2019 federal election on Instagram.
30% O. O (O.) Interacted with Interacted Interacted with Posted related Sent related Interacted Interacted Interacted with Other None of the related posts content with posts with posts with related related posts made direct messages related posts made above made by political made by people by journalists made by political nosts made by non-partisan they know and other candidates organizations by news outlets public figures
Figure 1. How respondents engaged with political information related to the election on Instagram
Methods of engagement
### What we found
This study's second key finding is that respondents' political engagement on Instagram was prompted by a range of diverse motivations (**Figure 2**). Some of these rationales were related to respondents' political lives ("to engage with others on the platform who share their political views"), while some were linked to social pursuits ("as a form of entertainment"). Yet,
there was some variance in respondents' alignment with certain rationales for engagement. Just over a third of respondents "strongly agreed" that their engagement was motivated by a desire to become a more informed voter. This was more respondents than any of the surveyed motivations' respective levels of agreement (strongly disagree to strongly agree).
Figure 2. Why respondents used Instagram to engage with political information related to the election
Responses of 'not applicable' and 'I don't know' were excluded from this graph
### What this means for the election
- This study gave initial indications that offline social relationships between Instagram users may drive political engagement on the platform more readily than professional communicators, including members of political candidates' social media teams. While further research is needed to determine the extent these respondents' behaviours are reflected nationally, this finding does highlight the possibility that social capital may play a greater role than political, industry, or third-sector financial resources in political engagement on Instagram.
- Respondents' wide-ranging motivations for political engagement could signal the prospect that such activity is not prompted by a single desire or use.
Rather, this breadth may be brought about by a range
- of overlapping and sometimes tangential factors, to fulfill a span of different gratifications, some entirely unrelated to politics. Yet, it is also notable that respondents were most aligned with the suggestion that their engagement was motivated by a drive to increase political knowledge. This finding could suggest that, despite the breadth of content on the election available across Canada's media environment, respondents still felt as though they lacked a complete picture of their political options.
- As the subset of data that informed this study was not nationally representative, this study was limited in that it could only provide initial indicators on how Canadians might use Instagram to engage politically.
### What's next
- Civil society groups whose mandates include digital literacy should integrate a discussion of political engagement on Instagram into program materials and initiatives. As young people are more active users of the platform, these efforts should target teenagers and young voters. These initiatives might address how these groups can meaningfully develop their political knowledge and participation on Instagram, as a way to become more engaged citizens.
- Civil society groups that share nonpartisan political information online should consider how they can most effectively use Instagram to disseminate this content. This study raises the possibility that individual users may be more inclined to engage politically on the platform with people from their offline social circles. These organizations may want to consider strategies that encourage individual users to screen-grab or otherwise independently disseminate civil society groups' content, to better connect with individual Canadians
- Future research should use alternative methods to investigate the questions raised in this study. In-depth interviews, for example, would allow participants to describe in detail how they engage politically on Instagram, and what motivates them to do so. Such an approach would also allow researchers to probe respondents further on motivations that are distinct from those outlined in this survey.
- Researchers should study how Canadians use Instagram to engage with political information related to the federal government's mandate and ongoing activities outside of campaign periods. How does this engagement reflect or differ from the engagement that takes place during an election campaign?
# The Quality of Online News and Citizens' Perceptions of Pre-Election Polls
#### **Jean-François Daoust**,
University of Edinburgh, Politics and International Relations, Lecturer (Assistant Professor) jf.daoust@ed.ac.uk
#### **Frédérick Bastien**,
Université de Montréal , Department of Political Science, Associate Professor
# **What we wanted to know**
During election campaigns, citizens can access tons of political news. But how do they assess the quality of online news? This study investigates whether citizens' confidence and understanding of online news are impacted by 1) its source (i.e. who shares the news) and 2) its presentation format (i.e. how it is presented). Following the *Canada Elections Act* (*Canada, 2019*), news media must meet certain standards when they publish public opinion polls during federal election campaigns: notably, a certain amount of methodological information (e.g., the name of the pollster, the sample size and the margin of error) is required. However, previous research has shown that it is a challenge for journalists to accurately interpret margins of error when they make claims about the actual state of the race between parties (*Pétry and Bastien, 2013*). It is important to understand citizens' ability when it comes to interpreting preelection polls in online news. **We examine:**
Whether who shares information about methods and how the information is shared have an impact on citizens' perceptions of the news.
# **How we did it**
Survey experiment: embedded in the 2019 Canadian Election Study which ran from September 20th to 28th, in total there were 5329 (eligible to vote) respondents who took the survey.
Experiment: Respondents were exposed to a Facebook-style post showing a (real) pre-election poll.
We compared citizens' perceptions (reliability of the news, etc.) depending on what information they were exposed to in order to understand what the effect of news source and amount of methodological information displayed is.
The treatment that respondents were exposed to was assigned randomly and varied both on two features: the source of who disseminated the news (either CBC/Radio-Canada or an ordinary citizen named Michael Forrest) and the amount of methodological information displayed (none, some, or a lot).
Because of the random attribution, these differences are the only things that differ across the groups.
# **Key terms and what they mean**
#### Source of the online news
Individual or organization who disseminated the news. In treatments 1, 3, and 5, the election poll results appear to originate from CBC/Radio-Canada, the news media trusted by the largest proportion of Canadian population (*Brin 2019*), whereas they look like they have been shared by an ordinary citizen in treatments 2, 4, and 6.
#### Methodological information in the news
Confidence intervals and methodological note about the survey, as required by the Canada Elections Act. In treatments 1 and 2, no methodological information is displayed. In treatment 3 and 4, the margin of error is graphically shown. In treatment 5 and 6, the margin of error is graphically shown along with a detailed methodological note. **Figure 1** displays treatments 1 and 6 to illustrate these differences.
Figure 1. Treatments 1 and 6 in the survey experiment
## Perceived reliability
Confidence of respondents into the reliability of the poll results that they were exposed to.
#### Perceived state of the race
The party, if any, citizens believe is ahead according to the pre-election poll results that they were exposed to.
#### Ban support
Opinion of respondents about the relevance of prohibiting the publication of the new pre-election polls during the last week of the campaign.
### What we found
First, a clear majority (69%, n=4693) of respondents believed that the poll was reliable. Second, although there was no party that had a statistically significant lead in the poll, about 80% (n=4685) of the sample thought one party was leading. Third, although surveyed Canadians found the information reliable, about 60% (n=3817) of them agreed with the statement that they should be banned in the last week of the campaign.
Did the source of the information, whether it is an ordinary citizen (Michael Forrest) or a credible news organization (CBC/Radio-Canada), make a difference?
As shown in **Figure 2** (panel a), it did not. There was no substantial variance in terms of perceptions of reliability and polls' restriction. It is also worth noting that it did not affect citizens' propensity to claim that one party was ahead (not shown in **Figure 2**).
Panel A Panel B 0.8 0.8 Mean (0 to 1) Mean (0 to 1) 0.7 0.6 0.5 0.5 Π4 0.4 CBC/Radio-Canada Michael Forrest No info Some info All info Source of the news Methodological information Reliable Ban polls Reliable Ban polls
Figure 2. Canadians' opinions of pre-election polls' reliability and their regulation
Did the addition of the margin of error in the graph and a detailed methodological note have an impact?
As shown in panel b of **Figure 2**, there was, again, no substantial difference on perceptions of polls' reliability, ban support
and which party was ahead (not shown). These results held among many different types of subgroups based on age, gender, education, political interest, ideology, and partisanship. For instance, the most educated citizens or the most interested in politics are not more sensitive to these treatments.
### What this means for the election
- The results should motivate us to understand why a layperson is perceived as equally reliable as a professional news media organization when providing information. This is quite odd inasmuch as CBC/Radio-Canada, featured in our experiment, is perceived as the most credible news media organization by the largest proportion of Canadians. This suggests that citizens are far from immune to "fake news" and disinformation, which can easily be shared in the digital public sphere by a layperson.
- We should be concerned that the application of the *Canada Elections Act*, that is, providing specific methodological information about a public opinion poll, did not improve citizens' capability
- to accurately perceive the state of the race. No party was ahead in the race. However, the fact that neither the inclusion of the margin of error in the graph nor the methodological information enabled citizens to more accurately interpret the poll results warns us that citizens' news literacy is problematic.
- Research testing the source of the news and/or methodological features of pre-election polls using a clean experimental design is very limited (*Kuru et al.*, 2017, 2019), and to our knowledge this is the first such study ever conducted in Canada. Our results are robust and important, but they raise major concerns regarding democratic citizens.
### What's next
- Support development of news literacy, both through formal and popular education.
Citizens have trouble assessing the reliability of online information based on its source: in our experiment, they found the information shared by the public broadcaster as reliable as the news shared by an unknown citizen.
- Support continuous training of journalists so they are equipped to interpret statistical information or other data generated by research methods from the social sciences.
Journalists play a vital role in collecting, interpreting, and assessing the reliability of such information. Our results point out citizens' lack of ability to interpret the results of election polls, especially when the race is "too close to call", that is, within the margin of error.
- Put forward public research on how citizens perceive pre-election polls. They are increasingly popular during election campaigns. Sharing polling information is very easy for any ordinary Canadian, who can cherry-pick what they want to present. Our results are worrisome for Canadian democratic citizenship and so future public research should shed light on these issues.
# Flow of Disinformation and Misinformation
# What you need to know
A strong democracy requires voters to have access to trustworthy information about their government, candidates and their promises. Traditionally news media play a key role in keeping the public informed but the spread of misinformation and disinformation online increasingly challenges this foundational role.
Misinformation is the spread of false information inadvertently while disinformation is false information intentionally and maliciously created, spread and manipulated (*Weedon et al., 2017*). The European Commission describes online disinformation as "using social media to control the political narrative or to radicalize, recruit and direct proxy actors" (*European Commission, 2015*).
False information can be created for financial, political, social and psychological motivations (*Wardle, 2018*) and spread by a variety of actors including governments, for-profit companies, social media platforms, mainstream and alternative media outlets and audiences. These actors can challenge the traditional gatekeeping function of established journalists (*Bennett & Livingston, 2018*).
The flow of false information online is a cyclical process (*Bennett & Livingston, 2018*), where the audience also becomes an agent of amplification, both wittingly and unwittingly (*Duncombe, 2019*). Mainstream media then play a role in amplifying false information, intentionally or not, including through fact-checking efforts (*Wardle, 2018*). Emotion also plays a role in increasing the flow of misinformation and disinformation online, particularly anger and anxiety (*Weeks, 2015*), as it can depress information seeking, increase selective exposure (Valentino et al., 2008), and decrease the effectiveness of fact-checking (*Duncombe, 2019*).
# What we still don't know
- Who is running disinformation campaigns? While disinformation campaigns are often associated with foreign (commonly Russian) sources in the form of "information operations" (*Bennett & Livingston, 2018*), domestic actors can also be involved.
- What is the role of platforms and the algorithms they design to support the flow of information online? Social media and search engines play a crucial role in sifting through large amounts of information to surface what people will find useful. They make choices about what is prioritized on people's screens but we know little about how that works and how it is changing.
- How do niche/non-mainstream social media sites interact with more mainstream social media sites and online news publications? A lot of social media research is focused on Twitter and Facebook because data is more easily accessible from these sites. However, to understand the overall ecosystem of false information in Canada, research is needed on these other social media sites .
- How can news media and journalistic sources avoid amplifying false information? Guidelines and strategies are needed to help mainstream media determine when they should cover potential false stories.
- What issues are most important to Canadians? What issues incite the most emotional reactions among Canadians? There is a lack of study into the specific Canadian context.
# What you'll learn from the projects
False information can originate in non-mainstream internet spaces before inserting itself into the Canadian election discourse on social media and online news platforms. Alt-right media in Canada adapts their communications styles to mainstream and non-mainstream media sites to recruit members and strengthen their communities using different digital platforms (*Elmer, et al.*). **[PAGE 58](#page-57-0)**
Emotionally charged issues in Canada, such as national identity, racism, and free speech, are expressed on social media both directly and indirectly through "deep stories". National and historical narratives lay the groundwork for emotional flashpoints, which can contribute to political polarization on social media (*Boler*). **[PAGE 62](#page-61-0)**
Media manipulation campaigns employed polarizing issues, including immigration and racism, to provoke strong reactions. However, these campaigns were not amplified by the Canadian mass media system and thus failed to become viral (*Donovan, et al.*). **[PAGE 66](#page-65-0)**
# Fringe Politics: The Deep Web's Impact on the 2019 Canadian Election
#### **Greg Elmer,**
Ryerson University, School of Professional Communication, Professor gelmer@ryerson.ca
#### **Marc Tuters**,
University of Amsterdam, Department of Media Studies, Associate Professor
#### **Anthony Burton**,
York\Ryerson Universities, Joint Programme in Communication and Culture, M.A. Student,
#### **Melody Devries**,
York\Ryerson Universities, Joint Programme in Communication and Culture, Ph.D. Candidate
#### **Ganaele Langlois**,
Associate Professor, Department of Communication Studies, York University
#### **Stephen J. Neville**,
Ph.D. Student, Joint Programme in Communication and Culture, York\Ryerson Universities
#### **Sabrina Ward-Kimola**,
M.A. Student, Joint Programme in Communication and Culture, York\Ryerson Universities
### **What we wanted to know**
We investigated how political memes, language, and shared political objects (videos, photos, images, graphics, posts, etc.) from fringe websites became insinuated into mainstream political conversation on more established social media platforms and news properties in discussions of the 2019 Canadian federal election. In contrast to the popular theory of "fake news" as the product of foreign interference, our hypothesis was that much of the democratically disruptive content making its way to social media and news platforms originates on nonmainstream internet spaces such as 4chan/pol/ and Reddit. There is a distinct lack of critical scientific study in Canada about how extremist content makes its way on to mainstream platforms during election cycles; how this content is picked up by commentators on mainstream platforms; and the effect that this has on contemporary political debates and elections. This study provides insight on how marginal political actors' dark web content intervened, or was actively co-opted by other political interests and groups, to influence the fall 2019 election. The findings will contribute to ongoing discussions about the governance and regulation of elections, political parties and candidates in the context of online media properties and platforms.
#### **This overall study was further broken down into four distinct projects:**
The first project focused on how GoFundMe campaigns have emerged as sites of electoral debate and fundraising, particularly on the part of individual, nonpartisan individuals and some candidates.
The second project investigated how "adversarial citizenship" in Canada is being produced during the federal election through a set of online media practices.
The third project investigated Rebel Media's rapid ascension as both a major entity within the Canadian media sphere and the global "alt-right" media sphere. The project asked: is there an intersection between the affordances of new media platforms and the dissemination of "alt-right" ideology? What is the "style" of this new alt-right media sphere?
The fourth project developed distinct behaviour-based archetypes that describe and qualify the varying types of conservative or far-right users that engage with content on mainstream as well as fringe websites.
# **How we did it**
Software-assisted digital methods (*Rogers, 2015*): Using manual and automated software, we scraped data from internet platforms (such as GoFundMe and Reddit) and tracked the circulation of digital objects (e.g. memes, videos, posts, and links) from one platform to the next. This method is also referred to as 'cross platform studies' (*Elmer & Langlois 2013*).
Online discourse analysis: analyzed images and text to understand ideological and rhetorical strategies.
Qualitative ethnographic analysis of online users.
### **Key terms and what they mean**
#### Adversarial citizenship
A set of practices charged with feelings of anger, hostility, fear, and resentment towards an imaginary enemy, as reflected in online platforms by doxing, disinformation, and hate speech.
#### Archetype
A category of user-type that indicates similarity in terms of political beliefs, social features and positionality (age, gender, economic status, personality, technical literacy, etc.). In this project, archetypes are based on detailed qualitative observation that indicates patterns of behaviour and worldview.
#### Fringe financing
Within the context of electoral financing, pertains to the use of alternative modes of accruing donations for political and electoral goals, such as through GoFundMe.
#### Ideological recruitment
The processes by which online users are drawn to identify with the logics, narratives, and rhetoric of varying far-right belief systems.
#### Vox pop
A style of YouTube video wherein a host attends a public political event and attempts to interview attendees in the spur of the moment. This style is popular on alt-right commentary channels, especially the Rebel.
#### Reactionary media
Non-mainstream media entities whose primary product is political commentary reacting to contemporary cultural and political issues. The politics on display are often themselves reactionary.
#### Alt-right
Qualifies contemporary forms of political conservatism differentiated from "New Right" movements of past neoconservativisms. The contemporary "Alt-Right" is characterized by a rejection of traditional conservative and liberal politics, progressive knowledge, and governance structures. They hold values rooted within but not limited to white supremacy, denial of minority rights and equity initiatives, and anti-immigration sentiments.
# **What we found**
Our findings reveal that profound changes are under way in how different digital platforms — be they established media institutions (e.g. the CBC), social networking sites (Facebook, Reddit), crowdfunding platforms, or fringe, alt-right media — are mobilized by alt-right political actors in Canada. Altright movements are proficient at multiplying site-specific communicative strategies to both recruit new participants and strengthen their existing micro-communities.
Alt-right discourse online generally tends to present a profound sense of socio-economic betrayal from the state, which is further seen as symptomatic of a Canadian nation in existential crisis. Alt-right members on sites such as Facebook but also on marginal sites like 4chan and Reddit enclose themselves within a negative worldview mired by a pervasive sense of fighting multiple adversaries, from the Prime Minister to immigrants, Antifa, LGBTQ people, and climate change activists.
# **What we found**
This antagonism towards the state and mainstream liberal culture underlies the content and themes of Rebel Media, a YouTube channel and news website founded by Canadian lawyer Ezra Levant in 2015. By crafting "microcelebrities" through the narrative continuities and low overhead of YouTube as a media platform, the Rebel has quickly grown into a key media property of the global "new right new media" along with websites such as Breitbart and the Daily Caller.
Our digital archetype research used qualitative analysis to examine how users engage with right-wing to far-right content on social media sites such as Facebook, Reddit, and CBC article comment sections. This research showed that users often develop shared modes of political communication and habitual online practice which can further cement them within far-right worldviews. Moreover, this work showed how distinct
styles of political engagement corresponded to socio-cultural features, including socio-economic status, age, and geographic region (urban vs. rural). From this, we formed "character types" or archetypes that predicted reactions to contentious events, and more generally confirmed that politically charged online engagement shares deep relation with offline life, which has implications for how we address populations or assess the viability of policy changes and suggestions.
On the crowdfunding sub-project, our findings suggest the more successful campaigns are highly linked to other platforms, most notably YouTube. Further, evidence suggests that these highly linked campaigns are more likely to advance distrust of Justin Trudeau's persona, and contain anti-Islam sentiment or yellow vest rhetoric. (see **Figure 1** below).
Figure 1. Theme Distribution: Sucessful Campaigns
# What this means for the election
This study confirmed that politically charged online engagement shares deep relation with offline life, which has implications for how we address populations or assess the viability of policy changes and suggestions.
The 2019 election saw the full-fledged arrival of alt-right ideologies and communication tactics, both online and off, at the fringe of political discussions and through established party politics structures.
Political discussion online, especially around federal election time, plays a key role in defining the contemporary
state of Canada, in formulating (or attacking) shared values around what it means to be Canadian and in defining how to face complex global issues as a nation. The divisiveness of alt-right tactics and discourses negates the core of this process of formulating shared democratic worldviews.
As elections in other countries have shown, even if the 2019 election did not see a single alt-right candidate elected, we should nevertheless not delude ourselves that Canada, because of its multiculturalism and bilingualism, is immune from the rise of neo-fascist power.
### What's next
- Frameworks should be implemented by Elections Canada to better assess sources and amounts of party financing during election campaigns.
- Lawmakers should conduct broad consultation on the spectrum of online political violence which does not fall under the category of hate speech under current legislation (memes, political commentary and dark humour).
- Federal government and its agencies should regulate or redesign existing digital platforms to promote healthy civil discourses and protect vulnerable groups and individuals in Canada.
- Future research should analyze additional crowdfunding platforms such as Crowdpac, Patreon, and Kickstarter. Research should also consider the legal boundaries that alternative financing traverses when used for political ends.
# Affective Media, Social Movements, and Digital Dissent: Emotions and Democratic Participation in the 'Post-Truth' Era
#### **Megan Boler, PI**, 1
OISE/University of Toronto, Social Justice Education, Professor megan.boler@utoronto.ca
# **What we wanted to know**
The Cambridge Analytica/Facebook scandal brought to public attention how voters' emotions are manipulated by targeted digital propaganda. However, this is only one way that politicians, advertisers, and other actors target, incite, and exploit emotions within social media to influence public opinion and election outcomes. Such manipulation commonly seeks to inflame and polarize deep-seated emotions related to racial and national identities. While these 'post-truth' politics are increasingly well-documented in the U.S., we know less about the Canadian context and how digital propaganda threatens Canadian democracy. As well, scholars lag far behind corporate and private sectors in understanding emotional manipulation within social media.
#### **To better understand how digital propaganda targets emotions, we are exploring:**
Which emotional expressions and which deep stories are associated with racial and national identity and belonging, in the context of the 2019 Canadian federal election?
How do these emotional expressions and deep stories vary across the political spectrum?
What feeling rules and emotional expressions characterize the different political perspectives/ camps within Canada?
# **How we did it**
Qualitative and quantitative, exploratory, comparative mixed-methods research of emotional expressions within social media: across platforms of Twitter, Reddit, and YouTube. From each of the three platforms, our team manually surveyed approximately 1000 distinct tweets, comments, and posts from each platform. Our quantitative team analyzed 1.5 million tweets; as well as 412,752 comments from Far-Left reddit, and 335,304 comments from Far-Right reddit.
Discourse analysis of social media discussions related to the 2019 Canadian election2
Sentiment analysis and natural language processing using Vader, Python, and machine learning
Social network analysis
- Boler team members: Hoda Gharib, Department of Social Justice Education. Emma McCallum, MA student, Department of Social Justice Education. Michael Primrose, MA student, Department of Curriculum Teaching and Learning. Amanda Trigiani, Ph.D. student, Department of Social Justice Education. Kate Reyes, PhD student, Department of Curriculum Teaching and Learning. Míchílín Ní Threasaigh, MA student, Department of Social Justice Education. Arezou Soltani, Ph.D. student, Department of Social Justice Education. Sherry Yuan, undergraduate student, Computer Science. Peter Potaptchik, undergraduate student, Computer Science. Aakash Iyer, MA student, Mechanical and Industrial Engineering. Minjia Zhu, undergraduate, Industrial Engineering. Lionel Liang, MA student, Mechanical and Industrial Engineering. **1**
- The research phase of September 2019-March 2020 supported by Digital Ecosystem Research Challenge is but the initial phase of a three-year research project that will compare the Canadian 2019 election with the U.S. 2020 election. **2**
# **Key terms and what they mean**
#### Feeling rules
The rules regarding which particular feeling(s) may be appropriate within given social settings, which also demarcate the direction, duration, intensity, and/or objects of emotions considered appropriate in any situation (*Hochschild, 1983*).
#### Deep stories
The "feels-as-if accounts" that underlie all political beliefs and ideologies. A deep story describes pain, blame, and points to a rescue narrative; it provides an emotional accounting system, establishing an accounting of who deserves sympathy, distrust, shame. "This system becomes a foundation for feeling rules – which establish what we believe we 'should and shouldn't' feel – now a key target of heated political battle." (*Hochschild, 2016, Global Dialogue*).
#### Social justice warrior
A pejorative term for an individual who repeatedly and vehemently engages in arguments about social justice on the Internet.
#### Cancel culture
This primarily internet-based phenomenon is "an attempt to take away someone's power and influence when the person in question expresses an 'objectionable opinion' or has 'conducted themselves in a way that is unacceptable.'" *(Sung, 2019).*
# **What we found**2
Our qualitative findings include insights regarding two key areas of relevance to Canadian elections and politics:
- **(a) hypocrisy and perception of political candidates' integrity,**
- **(b) political and partisan polarization.**
Regarding (a): Anger and perceptions of hypocrisy in relation to revelation of Scheer's American citizenship were significantly shared across political spectrum regardless of party or partisan differences. By contrast, **Figure 1** (below) shows much greater variation in citizens' anger regarding Trudeau blackface incidents.
Figure 1. Anger linked to perceptions of Trudeau's hypocrisy with the blackface scandal:
| Far left | Center and
center left | Far right,
center right, center |
|---------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------------|
| Sustained anger
towards Trudeau
and supporters
given 'deep
seated racism'
still unaddressed
by Trudeau
and Canadian
government. | Little if any
anger; blackface
incidents
'insignificant'
given
perceptions
of Trudeau's
multicultural
commitments. | Pronounced
anger towards
Trudeau for
perceived
hypocrisy,
overlooked by
his supporters. |
NB: The findings reported are preliminary, as this 4 month analysis reflects the initial findings of a 3 year research project. **2**
# **What we found**
**Figures 2** and **3** below represent our findings regarding feeling rules and deep stories that fuel polarization within Canadian politics and 2019 election. The deep stories are rooted not only in current events, but in long-standing perceptions/version of Canadian history, which vary significantly by region (*Western, Eastern*) and provincial history. Especially notable, Canadian deep stories were less visible within election debates per se than they were within the context of debates regarding Wexit and the firing of Don Cherry, shortly following the 2019 election.
Figure 2. Deep stories animating political perspectives regarding racial and national belonging in each sector of Canadian political spectrum:
| Far left | Center | Far right |
|-----------------------------------------------------------------|-----------------------------------|-------------------------------------------|
| 'We are all settlers on stolen land; | 'Canada is a nation of | 'Canada was better off back when |
| the only people who have right to | immigrants and a global model | "we only let in Europeans and the |
| complain about immigrants are First | for multiculturalism because we | odd Asian". Current immigration |
| Nations, Inuit, and Métis peoples. | value our cultural mosaic. We are | should accept far fewer immigrants, |
| Borders are a tool of oppression because | a tolerant people as evidenced | and only through proper channels; |
| capital is able to move freely across them | by our history from the North | need to ensure those admitted are |
| while people are not. We have a moral | Star Underground Railroad to | from 'compatible cultures', already |
| obligation to resettle as many refugees | UN Peacekeeping. Sure, Canada's | educated and prepared to work (hard |
| as possible. Canada's migrant worker | not perfect, but we are way | and for low pay); immigrants must be |
| programme is exploitative if we need | better than most other countries | willing to assimilate fully into, display |
| to import workers, they should be paid | (especially the US and those | appreciation for, and not try to change, |
| living wages, offered citizenship, and a
social safety net." | places with dictators).' | the dominant culture.' |
Table 3. Feeling rules and predominant emotions related to narratives of national and racial belonging vary across the political spectrum. These include the following, across the Canadian political spectrum:
| Far left | Center left | Center | Center right | Far right |
|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| Anger that
Canadian
multiculturalism
overlooks
structural and
systemic racism.
Hopeless
regarding
genuine
conversation
about systemic
racism. | Shame
regarding racism
in Canada.
Outrage
about Don
Cherry's racist
comments. | Shame and surprise
regarding Trudeau's
blackface and comment
on Singh's turban;
perception that these
racist acts are 'exceptions'
rather than reflecting
systemic racism in
Canadian culture.
Anger regarding Trudeau's
and Scheer's hypocrisies.
Hopeful and empathetic
regarding these 'learning
moments' and possibility
for cultural change. | Anger about
Trudeau's
hypocrisies.
Outrage
about Don
Cherry being
fired. | Intense national
pride.
Disgust and
anger with 'cancel
culture' and
perceived silencing
of 'free speech'
by 'social justice
warriors'.
Dismissal/
annoyance
towards political
correctness and
climate change
hysteria. |
# What this means for the election
- As our findings regarding deep stories and feeling rules evidence, to understand how political communication mobilizes emotions over social media it is crucial to consider both the micro-level creation and recreation of narratives through posts and comments on social media, as well as the more abstract and macro-level (and historical) narratives which structure the Canadian political spectrum. This interaction between macro-and micro-levels means that highly-charged issues like national identity may not be expressed directly, nor expressed directly in relation to the election per se.
- In the Canadian context, highly-charged issues like national identity and polarized views regarding racism and "free speech" became more apparent in the discourse surrounding unexpected events. After the 2019 election, the debates regarding Wexit and the firing of a long-time sports commentator Don Cherry, served as flashpoints that provided greater clarity regarding the concerns that polarize Canadian voters.
- Canadian voters have high expectations that politicians behave with integrity, and are offended and angered
- when they perceive hypocrisy in politician's behavior, actions, and pasts.
- Our investigation of deep stories and feeling rules reveals the extreme complexity of factors that contribute to political polarization, and the challenge of understanding such polarization. Political polarization is not only caused by an inability to speak across political divides, but by an inability to feel across these divides. This means that those researching elections must pay attention not only to the emotional flashpoints, but also to the manifold historical, cultural, and political factors which lay the groundwork for these flashpoint to occur.
#### Limitations related to our project:
At present, there are very few studies that use this "mixed-methods" approach; as well, there are very few studies that focus specifically on Canadian social media. Finally, we have discovered that the 'deep stories' in Canada are much less obvious than the deep stories in U.S. political social media. All of these have factors have posed limitations and challenges to our research.
# What's next
#### Policies and publics need to consider:
Developing media education and media literacy to include focus on understanding digital propaganda and specifically, the manipulation of emotions.
#### Future research needs to consider:
Developing nuanced, mixed-method models for sentiment analysis. Quantitative tools of sentiment analysis are extraordinarily rudimentary, and have extraordinarily low accuracy rates—often only 60% at best. Claims made on the basis of quantitative methods of sentiment analysis/NLP alone should be viewed skeptically, and should be validated through manual and qualitative coding.
Supporting investigative journalism into disinformation that includes research into how political and corporate advertising aims to influence voters' emotions within social media.
Investigating differences in the feelings rules and emotions of voters, particularly in terms of how these vary based on political perspective. To say 'voters are angry about x' misses critical differences in how the political function of an emotion may vary depending on political party, for example. There is no singular meaning of 'anger' in a political context. What voters are angry about and how that anger shapes voters' attitudes varies greatly, depending on political stance and identity.
# Media Manipulation, Memetic Campaigns, and the 2019 Canadian Elections
#### **Joan Donovan**,
Director and Lead Researcher of the Technology and Social Change Research Project at the Shorenstein Center for Media, Politics and Public Policy, Joan\_Donovan@hks.harvard.edu
**Brian Friedberg**,
**Nicole Leaver**,
**Jon Penney**,
Senior Researcher
Research Assistant
Senior Research Fellow
# **What we wanted to know**
The Technology and Social Change (TaSC) research project at the Shorenstein Center investigates media manipulation campaigns, trends, and techniques that seek to influence politics and culture. Throughout the Canadian electoral cycle, researchers at TaSC tracked the life cycles of media manipulation campaigns. These campaigns were propagated by a mixed group of actors, including political operatives, social movements, junk news agencies, and unaffiliated trolls. Given that disinformation campaigns peak during elections, our analysis culminated in three particular case studies that responded to **the following questions:**
Was there a discernible pattern to media manipulation efforts? How did disinformation spread and adapt throughout the 2019 election cycle?
What manipulation trends and techniques were employed?
What kinds of media manipulation campaigns were attempted? What were the methods of response? Have responses, if any, been successful?
Were there any evident or traceable effects of the media manipulation campaigns?
# **How we did it**
**Digital Ethnography and Open Source Intelligence** (OSI) (*Colquhoun, 2016*): We identified and documented misleading or false stories and memes with high engagement by analyzing communities on Reddit, Facebook, Twitter, YouTube, and anonymous message boards. We mapped the cross-platform spread of these materials by manually reviewing the content of these websites and platforms, and used OSI to establish the provenance of materials.
**Social Media Monitoring and Media Content Analysis**: Using MediaCloud, Crowdtangle and Netlytic, we gauged the scale and spread of particular stories, memes or slogans on Twitter and Facebook, and minor platforms like Gab and 4chan.
**Collaboration with Journalists**: Hosted an election-day "situation room" with reporters from Buzzfeed Canada, where we vetted potential disinformation in real time to support journalists reporting on elections.
# **Key terms and what they mean**
#### Disinformation
False or misleading information spread for the purpose of deception, including inaccurate statements about public figures, falsified news stories, or viral hoaxes.
#### Memetic Campaign
Memes are an integral part of contemporary visual communication and can be found on any platform that allows the sharing of text, images, or video. Meme campaigns are the organized sharing of sticky slogans and images to influence public opinion. These campaigns are often authorless and decentralized, thus making them difficult to attribute to a single source.
# **What we found**
We examined three cases of media manipulation and meme campaigns during the Canadian election, each of which provoked responses by the mainstream press. These cases followed discernible patterns, depending on the themes, scale, and scope of the campaign. Our life cycle model illustrates how campaigns are planned, spread, elicit reactions, and adapt to changing conditions. These three campaigns failed to provoke mass attention and, thus, evaded response from platform companies while amplifying disinformation and extremist content. Importantly, studying failed manipulation attempts can shed light on new counteractions for disarming the weaponization of social media.
Using digital ethnography and OSI techniques, our research explores the production and distribution of disinformation qualitatively. By following manipulation and memetic campaigns across platforms directly through real-time tracking, we provide insight into the tactics employed by a rising tide of bad actors and extremist movements online. By cataloguing their campaign cycles, our research seeks to name patterns of abusive behavior such as networked harassment, algorithmic manipulation, and hoaxing—that may be better regulated by platforms and legal bodies.
To analyze a case study, we map the life cycle of a media manipulation campaign through five phases: (**1**) **initial phase of ideation and discussion**, (**2**) **seeding the campaign across social web and platforms**, (**3**) **challenges by journalists, platform companies, activists and regulators**, (**4**) **observable changes in the media ecosystem**, and finally (**5**) **adaptations by media manipulators to the new environment**. Many media manipulation campaigns fail to garner significant responses from more powerful actors, so in our case studies we did not find evidence of responses or changes to the media ecosystem.
#### Media Manipulation
A loosely affiliated and/or coordinated effort to create misleading media coverage and falsely influence public perception using networked communication technologies.
#### Brigading
A manipulation technique used to shape public perception by positively or negatively affecting engagement metrics on partisan content by overwhelming comment sections on news articles, tweets, and posts. Brigading can occur organically when someone shares an unpopular opinion that is widely reacted against by individual users, or it can be orchestrated by using fake accounts or botnets at scale.
# The Buffalo Chronicle
*The Buffalo Chronicle* published four anti-Trudeau articles between October 7 and 17 based on false rumours and conspiracies circulating on social media, including an alleged scandal stemming from Trudeau's time teaching at West Point Grey Academy; allegations that Trudeau faked security threats during the campaign; and a rumour that the RCMP would be charging Trudeau in connection with in the SNC Lavalin scandal ("RCMP plans to charge Trudeau with obstruction in SNC Lavalin affair, following federal elections," 2019; "RCMP source says 'security risk' against Trudeau was contrived by PMO staffers," 2019; "Trudeau accusor lands a seven-figure NDA to keep quiet about West Grey departure," 2019; "Trudeau is rumored to be in talks with an accusor to suppress an explosive sex scandal," 2019). The site paid to promote these articles on Facebook to drive engagement, and the content was also shared on social media in conservative, conspiratorial, and far-right communities (*Silverman et al., 2019*). Each was quickly debunked in the mainstream press, after which a new article with a different piece of misinformation was published and promoted (*Lytvynenko et al., 2019*). *The Buffalo Chronicle* Facebook page and website were not subject to any significant response and the articles remain online.
We approached the *Buffalo Chronicle* as a single case study overall, but also pursued the three specific false rumours/conspiracies it promoted as embedded cases. Each provided insights on how disinformation develops and spreads—following the discernible "life cycle" pattern noted above—but also as to tactics and techniques used by various actors to do so.
# **What we found**
This can be seen, for example, with the false West Point Grey Academy scandal story:
1 **Seeding/Ideation:** This false scandal was first "seeded" online via posts and content by various actors on different social media platforms and political websites, including Twitter, Youtube, and Reddit, as early as 2018. Additional
*The Buffalo Chronicle*
related rumours about Trudeau's time at West Point Grey Academy were also posted on websites as early as 2015.
2 **Seeding Across Web/ Platforms:** Varying posts/tweets repeating similar "rumours" on these platforms continued through 2018 and 2019, but had limited reach until the election period, when more influential sites
and social media actors, particularly right-wing media sites and accounts on Twitter, Facebook, and YouTube, amplified it after the Trudeau blackface/brownface scandal came to light on September 18.
- 3 **Challenges/Responses by Journalists / Politicians:** actors associated with mainstream media outlets, including a journalist with *The Globe and Mail* asking Trudeau directly about the scandal in a press conference, amplified the rumoured scandal in early October. Mainstream political parties did so as well, as the Conservative Party of Canada issued a press release on October 7. *The Buffalo Chronicle*, a site that would later receive widespread media coverage for publishing false rumours and stories throughout the election, also amplified the story via a published story on October 7.
- 4 **Changes to Information Ecosystem:** On October 4, the headmaster of West Point Grey Academy during Trudeau's time there issued a statement denying the rumoured scandal and clarifying the circumstances surrounding Trudeau's departure. In the days following, multiple mainstream media outlets debunked the rumoured scandal.
- 5 **Adjustment by Manipulators:** various influential actors and sites on various platforms, particularly Twitter, adjusted by
amplifying new (not yet debunked) rumours about Trudeau, or (as in the case of *The Buffalo Chronicle*) published "new" false rumours about the scandal despite the denials/ debunking, or moved on to different false rumours and conspiracies about Trudeau.
*The Buffalo Chronicle*
Among the key findings here were recurring tactics and techniques in these cases. Among the more effective were:
- 1 **"trading up the chain"**, (*Krafft & Donovan, 2020; Marwick & Lewis, 2017*) which involves various actors formulating themes, narratives, rumours, and/or ideas in one context, such as Twitter or a Reddit thread—usually with smaller audiences or followings— and then shifting these narratives to other platforms or contexts in order to gain interest or amplification from more influential actors, like news outlets or right-wing social media figures, for greater impact;
- 2 **amplification**, where platforms, news outlets, or other actors with larger audiences/following report on, cover, or repeat disinformation, increasing its audience and impact (*Donovan & Boyd, 2019*); and
- 3 **manipulating frames**, (*Marwick & Lewis, 2017*), wherein actors reframe media narratives, news, and/or information to continue propagation, amplification, or "trading up the chain".
We also found that, despite the intensive media coverage the Buffalo Chronicle received for creating and spreading disinformation during the election, it was not, in fact, primarily an originator of disinformation but an amplifier of previously seeded, framed, and amplified false rumours and conspiracies found on social media. If anything, the *Buffalo Chronicle* was effective at identifying such false stories, slightly manipulating the framing of the story or scandal, and amplifying them through established channels.
# **What we found**
### Racist Trudeau Memetic Campaigns
- 1 We documented several memetic campaigns to create and distribute crudely doctored photos of Trudeau in black- or brownface and different racist consumes (*MacPherson, 2019*).
- 2 Many of these memes originated on Reddit and 4chan and were later shared more widely on Facebook, Twitter and Instagram. By tracking communities that produce political memes during election cycles in real time, we observed how memetic campaigns adapt to breaking news cycles, while evading content moderation (*Green, 2019*).
- 3 Though largely ignored by politicians, mainstream press provided some more critical commentary on these memetic campaigns, with The Star Vancouver declaring Trudeau had "lost the meme war" (*Peng, 2019*).
Again, as with the *Buffalo Chronicle* case study, we treated this as a single overall case study, but then also pursued individual instances of faked or racist Trudeau photos memes as embedded case studies, including "Trudeau as Hitler" memes or cheap fake "new" Trudeau blackface photos. To varying degrees, we found these instances followed some stages of the "life cycle" pattern above, but, as noted, were often ignored by media and politicians, possibly because there was little coordination among actors involved. Images were seeded and shared by a smaller
# People's Party of Canada (PPC) YouTube Brigading
We monitored YouTube closely, and selected videos discussing Trudeau or Bernier from five mainstream Canadian news YouTube channels—*The Globe and Mail*, the *Toronto Star*, the *National Post*, CBC News and CTV. Videos featuring Bernier, or discussing Bernier and/or the PPC, were positively brigaded, where top-rated comments were supportive of the candidate. Videos featuring Trudeau, or discussing Trudeau and/or the Liberal Party, were negatively brigaded, where top-rated comments were critical of the candidate and party. Our findings suggest that positive online engagement did not directly correlate to voting outcomes, as Bernier and his party lost all elections (*Ling, 2019*). However, positive social media engagement compelled mainstream journalists to cover the PPC, and continued to attract more journalists to wedge issues central to this party.
*Just For Trudeau / Twitter Account*
number of actors—often from far-right communities on Reddit. It may also be the case that media outlets and larger social media accounts were engaging in "strategic amplification" (*Donovan & boyd, 2019*), or strategic silence, in ignoring these memes as their content was often blatantly racist or easily debunked.
We found key tactics and techniques here to include "trading up the chain," whereby memes were created in one context, like Reddit and then moved to others. Use of "cheap fakes," that is, editing photos in simple, cheap, and unsophisticated ways, that convey false information or messages. Here, actors cheaply edited photos of Trudeau to create fake images of Trudeau in "Nazi uniform" or fake black- and brownface photos.
The Globe and Mail. URL : https://www.youtube.com/user/TheGlobeandMail; Toronto Star. URL :https://www.youtube.com/TorontoStar; National Post. URL : https://www.youtube.com/user/nationalpost; CBC News. URL :https://www.youtube.com/user/cbcnews; CTV News. URL : https://www.youtube.com/ctvnews." **1**
### What this means for the election
Particular polarizing issues, including immigration and racism, provoked strong reactions in this election cycle. Part of the challenge was Trudeau's widely publicized history of black- and brownface incidents, which created a media ecosystem ripe for exploitation by bad actors seeking to spread false rumours, confusion, and even hate. Moreover, our study of YouTube brigading in favor of the People's Party of Canada illustrated how a small number of influential online actors, often with ties to the far right, were able to amplify, drive, and set mainstream media agendas on specific wedge issues, all hallmarks of disinformation campaigns. Nevertheless, our study found no evidence to suggest that any manipulation campaign impacted the election's integrity or produced unexpected electoral results. In fact, Canadian reporters seemed to be attuned to the
threat of amplifying falsehoods and were strategically silent on some potentially viral stories (*Wardle, 2019*).
For ethnographic researchers, there remain important limitations when studying the intent and impact of disinformation and memetic campaigns. First, without access to subjects, we cannot always ascertain motivation for spreading disinformation or racist material, particularly among pseudonymous communities. Second, measuring the distribution of disinformation or memes is limited by the available social media analytical tools, which rely on access to major platforms. Finally, we cannot definitively state that we identified and catalogued every instance of disinformation relating to the Canadian elections, just those with visible effects on social media and mainstream press.
# What's next
# Further research on far-right extremism in Canada
The resurgence of far-right mobilization requires a committed federal strategy for countering extremism and radicalization online. To help formulate this strategy, we recommend more research into how far-right extremist movements amplify and spread disinformation to reach their political goals. This should include crosssector collaboration, and continued accountability to marginalized communities that are the targets of disinformation.
# Better training for journalists covering far-right movements and disinformation
Journalism is caught in a paradox with online disinformation. Journalists are both the first line of defense against disinformation and also a major target for malicious actors who want to spread manipulation campaigns. Attempts to manipulate racial discourse are particularly difficult to navigate for journalists. We recommend that journalists covering far-right movements take deliberate and strategic steps to avoid spreading disinformation in their coverage of hate movements.
# Improved access to platforms' data for journalists and researchers studying disinformation and media manipulation
While social media analysis tools, like Crowdtangle and Netlytics, offer some insight into how disinformation spreads online, data showing the full distribution on social media platforms is not available for outside auditing. Capturing the true spread of disinformation and hate speech online for academic research requires greater transparency from platform companies so that researchers and civil society groups can audit the impact of unchecked misinformation.
# Agenda Setting and Issue Framing
# What you need to know
The battle to set the topic and tone of public discourse during an election is paramount for all campaigns. Agenda-setting and issue framing are two key strategies for impacting what topics are on the minds of potential voters, largely via news media and social media dissemination and discussions.
Each political actor in a campaign has their own agenda of issues and topics they care about and that they want the public to be aware of. Agenda-setting is the process by which certain issues become more salient in public discourse. The idea here is that news media, for example, can tell the public what to think about, but not what to think (*McCombs & Shaw, 1972*). While they are not the only agenda-setters, what the news media covers during election campaigns impacts the salience of attitudes towards specific political issues (*Entman, 1993*). Issue framing describes the practice of the media placing emphasis on certain aspects or angles of an issue to promote a particular interpretation (*Scheufele & Tewksbury, 2007*). Issue framing has been studied in the context of both mainstream news media and social media (*Wasike, 2017*). Political actors use issue framing as a rhetorical strategy to emphasize specific aspects of an issue to support their own position (*Riker et al., 1996*).
In the past, broadcasters played the central role in agenda setting and issue framing. However, social media is more dynamic and interactive. For example, politicians who post on social media in reaction to news media and broadcasters who then cover these posts exist in a cyclical or symbiotic relationship where both are agenda setters and issue framers (*Conway et al., 2015*).
# What we still don't know
How do agenda-setting and issue framing work across different online platforms? Much work compares Twitter to online news articles (*Egbunike & Olorunnisola, 2015*) but less has been done comparing the effects across different social media platforms.
Do agenda-setting and issue framing techniques and impacts vary by issue? Agenda setting and issue framing practices vary widely on different online channels (social media platforms and online news publications), when published by different actors (politicians, activists, interest groups) and when covering different issues. Since these practices cannot easily be generalized, more research is needed on an issue specific basis.
# What you'll learn from the projects
Public opinion suggests climate change and environmental issues were important issues to Canadians during the 2019 federal election. Social media posts about climate change and environmental issues by party leaders and supporters differed by political party (*Boulianne, et al.*). **[PAGE 74](#page-73-0)**
As compared to newspaper headlines, Twitter framed the issue of migration in Canada in predominantly negative terms, specifically migration was framed as threatening, fearsome and hostile. These findings demonstrate how social media environments can shape the public discussion of socio-political issue (*Walsh, et al.*). **[PAGE 78](#page-77-0)**
Political party leaders use moral rhetoric to frame key election issues according to their party positions. While appeals to moral rhetoric are used by all party leaders, its application differs for each party across key issues including for example the Trudeau Blackface scandal and the environment (*Lalancette and Gosselin*). **[PAGE 82](#page-81-0)**
Discussion about Indigenous issues by political candidates during the 2019 federal election differed according to political party. Overall, Indigenous issues did not stand out as a key election issue on Twitter, suggesting this was a missed opportunity to engage in a conversation on Indigenous issues (*Savard, et al.*). **[PAGE 86](#page-85-0)**
Indigenous candidates frequently discussed environmental issues and Indigenous specific issue topics including missing and murdered Indigenous women, UNDRIP and reconciliation. A major issue theme identified in Twitter conversations about Indigenous candidates was around fraudulent claims of Indigneous identity which fueled harmful discourses (*Leonard, et al.*). **[PAGE 90](#page-89-0)**
# Climate Change, Social Media, and Political Parties
#### **Shelley Boulianne**,
MacEwan University, Associate Professor, sjboulianne@gmail.com
#### **Stephanie Belland,**
MacEwan University, Research Assistant
#### **Anders Olof Larsson**,
Kristiania University College, Professor
# **What we wanted to know**
In 2018-2019, world leaders wavered on their commitments to international climate change agreements, while millions of people marched in the streets to encourage action on climate change. These international events, as well as domestic debates about pipelines and the economy, contributed to climate change being a key issue in the 2019 Canadian federal election. Existing research demonstrates that climate change views are highly correlated with political affiliation (Boulianne & Belland, 2019; Hornsey et al., 2016). We used survey data to examine how environmental concerns and climate change opinions factored into people's voting preferences leading up to the election. Then, we examined political leaders' posts on Facebook, Twitter, and Instagram to determine the frequency with which they posted about climate change. **Our research question:**
Are there party differences in attention to and concern about climate change?
# **How we did it**
Survey data collected from an online panel of 1,500 Canadians, September 2019.
Social media posts by five political leaders, collected using CrowdTangle and Rtweet, July 22 to October 22, 2019.
# **Key terms and what they mean**
#### Party support
Survey respondents were asked a series of questions about which party they were thinking of voting for in the upcoming election. People who said they were undecided were further prompted with a question about whether they were leaning toward a party and if so, which one. The question was asked differently in Quebec, as the Bloc Québécois was included in the response options. People's Party of Canada was not offered as an option and made occasional appearances in the "other party" category.
#### Importance of climate change
Survey respondents were asked, "How important are candidates' positions on environmental issues when making your choice about which party to vote for?"
#### Concern about climate change
Respondents were also asked about their level of concern for environmental issues generally and about climate change specifically. The responses were identical. For brevity, we present the responses to the question specifically about climate change.
#### Climate change related social media posts
We searched social media posts for references to "climat" which covered references to climate change in both English and French. Two human coders reviewed each reference to confirm that the reference related to climate change.
### **What we found**
Environmental issues were important for Canadians in this election. **Figure 1** demonstrates that very few Canadians (1 below; lightest colour in graph) reported that candidates' environmental positions were "not at all important" when making their voting decisions. The average score for importance, on a scale ranging from 1, "not at all important," to 7, "very important," was 5.
**Figure 2** outlines how party supporters differed in their views about the importance of candidates' positions on environmental issues. Conservative Party supporters considered the environment to be less important, on average, than supporters of other parties. Undecided voters were also distinctive in the low degree of importance they attached to the environment.
**Figure 2** also reports average levels of concern about climate change. Conservative Party supporters were distinctive in their low levels of concern, on average, about climate change, relative to other parties' supporters. The average among undecided voters was similar to that of Conservative Party supporters.
Figure 1. Importance of Candidates' Environmental Positions
Figure 2. Importance of Environmental Positions
### **What we found**
**Figure 3** outlines the number of posts that each party leader made to their social media profiles and the percentage of those posts that related to climate change. The study counted English and French versions of the same posts as multiple posts. The study excluded retweets and tweets consisting of quotes from other sources with no additional content.
The study found that the Conservative Party leader posted the least about climate change compared to other party leaders.
In sum, Canadians viewed the environment as an important factor in their voting decisions. Conservative Party supporters indicated the environment was of relatively little importance to them when choosing which party to vote for. These supporters were also not as concerned about climate change as other parties' supporters. Finally, the Conservative Party leader did not post much about climate change on social media, making only a handful of references during the campaign.
Figure 3. Social Media Posts by Political Leaders
| Figure 3. Social Media Posts by Political Leaders | | | | | |
|-----------------------------------------------------|-----------------------------------------------------|------------------------------------------------------|--------------------------------------------------------|-----------------------------------------------------|--|
| 2.6%
19 / 727
mentions on all
social media | 0.2%
5 / 2096
mentions on all
social media | 28.5%
69 / 242
mentions on all
social media | 11.3%
211 / 1872
mentions on all
social media | 9.2%
90 / 979
mentions on all
social media | |
| 0%
0
mentions on
Instagram | 0%
0 / 191
mentions on
Instagram | n/a
0
mentions on
Instagram | 13.5%
12 / 89
mentions on
Instagram | 6.1%
7 / 115
mentions on
Instagram | |
| 1.9%
11 / 569
mentions on
Twitter | 0.3%
3 / 1131
mentions on
Twitter | 26.7%
40 / 150
mentions on
Twitter | 10.7%
131 / 1224
mentions on
Twitter | 9.2%
51 / 556
mentions on
Twitter | |
| 5.1%
8 / 158
mentions on
Facebook | 0.3%
2 / 774
mentions on
Facebook | 31.5%
29 / 92
mentions on
Facebook | 12.2%
68 / 559
mentions on
Facebook | 10.4%
32 / 308
mentions on
Facebook | |
| Yves-François
Blanchet,
Bloc Québécois | Andrew Scheer,
Conservatives | Elizabeth May,
Green | Justin Trudeau,
Liberals | Jagmeet Singh,
NDP | |
# What this means for the election
- The study suggested that climate change and environmental issues were important considerations for voters in the 2019 Canadian federal election. Leaders and supporters of different parties showed varying levels of concern and attention to climate change. The Conservative Party leader and supporters were distinctive in their low levels of attention to, and concern for, climate change. While the leader and supporters aligned on this lower level of importance, a large majority of Canadians were concerned about climate change and environmental issues. As for climate change as a policy issue, supporters and the leader of the party that won the most seats (Liberal Party) had high levels of attention to, and concern about, climate change.
- The study's data were limited by the timeline of survey data collection. We asked about the importance of climate change to voters' decisions a few weeks before the election. We do not know if voters' views changed as the election progressed. Also, the survey did not ask about support for the People's Party of Canada, which is important considering the policy area under discussion. The leader of the People's Party of Canada was the only leader to deny the existence of climate change. The study did not examine whether supporters of this party also denied the existence of climate change.
### What's next
- Public opinion surveys support making climate change a priority, and suggest that the new government has a mandate to act on the issue.
- Future research: examine which social media posts about climate change received more interactions, such as likes, retweets, comments, and shares (Larsson, 2019). This additional analysis will help us understand which initiatives might have greater public support.
- Future research: examine all of the leaders' social media posts to understand which posts receive the most interaction. This additional analysis will help us understand the relative importance of climate change, compared to other policy areas, such as the economy, immigration, and healthcare.
# Digital Media, Migration, and the Canadian Federal Election
#### **James Walsh**,
University of Ontario Institute of Technology, Assistant Professor of Criminology, james.walsh@uoit.ca
#### **Aziz Douai**,
University of Ontario Institute of Technology, Associate Professor of Communications and Digital Media Studies
#### **Amin Ibrahim**,
University of Ontario Institute of Technology, Associate Teaching Professor of Business and Information Technology
### **What we wanted to know**
Unlike the U.S. and Europe, since the 1990s, Canada has avoided a nativist backlash, with support for diverse, largescale migration found across the political spectrum. However, recent developments, whether the rise of the populist People's Party or growing polarization within the electorate (*EKOS Politics 2019*), suggest the issue increasingly represents a political flashpoint and source of, often acrimonious, conflict and debate, outcomes that are potentially reflected in and amplified by social media communications (*Flores-Yeffal et al 2019*). In assessing the extent of such developments, this study systematically analyzes the participants, themes, and engagement patterns that define online discussions about the contentious and multi-faceted issue of migration during the 2019 Canadian Federal election.
#### Ultimately, **this research investigates the following specific questions:**
How are social media used to discuss migration and refugee issues?
Which claimants and themes predominate?
What are digital communications' implications for the nature and tenor of political discourse and debate?
# **How we did it**
Analyzed a representative sample of 2273 tweets containing election-related hashtags (#cdnpoli, #elxn43 etc.) and migration-related keywords (immigration, refugee, migrants, etc.) posted during the 2019 election (September 11–October 21, 2019).
Compared the sentiments and themes that characterize social media communications with newspaper coverage. This included all 215 stories published during the election with headlines containing migration-related keywords. Leading national (*Globe and Mail, National Post*) and regional (*Vancouver Sun, Calgary Herald*, etc.) newspapers featured in the database Canadian Newsstream were targeted.
Evaluated engagement patterns (likes, retweets, etc.) to gauge message influence and impact.
# **Key terms and what they mean**
#### Sentiment
Tweets and headlines were coded to reflect their predominant emotional valence. Statements expressing sympathy and/or support for migrants were coded 'positive' and assigned sentiment scores of +1. Statements depicting migration as a threat to, among other things, national security, sovereignty, and identity, were coded 'negative' and assigned scores of -1. Factual, neutral, and ambiguous statements were coded 'other' and assigned scores of 0. To capture differences in potential visibility, scores were weighted by newspapers' daily readership and the number of followers per Twitter account.
#### Theme
All negative statements were re-coded 'risk' while positive messages were coded either 'victim' or 'contributor' with the former highlighting migrants' marginalization and the latter accentuating their societal contributions.
#### Engagements
The number of engagements ('likes' and 'retweets') per tweet was recorded and used to capture message impact and reception.
# **What we found**
During the election, 63.4% (n=1141) of migration-related tweets contained user-generated content and ordinary citizens posted 78.1% (n=1775) of all messages (**Figure 2**). Further, 42% (n=973) of tweets were coded 'negative' versus 40% (n=904) and 18% (n=396) 'positive' and 'other'. 'Risk' was the leading theme in general and, excluding two weeks, throughout the election while positive messages were considerably more likely to emphasize migrants' vulnerability rather than their contributions (**Figure 1**).
Figure 1. Weekly Proportion of Tweets Expressing Positive or Negative Sentiments
Note: 42.8% of all tweets were coded Risk versus 31.9% for Victim and 7.9% and 17.4% for Contributor and Other (not featured), respectively.
Figure 2. Profile of Twitter Users
Original content is defined by the percentage of messages that were not shared or 'retweeted' from other accounts
# **What we found**
To contextualize these findings, the average sentiment of tweets was compared with the average sentiment of newspaper headlines. Twitter communications were found to be negative both in absolute and relative terms (**Figure 3**). To account for differences in message visibility and the number of individuals exposed to specific communications, sentiment scores were adjusted to capture variations in the number of followers per Twitter account. While weighting tweets by audience size increased the mean sentiment, suggesting positive messages
were disproportionately posted by more popular accounts and thereby exposed to a greater number of users, tweets remained negative compared to newspaper headlines (**Figures 5** and **6**). Finally, statistical analysis was conducted to assess whether message sentiment meaningfully influenced engagement patterns. As conveyed in **Figure 4**, while sentiment had no statistically significant effect on the number of likes, messages coded negative were more likely to be retweeted.
Figure 3. Mean Sentiment Expressed in Tweets, Headlines, and Public Opinion
| Source | Mean Sentiment | Weighted Mean
Sentimenta | N |
|-----------|----------------|-----------------------------|------|
| Tweets | -0.03 | 0.23 | 2273 |
| Headlines | 0.29 | 0.37 | 215 |
Sentiment scores were weighted by multiplying each assigned score by the corresponding number of followers per Twitter account and daily readers per newspaper. The mean was calculated by dividing the sum of weighted sentiment scores by the total number of followers and readers contained in each sample. **a**
A Chi-Square test of independence revealed significant differences in sentiment scores (321.18, p <.001)
Figure 4. Mean Number of Engagements by Message Sentiment
| Sentiment | Retweets | Likes | Total Engagements |
|-----------|-------------|-------------|-------------------|
| Positive | 29.6 (80.0) | 13.5 (89.5) | 43.2 (150.7) |
| Negative | 45.6 (78.6) | 9.2 (73.5) | 54.7 (129.4) |
| Other | 11.2 (42.2) | 11.2 (68.4) | 22.3 (108.3) |
A one-way ANOVA test revealed statistically significant differences in the mean number of retweets (F=32.0, p<.001) and total engagements (F=8.1, p<.001) across groups. A post-hoc Tukey HSD test revealed that, for retweets, all groups differed significantly at p<.001. For total engagements, post-hoc tests revealed statistically significant differences of p<.05 between the positive and other groups, as well as, differences of p<.01 between the negative and other groups. No statistically significant differences in the mean number of likes across groups was found (F=.70, p=.497).
Figure 5. Weekly Mean Sentiment Expressed in Tweets and Headlines 0.8 0.6 0.4 0.2 0.0
# What this means for the election
Our research findings offer, at best, partial and qualified support for claims of social media's democratic potential. While Twitter was overwhelmingly used by citizens to directly shape political discourse, tweets disproportionately portrayed migration in threatening, fearsome, and hostile terms. Moreover, although positive messages were potentially exposed to more individuals through follower networks, tweets containing negative sentiments were disproportionately retweeted. Thus, despite broadening opportunities for information acquisition and robust political expression and participation, social media also contain illiberal tendencies, with negative, exclusionary discourse displaying outsized visibility and influence.
Additionally, our results suggest that claims of widespread support for diverse, large-scale migration amongst
Canadians (*Bloemraad 2012*, *Reitz 2014*) are less applicable to digital platforms, with Twitter communications evincing considerable rancor and division. By permitting ordinary people to communicate alongside powerful actors, Twitter offers a barometer of popular sentiment, allowing complex political processes to be documented and reconstructed in entirely novel ways.
In its current form, our research displays the following limitations. First, focusing on a single country and electoral cycle makes it difficult to discern our findings' uniqueness. Further, without data on users' political orientations, it is not possible to document ideological cleavages within the electorate. Finally, in the absence of a more detailed coding scheme, knowledge of migration's thematic framing remains limited.
### What's next
- To better capture migration's representation across the political spectrum, future research will consider how specific candidates, political parties, and their supporters approached the issue. Doing so will deepen understanding of migration's construction as a public issue and identify the extent to which conflict and polarization are shaped by partisan differences.
- The project's coding scheme will be elaborated to comprehensively document migration's construction as a public issue. Beyond identifying the emotional valence and general theme of tweets, future research will also identify how migration in its various forms (permanent/ temporary, voluntary/forced, regular/irregular, etc.) is associated with specific topics and frames (economic growth, unemployment, national security, social and cultural identity, environmental issues etc.). Such refinements will assist in developing a wider appreciation of the meanings attached to migration and refugee issues.
- To document Twitter's use in distorting communications and circulating misinformation and incendiary claims, future research will employ specialized software to establish the likelihood users are 'bots'—fake accounts deployed to engineer and manipulate online conversations. When combined with data concerning account type, message theme, and engagement patterns, such information will help establish the extent, profile, and impact of bot activity.
- Together, the results of this project are relevant to several stakeholders. In addition to helping policymakers monitor and identify shifts in public opinion, its findings can assist civil society groups in developing more efficacious strategies of messaging and persuasion. Finally, by documenting patterns of media manipulation, the project may be of interest to social media platforms themselves.
# Moral Rhetoric and the Digital Environment: The 2019 Federal Election Campaign
#### **Mireille Lalancette**,
Université du Québec à Trois-Rivières, Ph.D., Full Professor in Political Communication, mireille.lalancette@uqtr.ca
#### **Tania Gosselin**,
Université du Québec à Montréal, Ph.D., Associate Professor in Political Science, gosselin.tania@uqam.ca
### **What we wanted to know**
This project explores the moral rhetoric in social media during the 2019 federal election campaign through a qualitative and quantitative content analysis.
#### **Three questions guided our research:**
What moral values do party leaders use on Twitter?
Does moral rhetoric depend on the ideology of a political party?
What moral considerations do citizens use to justify their political preferences?
Studies on moral language in politics deal almost exclusively with the American context, showing that moral rhetoric is linked to its bipartisan dynamics. This research explored whether the same moral foundations shape political discourse in the Canadian multiparty system. A portrait of the moral rhetoric used in social media allows us to better assess the importance of language and the moralization of issues within discourses, as well as their role in partisan competition dynamics.
# **How we did it**
First, we carried out a qualitative analysis of all the campaign tweets from the six main Canadian party leaders (2,630 tweets) posted between September 11th and October 21st inclusively) and then proceeded with a quantitative analysis of the citizens' answers to survey questions on the reasons why they would vote (or not) for a specific leader or political party.
# **Key terms and variables**
In order to explore the moral rhetoric, we relied on the moral foundations theory (MFT) (*Graham, Haidt & Nosek, 2009*), which offers six dimensions related to the moral universe:
This theory was used to study the moral rhetoric in social media (*Sterling & Jost, 2016*), blogs (*Sagi & Dehghani, 2014*) and tweets (*Kaur & Sasahara, 2016; Grover et al., 2019*). Other academics have reviewed the impact of moral framings in partisan discourses on the support for public policies (*Clifford et al., 2015; Clifford & Jerit, 2013; Severson & Coleman, 2015*). These moral dimensions are tied to partisan identity, preferences, and political behaviour. Liberals more often rely on the dimensions of care and fairness, while conservatives tend to highlight authority and purity. The values put forward can also depend on the issue: for example, the sacred dimension of nature can be raised by liberals regarding environmental policy (*Frimer et al., 2016*).
### **What we found**
The results show a strong presence of moral rhetoric, particularly in Andrew Scheer and Justin Trudeau's tweets. While the MFT's moral foundations are not as clearly associated with the left-right political spectrum in Canada as they are in the United States, their use by Canadian party leaders is nonetheless consistent with ideology, as well as with issues more specifically supported by regional or left-wing parties. Among right-wing parties, the use of moral values aligns with a general ideological perspective, without necessarily being anchored to a specific issue.
Present in about 80% of the Liberal and Conservative leaders' tweets, the value of care/harm is the most commonly used. Loyalty and authority are those most often mobilized by Andrew Scheer and Maxime Bernier. Yves-François Blanchet also refers to authority: his focus on respecting laws is linked to the status
of Quebec. Loyalty is also a dimension addressed by Jagmeet Singh and Elizabeth May in reference to specific communities, including the riding. The negative counterpart of loyalty is voiced through the denunciation of deceit: Justin Trudeau's blackface is sometimes presented in those terms, as well as the "betrayal" of the ruling parties regarding the fight against climate change. Also, results suggest that environmental issues, frequently addressed in Elizabeth May and Justin Trudeau's discourse, are often depicted in terms of purity and degradation. The liberty/oppression dimension is barely mentioned by the leaders, except for Maxime Bernier.
Citizens use fewer words associated with moral values than politicians do, and without a clear connection to ideology. Opinions about leaders seem fairly similar in terms of moral rhetoric.
Figure 1. Moral values in the federal party leaders' tweets
### What this means for the election
Analyses show that leaders heavily resort to moral rhetoric on social media beyond issues that are typically considered to be about morality such as abortion and medical assistance in dying. The results echo the words of Kreitzer et al. (2019), for whom all policies and issues are subject to a "moralization" that varies across time and space.
Looking at the moral dimension of discourse within the socio-digital environment is relevant to the study of elections because using an emotional and moral language can increase the circulation of messages on social media (*Valenzuela et al.*, 2017); Canadians also rely more and more on online content to stay informed (*Newman*, 2019; *Lalancette*, 2018).
This project shines a light on a still uncharted aspect in Canada: the use of moral language in the discourse tied to politics. Consequently, the analyses conducted as part of this project represent a starting point. However, our analyses show that if MFT's moral foundations prove to be relevant to describe the 2019 federal campaign, the study of moral rhetoric requires combining multiple automated and qualitative content analysis methods. This multimethod approach should be applied to the available data and those that will be added following the Digital Democracy Project.
### What's next
Moral rhetoric often elicits strong emotional reactions (*Lipsitz*, 2018). The emphasis on moral values in elite discourse on salient issues can contribute to the polarization of citizens' political attitudes (*Clifford et al.*, 2015), namely by limiting the willingness to compromise and to weigh the consequences of public policies (*Ryan*, 2017, 2019). However, moral and emotional arguments can also convince and rally beyond political lines (*Feinberg & Willer*; 2013, 2015). The strong presence of moral language witnessed during the 2019 federal election campaign thus warrants consideration. Combined with the rise of partisan differentiation, especially on the right of the Canadian political spectrum (*Kevins & Soroka*, 2018), it could lead to growing polarization.
- Political players should consider the potential consequences of moral rhetoric before resorting to it in a systematic and strategic fashion. While those consequences can be positive, they can also contribute to the polarization of partisan positions, as well as to the demobilization of some citizens in the face of a political agenda perceived to be "too conflicting."
- The strong presence of moral rhetoric in political discourse on social media suggests that the issue should be included in media literacy courses. It would enable citizens to evaluate critically not only the reliability of political information, but also the way it is communicated.
# Federal Election Candidates' Interest in Indigenous Issues
#### **Jean-François Savard,**
National School of Public Administration (ENAP), Associate Professor, Jean-Francois.Savard@enap.ca
#### **Mathieu Landriault**,
Observatoire de la Politique et de la Sécurité de l'Arctique, Director
#### **Isabelle Caron**,
Dalhousie University, Associate Professor
#### **Christian Rock**,
Member of the Innu First Nation
### **What we wanted to know**
We wanted to know how the candidates from the five political parties with representation in the House of Commons at the time of the dissolution of the 41st Parliament used their Twitter accounts to address Indigenous issues during the federal election. The goal was to determine who was posting messages on Indigenous issues and which topics were addressed in these communications.
A second goal aimed at highlighting the characteristics associated with each candidate tackling Indigenous issues to a greater or lesser extent. The candidates' gender and party affiliation, the nature of their riding, the communications strategies of Indigenous groups, and the influence of the party leaders' national campaigns were all variables taken into account to explain the variation in the number and type of messages.
# **How we did it**
We established a list of the candidates' Twitter accounts.
By using "R" and the rtweet package, we retrieved, every day from September 11th to October 21st, the tweets published directly from Twitter's API platform (total of 99,000 tweets).
We automatically analyzed these tweets using a platform that we created.
# **Key terms and what they mean**
#### Political parties selected
Parties with representation in the House of Commons at the time of its dissolution (New Democratic Party, Conservative Party of Canada, Liberal Party of Canada, Parti Québécois, Green Party).
#### Candidates
Individuals seeking election in one of the five selected parties and having a Twitter account at the time the election was called.
#### Frequency of tweets
Total number of tweets published and related to Indigenous issues.
#### Frequency of words
Number of times a specific word is repeated in the tweets selected for analysis.
#### Keywords
Words associated with a theme or central idea that characterizes a tweet. Those words are preceded by the pound sign (#) in the text of those tweets.
#### Frequency of keywords
Number of times a keyword is repeated in the tweets selected for analysis.
#### Theme analysis
Hierarchical cluster analysis of words found in the tweets selected that helps identify groups of related words, thus revealing the topics addressed through the set of tweets.
#### Popularity analysis
Analysis that takes into account the number of tweets republished and liked by other users.
# **What we found**
The most popular words were "crise" (crisis) in French and "reconciliation" in English. The NDP candidates published the largest number of tweets, followed by the candidates from the Green Party and Liberal Party. The CPC and the Bloc Québécois showed little interest in Indigenous issues. Also, the number of tweets on Indigenous issues were the highest in urban ridings.
Figure 1. Number of tweets per political party
Considering our first research question, reconciliation stood out as a common topic on Twitter. Reconciliation was then divided into several sub-issues such as childcare, employment and residential schools. Access to drinking water was also a prominent topic. Finally the impact of climate change on Indigenous communities was one of the main issues of this campaign.
Figure 2. Number of tweets per riding
Considering our second research question, political affiliation represented the clearest explanatory variable. Candidates from the NDP and Green Party talked the most about Indigenous issues on Twitter. It also appears that the candidates from ridings located in or near urban centres with large Indigenous populations were more sensitive to those issues. **Figure 2** shows the four ridings where the number of tweets addressing Indigenous issues were the highest as compared to the national average. Those ridings were all located in or near an urban centre.
# **What we found**
Our research has shown that the interest given to Indigenous issues on Twitter was closely related to specific events that received coverage from mainstream media. For example, spikes on October 5th, 7th and 8th were respectively due to Jagmeet Singh's visit in the community of Grassy Narrows, as well as the leaders' debate in English and then in French. We also show women were more interested in Indigenous issues than men, but the topics discussed were generally the same. Finally, no candidate tackled the issue of missing and murdered Indigenous women and girls, even though the commissioners from the National Inquiry tabled their report just a few weeks before the election was called.
# What this means for the election
The candidates' partisan affiliation was the best predictor of whether a candidate was going to address Indigenous issues and of the topics they were going to deal with. The local candidates' communications strategies closely followed by those of the political party leaders' national campaign.
Although Indigenous issues were at the heart of the political communication of the outgoing government, it was difficult for those topics to stand out during the 2019 election campaign. The candidates showed little interest in addressing those issues or in making them national issues of their campaign.
The lack of interest has potentially limited the amount of reflection and questioning about the future direction for the recognition of Indigenous rights and reconciliation with Indigenous peoples. In that respect, this is a missed opportunity by the candidates, and parties, to contribute to this discussion.
This project had certain methodological limitations: the candidates' communications on Twitter only represent one type of political communication; The local campaigns are far more complex and involve in-person debates, door-to-door canvassing, and campaigns in mainstream media. This research project did not have the necessary resources to take into account all types of communications.
### What's next
- It would be interesting to analyze how the candidates' political communications on social media platforms like Twitter resemble or differ from other types of communications from those same candidates. Such a study would help highlight the functions and roles played by these media in all of the candidates' communications.
- The dominance of the national campaign is significant and had an impact on the local candidates' independence. Political parties must offer greater independence to their candidates so they might discuss issues specific to each riding. This could improve the quality of the public.
- Election campaigns are a great opportunity to discuss and debate core issues in Canadian society. In the years preceding this election, interest in issues facing Indigenous populations was paramount, however, these same issues were given little attention during this election. The lack of attention dedicated to these issues from the Bloc Québécois and CPC candidates is particularly disturbing. This silence should incite a reflection on our society and bring the parties, especially the Bloc Québécois and CPC, to question themselves.
# A Study of Indigenous Voices Online in the 2019 Canadian Federal Election
#### **Kelsey Leonard**,
(Shinnecock Nation), McMaster University, Department of Health, Aging & Society, Post Doctoral Fellow leonardk@mcmaster.ca
#### **Chelsea Gabel**,
(Red River Métis from Rivers, Manitoba), McMaster University, Department of Health, Aging & Society, Associate Professor
#### **Marrissa Mathews**,
(Omushkegowuk Cree, Weenusk First Nation), McMaster University, Department of Political Science, PhD Student
# **What we wanted to know**
We examined online activity centered on Indigenous issues during the 2019 Canadian Federal Election. We aimed to understand the experiences of Indigenous voices online during the election and how digital democracy threats to Indigenous Data Sovereignty attempted to diminish Indigenous rights and self-determination. We also sought to identify the key issue trends of Indigenous online users in digital spaces. The study analyzed digital content posted to Twitter and identified topics of salience for Indigenous Twitter users and Indigenous candidates during the election. **Our study addresses the following research questions:**
What were the most frequently used Indigenous hashtags among the general population of Twitter users during the 2019 Federal Election campaign season (August 1 to December 1, 2019)?
What political issues were most frequently discussed by Indigenous candidates active on Twitter in the 2019 federal election?
How were Indigenous candidates talked about on Twitter?
# **How we did it**
Indigenous Digital Research Methodology: drawing from principles of **Indigenous Data Sovereignty** (*Kukutai and Taylor 2016*), **Digital Bundles** (*Wemigwans 2018*), **Decolonizing Methodologies** (*Smith 1999*), and **Storywork** (*Archibald 2008*).
Tweet collection: Collected **Twitter data** on two sets of accounts, Indigenous candidates and the general population, from August 1 to December 1, 2019. We identified **62 Indigenous candidates** in the election, but **only 45 had Twitter accounts**.
Tweet Selection: We recorded a tweet if it contained the following hashtags: #Attawapiskat #FirstWater #WaterCrisis #GreenNewDeal #StopTMX #reconciliation #MuskratFalls #indigenous #MMIW #MMIWG #UNDRIP #TMX #Transmountain #GrassyNarrows #SNCLavalin #FlintNorth #waterislife #fnpoli #indigenousvotes #IndigVote #NativeVote #indigpoli #Unistoten #tinyhousewarriors #notransmountain #nomancamps #MétisNation #FirstNations #Metis #Inuit #IdleNoMore #FNMI #abpoli.
We also, in a separate collection, identified Twitter posts (August–December 2019) from Indigenous candidates. For candidates, we included all original tweets, as well as retweets and replies that were made by these accounts: Total of **8961 tweets**.
Content Analysis: Tweets were coded using **thematic analysis for salient electoral issues** identified by Indigenous candidates as determined by frequency of political issues in post collection.
# Key terms and what they mean
#### Indigenous Data Sovereignty
"The right of an Indigenous nation to govern the collection, ownership, and application of its own data. From this perspective, Indigenous governance of data assets is about stewardship for both present and future generations" (*Kukutai & Taylor 2016, pp.132*).
#### (i) First Nations Water Crisis
In Canada, many First Nations communities across the country do not have access to safe drinking water and live under boil-water advisories.
#### **Indigenous Electoral Politics**
Indigenous electoral politics in the Canadian context considers disparities in voter turnout and Indigenous Peoples' decision-making processes for advancing Indigenous policy agenda items often further complicated by limited electoral infrastructure on reserve, party Indigenous platforms, and colonial power dynamics.
#### Indigenous Peoples
Indigenous communities, peoples and nations are those which, having a historical continuity with pre-invasion and pre-colonial societies that developed on their territories, consider themselves distinct from other sectors of the societies now occupying those territories or parts of them. "Indigenous Peoples" as used in this project includes Aboriginal, Inuit, and Métis communities throughout what is currently known as Canada; individual members of First Nations; and status and non-status Indians, including those living on or off reserve.
#### **W** UNDRIP
The United Nations Declaration on the Rights of Indigenous Peoples (UNDRIP) is an international instrument adopted by the United Nations on September 13, 2007. Canada has yet to fully implement UNDRIP nationally.
#### What we found
**Figure 1.** Examples of Frequency of Hashtag Use Among Twitter Users During 2019 Canadian Federal Election From August to December 2019
The study found that the general Twitter population used the following hashtags most frequently: #mmiwg, #mmiw, #fnmi, #fnpoli, #tinyhousewarriors, #notransmountain, and #grassynarrows. We hypothesized that key political discourse points from the 2015 election and campaign promises of the incumbent administration, such as eliminating drinking water advisories in Indigenous communities and implementing the United Nations Declaration on the Rights of Indigenous Peoples, would have a high posting frequency in the social media activity of online users and Indigenous candidates. As Figure 1 shows with the examples of water and reconciliation, hashtags related to these issues (such as #mmiwg, #reconciliation, #waterislife, and #indigpoli) did spike in October and leading up to the election. Other highly used hashtags were unexpected including #abpoli and #greennewdeal. We did find that some hashtags were used for two different purposes. For example, #abpoli was used
to refer to Aboriginal politics as well as Alberta politics, and #greennewdeal was frequently used in discussions relating to American elections, as well as by Indigenous candidates for environmental issue advocacy.
Indigenous electoral politics discourse in the 2015 federal election was dominated by questions of Indigenous Peoples' voter turnout and participation (*Gabel et al. 2016*). Hashtag campaigns such as #nativevote, #indigvote, and #indigenousvotes were created to raise awareness and increase Indigenous voter participation. The use of these hashtags continued in the 2019 election, peaking before election day for each hashtag. **Figure 1** provides the example of #indigvote hashtag. But we found the 2019 election saw more diversity of Indigenous issue framing, indicating the discourse of Indigenous electoral politics is evolving beyond simply participating in Canadian electoral politics and positions of "to vote or not to vote."
### **What we found**
Next we investigated the political issues most frequently discussed by Indigenous candidates. Of 62 Indigenous candidates in this election, 18 did not have Twitter accounts. None of the Indigenous candidates without Twitter accounts were elected.
Of the hashtags we sampled, the top five (5) most frequently used by Indigenous candidates are (See **Figure 2**): #reconciliation, #indigenous, #firstnations, #undrip, and #greennewdeal. Use of hashtags evolved over time, for example, Indigenous candidates' use of #greennewdeal, #reconciliation, and #undrip peaked near election week (See **Figure 3**).
Figure 2. Frequency of Identified Hashtag Use Among Indigenous Candidates During 2019 Canadian Federal Election
Figure 3. Frequency of Hashtag Use Among Indigenous Candidates During 2019 Canadian Federal Election from August to December 2019
# **What we found**
It is also helpful to look beyond simple hashtag counts. Although water crisis hashtags were not frequently used by Indigenous candidates, content analysis found that Indigenous candidates referenced Indigenous water issues in 62 posts on Twitter from August 1 to December 1, 2019. The NDP used the First Nations water crisis as a political action point during its campaign, which was frequently referenced by Twitter users with hashtags such as #GrassyNarrows and #watercrisis (*Canadian Press, 2019*). This is particularly relevant given that an APTN and Environics poll during the election found that drinking water and climate change were ranked by Indigenous Peoples as their top issues, which may have linkages to the 44% of Indigenous candidates who tweeted or retweeted content on protecting the environment. This was also the first ever federal election opinion poll of Indigenous Peoples by an Indigenous news organization in Canada (*Environics Research, 2019; Martens, 2019*).
We also considered how Indigenous candidates were talked about by others on Twitter. Our search query for #Indigenous #elxn43 and #Metis found a newly trending Indigenous issue topic in the 2019 Federal Election. This emerging discourse included accusations of ethnic fraud perpetrated by certain candidates, who claimed Indigenous identity without any evidence to support their claims (*Bascaramurty, 2019; Barrera and Deer, 2019*). The tweets shown here illustrate two examples. Candidates who had their Indigenous identity claims challenged in news media and on Twitter included Celine Laquerre, Marc Serré, George Canyon, Amanda Kistindey, Jocelyn Rioux, and Kathy Laframboise.
Lastly, an important issue raised by Indigenous voices active on Twitter during the election using the #elxn43 hashtag and referencing Indigenous Peoples pertained to the common reference by non-Indigenous candidates of ownership of Indigenous Peoples. As seen in the tweet by Sarah Harney, many Indigenous People find it extremely offensive when candidates or elected officials say "Our Indigenous People" or "Canada's Indigenous People" implying that Indigenous Nations are not sovereign but rather under the colonial patronage of the Canadian state. The saliency of this tweet for Indigenous voices online is representative given it was one of the most retweeted and liked Indigenous Twitter posts during the election.
# What this means for the election
During this election 29% of Indigenous candidates did not use Twitter to engage in electoral politics. Indigenous candidates had a high frequency of posts regarding the environment (although not always water specifically) and the UNDRIP compared to other Indigenous issue topics. Notably, among the hashtags we identified, #greennewdeal was one of those used most frequently by Indigenous candidates. This speaks to the importance of environmental issues for Indigenous Peoples and Indigenous candidates but also the way in which Indigenous candidates and their parties are influenced by American environmental politics.
Additional content shared by Indigenous candidates on Twitter during the 2019 Canadian Federal Election included Missing and Murdered Indigenous Women and Reconciliation using affiliated hashtags.
In the next phase of this research, we are conducting
interviews to determine the level of incivility and online abuse Indigenous Twitter users may have experienced during the election, as well as identifying particular actors.
An important concern raised by the study is whether ethnic fraud by political candidates jeopardizes elections' integrity. Some Indigenous People already perceive elections' integrity to be compromised. This is shown by the significant debate over whether or not Indigenous People should even vote at all, a debate which can be exacerbated by allegations of fraudulent claims of Indigenous identity. High-level party remedies for these concerns may provide meaningful acts of reconciliation and help increase Indigenous voter participation.
The largest limitation to this study is that some data collection and analysis is still ongoing and so new findings will likely emerge to add more context and support to these initial findings.
# **What's next**
#### Policy Recommendations:
- Indigenous candidates may benefit from enhanced digital activity including updated social media accounts throughout elections. 1
- All candidates and elected officials can improve Indigenous public relations by stopping all reference to Canadian ownership/dominion over Indigenous Peoples. 2
- All parties should consider implementing policies to prevent ethnic fraud and false claims of Indigenous identity by their candidates. 3
#### Future research:
- Indigenous Online Visions of the Future (what do Indigenous People see their use of the internet during elections leading towards?) 4
- Assessment of digital media support and resources available to Indigenous candidates to build their digital personas during elections. Also examining Indigenous candidate digital literacy and use of digital tools for effective online engagement and outreach (e.g. hashtags, ads, etc.) 5
**Allcott, H., & Gentzkow, M.** [\(2017\). Social media and fake news in the 2016 election. Journal of Economic Perspectives, 31\(2\),](https://www.aeaweb.org/articles?id=10.1257/jep.31.2.211) 211–236. https://doi.org/10.1257/jep.31.2.211
**Andrey, S., & Bardeesy, K.** (2019, October 17). Liberals outspend all other campaigns combined on local Facebook advertising. [https://www.ryersonleadlab.com/liberals-outspend-all-other-campaigns-combined-on-local-facebook-advertising.](https://www.ryersonleadlab.com/liberals-outspend-all-other-campaigns-combined-on-local-facebook-advertising)
**Awan, Imran.** [\(2016\). Islamophobia on social media: A qualitative analysis of the Facebook's walls of hate. International Journal of](https://zenodo.org/record/58517#.Xk3ZLUNOnjA) Cyber Criminology, 10(1), 1–20
**Bachner, J., & Hill, K. W.** [\(2014\). Advances in public opinion and policy attitudes research. Policy Studies Journal, 42, S51–S70.](https://onlinelibrary.wiley.com/doi/abs/10.1111/psj.12052)
**Barrera, J., & Deer, J.** [\(2019, October 10\). 4 federal candidates accused of Indigenous identity appropriation by Halifax academic.](https://www.cbc.ca/news/indigenous/federal-candidates-claims-indigenous-identity-1.5314614) CBC News.
**Bascaramurty, D.** [\(2019, October 20\). Candidates find themselves at centre of debate on Métis identity. The Globe and Mail. https://](https://www.theglobeandmail.com/canada/article-candidates-find-themselves-at-centre-of-debate-on-metis-identity/) [www.theglobeandmail.com/canada/article-candidates-find-themselves-at-centre-of-debate-on-metis-identity/](https://www.theglobeandmail.com/canada/article-candidates-find-themselves-at-centre-of-debate-on-metis-identity/)
**Bennett, C. J., & Lyon, D.** [\(2019\). Data-driven elections: Implications and challenges for democratic societies. Internet Policy Review,](https://policyreview.info/data-driven-elections) 8.
**Bennett, W. L., & Livingston, S.** (2018). The disinformation order: Disruptive communication and the decline of democratic [institutions. European Journal of Communication, 33\(2\), 122–139.](https://journals.sagepub.com/doi/10.1177/0267323118760317)
**Bloemraad, I.** (2012) Understanding 'Canadian exceptionalism' in immigration and pluralism policy. Migration Policy Institute.
**Boulianne, S., & Belland, S.** (2019). Who matters in climate change discourse in Alberta. In P. Prado, J. Pinto, & R. Gutsche (Eds.), Climate Change, Media & Culture: Critical Issues in Environmental Communication (75-94). Emerald Publishing.
**Boynton, S.** [\(2019, October 12\). Scheer defends conservative Chinese Facebook ads saying Trudeau will legalize hard drugs.](https://globalnews.ca/news/6025644/conservative-facebook-ads-drugs-liberals/) Global News.
**Brady, W. J., Wills, J. A., Jost, J. T., Tucker, J. A., & Van Bavel, J. J.** (2017). Emotion shapes the diffusion of moralized content in [social networks. Proceedings of the National Academy of Sciences, 114\(28\), 7313–7318.](https://www.pnas.org/content/114/28/7313)
**Brants, K.** [\(2005\). Guest editor's introduction: The Internet and the public sphere. Political Communication, 22\(2\), 143–146.](https://www.tandfonline.com/doi/abs/10.1080/10584600590933133) [https://doi.](https://doi.org/10.1080/10584600590933133) [org/10.1080/10584600590933133](https://doi.org/10.1080/10584600590933133)
**Brin, C.** [\(2019\). Digital News Report: Canada. Online: http://www.digitalnewsreport.org/survey/2019/canada-2019/](http://www.digitalnewsreport.org/survey/2019/canada-2019/)
**Brown, D.** (2016, November 29). We investigated those damning rumours about Fidel Castro being Justin Trudeau's real dad. Vice. [https://www.vice.com/en\\_ca/article/jmk95b/we-investigated-those-damning-rumours-about-fidel-castro-being-justin-trudeaus-real-dad](https://www.vice.com/en_ca/article/jmk95b/we-investigated-those-damning-rumours-about-fidel-castro-being-justin-trudeaus-real-dad)
**Canada Elections Act.** [\(2019\). Online: https://laws-lois.justice.gc.ca/eng/acts/E-2.01/](https://laws-lois.justice.gc.ca/eng/acts/E-2.01/)
**Canadian Press.** (2019, October 5). Jagmeet Singh commits to funding clean drinking water in Indigenous communities. The Globe [and Mail. https://www.theglobeandmail.com/politics/article-jagmeet-singh-commits-to-funding-clean-drinking-water-in-indigenous/](https://www.theglobeandmail.com/politics/article-jagmeet-singh-commits-to-funding-clean-drinking-water-in-indigenous/)
**Clifford, S.** [\(2017\). Individual differences in group loyalty predict partisan strength.](https://link.springer.com/article/10.1007%2Fs11109-016-9367-3) Political Behavior, 39(3), 531–552. [https://doi.](https://doi.org/10.1007/s11109-016-9367-3) [org/10.1007/s11109-016-9367-3](https://doi.org/10.1007/s11109-016-9367-3)
**Clifford, S., & Jerit, J.** [\(2013\). How words do the work of politics: moral foundations theory and the debate over stem](https://doi.10.1017/s0022381613000492) cell research. Journal of Politics, 75(3), 659–671. https://doi.10.1017/s0022381613000492
**Colquhoun, C.** [\(2016, July 14\) A Brief History of Open Source Intelligence. Bellingcat. https://www.bellingcat.com/resources/](https://www.bellingcat.com/resources/articles/2016/07/14/a-brief-history-of-open-source-intelligence/) articles/2016/07/14/a-brief-history-of-open-source-intelligence/
**Communication Security Establishment.** [\(2018\). Cyber Threats to Canada's Democratic Process. https://cyber.gc.ca/sites/default/](https://cyber.gc.ca/sites/default/files/publications/cse-cyber-threat-assessment-e.pdf) files/publications/cse-cyber-threat-assessment-e.pdf
**Community Standards.** [\(n.d.\). Facebook. https://www.facebook.com/communitystandards/inauthentic\\_behavior](https://www.facebook.com/communitystandards/inauthentic_behavior)
**Conway, B. A., Kenski, K., & Wang, D.** (2015). The rise of Twitter in the political campaign: Searching for intermedia agenda-setting [effects in the presidential primary. Journal of Computer-Mediated Communication, 20\(4\), 363–380. https://doi.org/10.1111/jcc4.12124](https://academic.oup.com/jcmc/article/20/4/363/4067549)
**Costello, M., Hawdon, J., & Ratliff, T. N.** (2017). Confronting online extremism: The effect of self-help, collective efficacy, and guardianship [on being a target for hate speech. Social Science Computer Review, 35\(5\), 587–605. https://doi.org/10.1177/0894439316666272](https://journals.sagepub.com/doi/10.1177/0894439316666272)
**Dahlgren, P.** (2004). Foreword. In W. van de Donk, B. D. Loader, P. G. Nixon, & D. Rucht (Eds.), Cyberprotest: New media, citizens and social movement (pp. ix–xiii). Routledge.
**Delacourt, S.** (2019, November 23). How the Liberals won — an inside look at the targeting and tactics that got Trudeau re-elected. Toronto Star.
**Donovan, J., & boyd, D.** [\(2019\). Stop the Presses? Moving From Strategic Silence to Strategic Amplification in a Networked Media](https://journals.sagepub.com/doi/10.1177/0002764219878229) Ecosystem. American Behavioral Scientist. https://doi.org/10.1177/0002764219878229
**Dubois, E., & Blank, G.** (2018). The echo chamber is overstated: The moderating effect of political interest and diverse media. [Information, Communication & Society, 21\(5\), 729–745. https://doi.org/10.1080/1369118X.2018.1428656](https://www.tandfonline.com/doi/full/10.1080/1369118X.2018.1428656)
**Dubois, E., Gruzd, A., & Jacobson, J.** (2018, September 18). When journalists report social media as public opinion. Policy Options.
**Duncombe, C. (2019).** [Digital diplomacy: Emotion and identity in the public realm. The Hague Journal of Diplomacy, 14\(1–2\), 102–116.](https://brill.com/view/journals/hjd/14/1-2/article-p102_8.xml) https://doi.org/10.1163/1871191X-14101016
**Egbunike, N., & Olorunnisola, A.** [\(2015\). Social media and the #Occupy Nigeria Protests: Igniting or damping a Harmattan storm?](https://www.ingentaconnect.com/content/intellect/jams/2015/00000007/00000002/art00004%3bjsessionid=2pe3gv7qgtehb.x-ic-live-03) Journal of African Media Studies, 7(2), 141–164. https://doi.org/10.1386/jams.7.2.141\_1
**EKOS Politics** (2019) Increased polarization on attitudes to immigration reshaping the political landscape in Canada. EKOS Research.
**Elections Canada** [\(2011/2015\). Registered party returns in respect of general election expenses. https://www.elections.ca/wpapps/](https://www.elections.ca/wpapps/WPF/EN/Home/Index?act=&returnStatus=&selectedReportType=&reportOption=) WPF/EN/Home/.
**Elmer, G., & Langlois, G.** [\(2013\). Networked campaigns: traffic tags and cross-platform analysis on the web. Information Polity 18,](https://content.iospress.com/articles/information-polity/ip000244) 43–56. https://doi.org/10.3233/IP-2011-0244
**Entman, R. M.** [\(1993\). Framing: Toward clarification of a fractured paradigm. Journal of Communication, 43\(4\), 51–58. https://doi.](https://academic.oup.com/joc/article-abstract/43/4/51/4160153?redirectedFrom=fulltext) org/10.1111/j.1460-2466.1993.tb01304.x
**Environics Research.** (2019). New survey highlights issues of importance to Indigenous Peoples leading up to the federal election. Environics [Research. https://environicsresearch.com/insights/new-survey-highlights-issues-importance-indigenous-peoples-leading-federal-election](https://environicsresearch.com/insights/new-survey-highlights-issues-importance-indigenous-peoples-leading-federal-election/)
**European Commission.** [\(2015\). Public Opinion in the European Union. https://ec.europa.eu/commfrontoffice/publicopinion/archives/](https://ec.europa.eu/commfrontoffice/publicopinion/archives/eb/eb83/eb83_first_en.pdf) eb/eb83/eb83\_first\_en.pdf
**Faris, R., Ashar, A., Gasser, U., & Joo, D.** [\(2016\). Understanding harmful speech online. SSRN Electronic Journal. https://doi.](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2882824) org/10.2139/ssrn.2882824
**Feinberg, M., & Willer, R.** [\(2015\). From gulf to bridge: When do moral arguments facilitate political influence?](https://journals.sagepub.com/doi/10.1177/0146167215607842) Personality and Social Psychology Bulletin, 41(12), 1665–1681. https://doi.org/10.1177/0146167215607842
**Fletcher, R., Cornia, A., & Nielsen, R. K.** (2019). How polarized are online and offline news audiences? A comparative analysis of [twelve countries. The International Journal of Press/Politics, 1–27 https://doi.org/10.1177/1940161219892768](https://journals.sagepub.com/doi/10.1177/1940161219892768)
**Florres-Yeffal N., Vidales, G., & Martinez G.** (2019) #WakeUpAmerica, #IllegalsAreCriminals: The role of the cyber public sphere in [the perpetuation of the Latino cyber-moral panic in the US. Information, Communication & Society, 22\(3\), 402–419. https://doi.org/10.1](https://www.tandfonline.com/doi/full/10.1080/1369118X.2017.1388428) 080/1369118X.2017.1388428
**Frimer, J.A., Tell, C.E., & Motyl, M.** (2017). Sacralizing liberals and fair‐minded conservatives: Ideological symmetry in the moral [motives in the culture war. Analyses of Social Issues and Public Policy, 17\(1\), 33–59. https://doi.org/10.1111/asap.12127](https://spssi.onlinelibrary.wiley.com/doi/abs/10.1111/asap.12127)
**Frimer, J.A., Biesanz, J.C., Walker, L.C., & MacKinlay, C.W.** (2013). Liberals and conservatives rely on common moral foundations [when making moral judgments about influential people. Journal of Personality and Social Psychology, 104\(6\), 1040–1059. https://doi.](https://psycnet.apa.org/doiLanding?doi=10.1037%2Fa0032277) org/10.1037/a0032277
**Gabel, C., Bird, K., Goodman, N. J., & Budd, B.** (2016). The impact of digital technology on First Nations participation and governance. The Canadian Journal of Native Studies, 36(2), 107–127.
**Goodman, N., Bastedo, H., LeDuc, L., & Pammett, J. H.** (2011). Young Canadians in the 2008 federal election campaign: Using [Facebook to probe perceptions of citizenship and participation. Canadian Journal of Political Science, 44\(4\), 859–881. https://doi.](https://www.cambridge.org/core/journals/canadian-journal-of-political-science-revue-canadienne-de-science-politique/article/young-canadians-in-the-2008-federal-election-campaign-using-facebook-to-probe-perceptions-of-citizenship-and-participation/C7C3F11B624EF23E4C30669FC3ADD91C) org/10.1017/S0008423911000783
**Gorrell, G., Bakir, M. E., Greenwood, M. A., Roberts, I., & Bontcheva, K.** (2019). Race and religion in online abuse towards UK [politicians: working paper. ArXiv:1910.00920 \[Cs\]. http://arxiv.org/abs/1910.00920](https://arxiv.org/abs/1910.00920)
**Government of Canada.** (2019). The Government of Canada's plan to safeguard Canada's 2019 election. https://www.canada.ca/en/ [democratic-institutions/news/2019/03/speech-thegovernment-of-canadas-plan-to-safeguard-canadas-2019-election.html](https://www.canada.ca/en/democratic-institutions/news/2019/03/speech-thegovernment-of-canadas-plan-to-safeguard-canadas-2019-election.html)
**Graham, J., Haidt, J., & Nosek, B.A.** (2009). Liberals and conservatives rely on different sets of moral foundations. Journal of [Personality and Social Psychology, 96\(5\), 1029–1046. https://doi.org/10.1037/a0015141](https://psycnet.apa.org/doiLanding?doi=10.1037%2Fa0015141)
**Green, M.** (2020, January 25) Why political memes—Which are virtually unregulated—Matter to this federal election. The Star. https:// www.thestar.com/vancouver/2019/10/05/why-political-memes-which-are-virtually-unregulated-matter-to-this-federal-election.html [Greenwood, M. A., Bakir, M. E., Gorrell, G., Song, X., Roberts, I., & Bontcheva, K. \(2019\). Online abuse of UK MPs from 2015 to 2019](https://arxiv.org/abs/1904.11230) [Working Paper]. ArXiv:1904.11230 [Cs]. http://arxiv.org/abs/1904.11230
**Grover, T., Bayraktaroglu, E., Mark, G., & Ha Rim Rho, E.** [\(2019\). Moral and affective differences in U.S. immigration policy debate](https://link.springer.com/article/10.1007%2Fs10606-019-09357-w) on Twitter. Computer Supported Coop Work, 28(2-3), 317–355. https://doi.org/10.1007/s10606-019-09357-w
**Gruzd, A., Mai, P., Recuero, R., & Soares, F.** [\(2020\). Manual coding of toxic and insulting tweets during the 2019 federal election in](https://dataverse.scholarsportal.info/dataset.xhtml?persistentId=doi:10.5683/SP2/9VSRHU) Canada, Scholars Portal Dataverse. V2 https://doi.org/10.5683/SP2/9VSRHU
**Guess, A., Nyhan, B., Lyons, B., & Reifler, J.** (2018). Why selective exposure to like-minded political news is less prevalent than you [think. Knight Foundation. https://kf-site-production.s3.amazonaws.com/media\\_elements/files/000/000/133/original/Topos\\_KF\\_White-](https://kf-site-production.s3.amazonaws.com/media_elements/files/000/000/133/original/Topos_KF_White-Paper_Nyhan_V1.pdf)Paper\_Nyhan\_V1.pdf
**Hallin, D. C., Mancini, P., Bennett, W. L., & Entman, R. M.** (2004). Comparing media systems: Three models of media and politics. [Cambridge University Press. http://ebookcentral.proquest.com/lib/concordia-ebooks/detail.action?docID=266614](https://ebookcentral.proquest.com/auth/lib/concordia-ebooks/login.action?returnURL=https%3A%2F%2Febookcentral.proquest.com%2Flib%2Fconcordia-ebooks%2Fdetail.action%3FdocID%3D266614)
**Hawthorne, J., Houston, J. B., & McKinney, M. S.** (2013). Live-tweeting a presidential primary debate: Exploring new political [conversations. Social Science Computer Review, 31\(5\), 552–562. https://doi.org/10.1177/0894439313490643](https://journals.sagepub.com/doi/10.1177/0894439313490643)
**Hochschild, A. R.** (1983). The managed heart: Commercialization of human feeling. University of California Press.
**Hochschild, A. R.** (2018). Strangers in their own land: Anger and mourning on the American right. The New Press.
**Hofstadter, R.** [\(1964, November\). The paranoid style in American politics. Harper's Magazine. http://harpers.org/archive/1964/11/the](https://harpers.org/archive/1964/11/the-paranoid-style-in-american-politics/)paranoid-style-in-american-politics/
**Hornsey, M. J., Harris, E. A., Bain, P. G., & Fielding, K. S.** (2016). Meta-analyses of the determinants and outcomes of belief in [climate change. Nature Climate Change, 6\(6\), 622–627. https://doi.org/10.1038/nclimate2943](https://www.nature.com/articles/nclimate2943)
**Hui, A., & Dickson, J.** (2020, October 22). Federal election 2019: Jagmeet Singh's comeback fails to materialize as NDP falls to fourth [party. The Globe and Mail. https://www.theglobeandmail.com/canada/article-federal-election-2019-jagmeet-singhs-comeback-fails-to](https://www.theglobeandmail.com/canada/article-federal-election-2019-jagmeet-singhs-comeback-fails-to-materialize/)materialize/
**Iyengar, S.** [\(2011\). The media game: New moves, old strategies. The Forum, 9\(1\). https://doi.org/10.2202/1540-8884.1425](https://www.degruyter.com/view/j/for.2011.9.1_20120105083457/for.2011.9.1/for.2011.9.1.1425/for.2011.9.1.1425.xml)
**Kaur R., & Sasahara K.** [\(2016\). Quantifying moral foundations from various topics on Twitter conversations. Proceedings of the 2016](https://ieeexplore.ieee.org/document/7840889) IEEE Big Data, 2505–2512. https://doi.org/10.1109/BigData.2016.7840889
**Kevins, A., & Soroka, S.** [\(2018\). Growing apart? Partisan sorting in Canada, 1992-2015.](https://doi.org/10.1017/S000842391700071) Canadian Journal of Political Science, 51(1), 103–133. https://doi.org/10.1017/S000842391700071
**Khan, F. H., Bashir, S., & Qamar, U.** [\(2014\). TOM: Twitter opinion mining framework using hybrid classification scheme. Decision](https://www.sciencedirect.com/science/article/pii/S0167923613002327?via%3Dihub) Support Systems, 57, 245–257. https://doi.org/10.1016/j.dss.2013.09.004
**Klein, O., & Muis, J.** [\(2019\). Online discontent: Comparing Western European far-right groups on Facebook. European Societies,](https://www.tandfonline.com/doi/full/10.1080/14616696.2018.1494293) 21(4), 540–562. https://doi.org/10.1080/14616696.2018.1494293
**Kowalski, R. M., Giumetti, G. W., Schroeder, A. N., & Lattanner, M. R.** (2014). Bullying in the digital age: A critical review and meta[analysis of cyberbullying research among youth. Psychological Bulletin, 140\(4\), 1073–1137. https://doi.org/10.1037/a0035618](https://psycnet.apa.org/search)
**Krafft, P.M., & Donovan, J.**(n/d). Disinformation by design: The use of evidence collages and platform filtering in a media manipulation campaign [Forthcoming]. Political Communication.
**Kreitzer, R., Kane, K., & Mooney, C.** [\(2019\). The evolution of morality policy debate: Moralization and demoralization.](https://www.degruyter.com/view/j/for.2019.17.issue-1/for-2019-0003/for-2019-0003.xml) The Forum, 17(1), 3–24. https://doi.org/10.1515/for-2019-0003
**Kukutai, T., & Taylor, J.** (Eds.). (2016). Indigenous data sovereignty: Toward an agenda (Vol. 38). Anu Press.
**Kuru, O., Pasek, J., & Traugott, M.** [\(2017\) Motivated reasoning in the perceived credibility of public opinion polls. Public Opinion](https://academic.oup.com/poq/article-abstract/81/2/422/3835465?redirectedFrom=fulltext) Quarterly, 81(2), 422–446. https://doi.org/10.1093/poq/nfx018
**Lalancette, M., & Raynauld, V.** [\(2017\). The power of political image: Justin Trudeau, instagram, and celebrity politics. American](https://journals.sagepub.com/doi/10.1177/0002764217744838) Behavioral Scientist, 63(7), 1–37. https://doi.org/10.1177/0002764217744838
**Lalancette, M.** [\(2018\). Les web-mises en scène des candidats aux élections québécoises de 2012: entre discrétion et confession.](https://www.erudit.org/fr/revues/ps/2018-v37-n2-ps03803/1048876ar/) Politique et sociétés, 37(2), 47-81. https://doi.org/10.7202/1048876ar
**Larsson, A. O.** [\(2019\). Right-wingers on the rise online: Insights from the 2018 Swedish elections. New Media & Society, 1–20. doi:](https://journals.sagepub.com/doi/10.1177/1461444819887700) 10.1177/1461444819887700
**Lee Kaid, L., Postelnicu, M., Landreville, K., Hyun Jung Yun, & LeGrange, A. G.** (2007). The effects of political advertising on [young voters. American Behavioral Scientist, 50\(9\), 1137–1151. https://doi.org/10.1177/0002764207300039](https://journals.sagepub.com/doi/10.1177/0002764207300039)
**Ling, J.** [\(2019, October 20\). How Maxime Bernier Lost His Seat. Foreign Policy. https://foreignpolicy.com/2019/10/20/canada-maxime](https://foreignpolicy.com/2019/10/20/canada-maxime-bernier-populism-peoples-party/)bernier-populism-peoples-party/
**Lipsitz, K.** [\(2018\). Playing with emotions: The effect of moral appeals in elite rhetoric. Political Behavior, 40\(1\), 47-78.](https://link.springer.com/article/10.1007%2Fs11109-017-9394-8) https://doi. org/10.1007/s11109-017-9394-8
**Loader, B. D., & Mercea, D.** [\(2011\). Networking democracy? Information, Communication & Society, 14\(6\), 757–769. https://doi.org/10](https://www.tandfonline.com/doi/abs/10.1080/1369118X.2011.592648) .1080/1369118X.2011.592648
**Lohmann, R. C.** [\(2014, January 28\). Trolling or Cyberbullying? Or Both? Psychology Today. http://www.psychologytoday.com/blog/](https://www.psychologytoday.com/ca/blog/teen-angst/201401/trolling-or-cyberbullying-or-both) teen-angst/201401/trolling-or-cyberbullying-or-both
**Lytvynenko, J., Silverman, C., & Chown Oved, M.** (2019, October 19) The Owner Of The Buffalo Chronicle Once Offered To Publish [Negative Political Stories For Money. BuzzFeed News. https://www.buzzfeednews.com/article/janelytvynenko/matthew-ricchiazzi](https://www.buzzfeednews.com/article/janelytvynenko/matthew-ricchiazzi-buffalo-chronicle-trudeau-claims)buffalo-chronicle-trudeau-claims
**MacPherson, A.** [\(2019, September, 23\) Dirty Tricks Behind Trudeau Brownface/Blackface Scandal? Free The Press Canada. https://](https://freethepresscanada.org/2019/09/23/dirty-tricks-behind-trudeau-brownface-blackface-scandal/) freethepresscanada.org/2019/09/23/dirty-tricks-behind-trudeau-brownface-blackface-scandal/
**Maher, S.** [\(2019, September 25\). Why the Liberals are outspending everyone on Facebook. Maclean's. https://www.macleans.ca/](https://www.macleans.ca/politics/ottawa/how-the-parties-will-try-to-reach-you-this-campaign/) politics/ottawa/how-the-parties-will-try-to-reach-you-this-campaign/
**Martens, K.** (2019, October 2). #Vote2019: Climate change and drinking water top Indigenous issues in federal election. APTN [National News. https://aptnnews.ca/2019/10/02/vote2019-climate-change-and-drinking-water-top-indigenous-issues-in-federal-election/](https://aptnnews.ca/2019/10/02/vote2019-climate-change-and-drinking-water-top-indigenous-issues-in-federal-election/)
**Marwick, A., & Lewis, R.** [\(2017\). Media manipulation and disinformation online. Data & Society Research Institute. https://datasociety.](https://datasociety.net/pubs/oh/DataAndSociety_MediaManipulationAndDisinformationOnline.pdf) net/pubs/oh/DataAndSociety\_MediaManipulationAndDisinformationOnline.pdf
**McCombs, M. E., & Shaw, D. L.** [\(1972\). The agenda-setting function of mass media. Public Opinion Quarterly, 36\(2\), 176–187. https://](https://academic.oup.com/poq/article-abstract/36/2/176/1853310?redirectedFrom=fulltext) doi.org/10.1086/267990
**Mead, D.** [\(2014, February 19\). People sure tweet "Fuck" a lot, finds science. Motherboard. https://motherboard.vice.com/en\\_us/](https://www.vice.com/en_us/article/8qxn8a/people-sure-tweet-fuck-a-lot-says-science) article/8qxn8a/people-sure-tweet-fuck-a-lot-says-science
**Merriam Webster.** [\(n.d.\). Definition of PRIVACY. Merriam Webster. Retrieved January 13, 2020, from https://www.merriam-webster.](https://www.merriam-webster.com/dictionary/privacy) com/dictionary/privacy
**Meyer, R.** [\(2017, February 3\). The rise of progressive "fake news." The Atlantic. https://www.theatlantic.com/technology/](https://www.theatlantic.com/technology/archive/2017/02/viva-la-resistance-content/515532/) archive/2017/02/viva-la-resistance-content/515532/
**Milner, R. M.** (2016). The world made meme: Public conversations and participatory media. MIT Press.
**Milner, R. M., & Phillips, W.** (2017). The ambivalent internet: Mischief, oddity, and antagonism online. Polity.
**Mina, A. X.** (2019). Memes to movements: How the world's most viral media is changing social protest and power. Beacon Press.
**Mishra, P., Yannakoudakis, H., & Shutova, E.** [\(2019\). Tackling online abuse: A survey of automated abuse detection methods. arXiv](https://arxiv.org/abs/1908.06024) preprint arXiv:1908.06024. https://arxiv.org/abs/1908.06024
**Munoz, C. L., & Towner, T. L.** [\(2017\). The image is the message: Instagram marketing and the 2016 presidential primary season.](https://www.tandfonline.com/doi/full/10.1080/15377857.2017.1334254) Journal of Political Marketing, 16(3–4), 290–318. https://doi.org/10.1080/15377857.2017.1334254
**Nadim M., & Fladmoe, A.** [\(2019\). Silencing women? Gender and online harassment. Social Science Computer Review. https://doi.](https://journals.sagepub.com/doi/10.1177/0894439319865518) org/10.1177/0894439319865518
**Newman, N., Fletcher, R., Kalogeropoulos, A., & Nielsen, R.K.** [\(2019\). Reuters Institute Digital News Report 2019. Reuters Institute.](http://www.digitalnewsreport.org/) http://www.digitalnewsreport.org/.
**[New Registry Requirements for Political Ads on Online Platforms.](https://www.elections.ca/content.aspx?section=pol&dir=regifaq&document=index&lang=e)** (n.d.). Elections Canada. https://www.elections.ca/content. aspx?section=pol&dir=regifaq&document=index&lang=e
**Nissenbaum, A., & Shifman, L.** (2018). Meme templates as expressive repertoires in a globalizing world: A cross-linguistic study. [Journal of Computer-Mediated Communication, 23\(5\), 294–310. https://doi.org/10.1093/jcmc/zmy016](https://academic.oup.com/jcmc/article/23/5/294/5075540)
**O'Callaghan, D., Greene, D., Conway, M., Carthy, J., & Cunningham, P.** (2014). An analysis of interactions within and between [extreme right communities in social media. ArXiv:1206.7050 \[Physics\]. http://arxiv.org/abs/1206.7050](https://arxiv.org/abs/1206.7050)
**Owen, D.** (2014). New media and political campaigns. In K. Kenski & K. H. Jamieson (Eds.), The Oxford handbook of political [communication \(Vol. 1, pp. 823–835\). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199793471.013.016\\_update\\_001](https://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780199793471.001.0001/oxfordhb-9780199793471-e-016)
**[Owen, T., Loewen, P., Ruths, D., Bridgman, B., Edelson, L., Gorwa, R., Jaim, D., MacLellan, S., McCoy, D., Merkley, E., Potter,](https://ppforum.ca/wp-content/uploads/2019/10/DDP-Research-Memo-6-Oct2019.pdf) A., Skazinetsky, B., Srivastava, R., & Zhilin, O.** (2019). Digital Democracy Project: Research Memo #6: Political Advertising. Public Policy Forum. https://ppforum.ca/wp-content/uploads/2019/10/DDP-Research-Memo-6-Oct2019.pdf.
**Paas-Lang, C.** (2019, October 3). Manning Centre a main financial backer for network of 'Proud' Facebook pages. Global News.
**Parmelee, J. H.** [\(2014\). The agenda-building function of political tweets. New Media & Society, 16\(3\), 434–450. https://doi.](https://journals.sagepub.com/doi/10.1177/1461444813487955) org/10.1177/1461444813487955
**Patriquin, M.** (2019). How Canada's far right is using anti-Muslim propaganda to target Trudeau. The Guardian. https://www. [theguardian.com/world/2019/oct/04/canada-far-right-anti-muslim-propaganda-targeting-trudeau.](https://www.theguardian.com/world/2019/oct/04/canada-far-right-anti-muslim-propaganda-targeting-trudeau)
**Peng, J.** [\(2019, October, 22\) Justin Trudeau won the election—But lost the meme war. The Star. https://www.thestar.com/](https://www.thestar.com/vancouver/2019/10/21/justin-trudeau-won-the-election-but-lost-the-meme-war.html) vancouver/2019/10/21/justin-trudeau-won-the-election-but-lost-the-meme-war.html
**Perrin, A. J., & McFarland, K.** [\(2011\). Social theory and public opinion. Annual Review of Sociology, 37\(1\), 87–107. https://doi.](https://www.annualreviews.org/doi/10.1146/annurev.soc.012809.102659) org/10.1146/annurev.soc.012809.102659
**Perry, B., & Scrivens, R.** [\(2016\). Uneasy alliances: A look at the right-wing extremist movement in Canada. Studies in Conflict &](https://www.tandfonline.com/doi/full/10.1080/1057610X.2016.1139375) Terrorism, 39(9), 819-841. https://doi.org/10.1080/1057610X.2016.1139375
**Perry, B.** (2017, September 13). Canada has seen an increase in right-wing extremism over the past year. Polic -makers and [practitioners seem ill-prepared to deal with it. Policy Options politiques, https://policyoptions.irpp.org/magazines/september-2017/the](https://policyoptions.irpp.org/magazines/september-2017/the-threat-of-right-wing-extremism/)threat -of -right - wing-extremism/.
**Pétry, F., & Bastien, F.** (2013). Follow the polster: Inaccuracies in media coverage of the horse-race during the 2008 Canadian [election. Canadian Journal of Political Science, 46\(1\), 1–26. https://doi.org/10.1017/S0008423913000188](https://www.cambridge.org/core/journals/canadian-journal-of-political-science-revue-canadienne-de-science-politique/article/follow-the-pollsters-inaccuracies-in-media-coverage-of-the-horserace-during-the-2008-canadian-election/BAF538901E2F8070440B84961224BA6F)
**Pew Research Center.** (2019, April 10). Share of U.S. adults using social media, including Facebook, is mostly unchanged since 2018. [Pew Research Center. https://www.pewresearch.org/fact-tank/2019/04/10/share-of-u-s-adults-using-social-media-including-facebook](https://www.pewresearch.org/fact-tank/2019/04/10/share-of-u-s-adults-using-social-media-including-facebook-is-mostly-unchanged-since-2018/)is-mostly-unchanged-since-2018/
**Phillips, W. (2015).** [This Is Why We Can't Have Nice Things: Mapping the Relationship between Online Trolling and Mainstream](https://www.jstor.org/stable/j.ctt17kk8k7) Culture. MIT Press. http://www.jstor.org/stable/j.ctt17kk8k7
**Piebiak, J. (2014).** Shit Harper Did: A community speaking truth to power? Concordia University.
**Pintak, Lawrence, Jonathan Albright, Brian J. Bowe, and Shaheen Pasha (2019).** #Islamophobia: Stoking Fear and Prejudice in [the 2018 Midterms. Social Science Research Council. https://www.ssrc.org/publications/view/islamophobia-stoking-fear-and-prejudice](https://www.ssrc.org/publications/view/islamophobia-stoking-fear-and-prejudice-in-the-2018-midterms/)in-the-2018-midterms/.
**Postill, J., & Pink, S.** [\(2012\). Social media ethnography: The digital researcher in a messy web. Media International Australia, 145\(1\),](https://journals.sagepub.com/doi/10.1177/1329878X1214500114) 123–134. https://doi.org/10.1177/1329878X1214500114
**Powell, A., Scott, A. J., & Henry, N.** (2018). Digital harassment and abuse: Experiences of sexuality and gender minority adults. [European Journal of Criminology, 147737081878800. https://doi.org/10.1177/1477370818788006](https://journals.sagepub.com/doi/10.1177/1477370818788006)
**Rabson, M.** [\(2019, September 7\). Environment minister now needs security as threats against Her Spike. HuffPost Canada. https://](https://www.huffingtonpost.ca/entry/catherine-mckenna-security-threats_ca_5d73d831e4b06451356ee2fe) www.huffingtonpost.ca/entry/catherine-mckenna-security-threats\_ca\_5d73d831e4b06451356ee2fe
**Raynauld, V., & Greenberg, J.** [\(2014\). Tweet, click, vote: Twitter and the 2010 Ottawa municipal election. Journal of Information](https://www.tandfonline.com/doi/abs/10.1080/19331681.2014.935840) Technology & Politics, 11(4), 412–434. https://doi.org/10.1080/19331681.2014.935840
[RCMP plans to charge Trudeau with obstruction in SNC Lavalin affair, following federal elections. \(2019, October 17\). The Buffalo](https://buffalochronicle.com/2019/10/17/rcmp-plans-to-charge-trudeau-with-obstruction-in-snc-lavalin-affair-following-elections/) Chronicle. https://buffalochronicle.com/2019/10/17/rcmp-plans-to-charge-trudeau-with-obstruction-in-snc-lavalin-affair-followingelections/
[RCMP source says 'security risk' against Trudeau was contrived by PMO staffers \(2019, October 17\) The Buffalo Chronicle. https://](https://buffalochronicle.com/2019/10/15/rcmp-source-says-security-risk-against-trudeau-was-contrived-by-pmo-staffers/) buffalochronicle.com/2019/10/15/rcmp-source-says-security-risk-against-trudeau-was-contrived-by-pmo-staffers/
**Reepschlager, A., & Dubois, E.** [\(2019, January 2\). New election laws are no match for the internet. Policy Options. https://](https://policyoptions.irpp.org/magazines/january-2019/new-election-laws-no-match-internet/) policyoptions.irpp.org/magazines/january-2019/new-election-laws-no-match-internet/
**Reitz, G.** (2014) Canada: New initiatives and approaches to immigration and nation building. In: Hollifield, J., Martin P. & Orrenius P. (Eds.) Controlling Immigration: A Global Perspective. Stanford University Press.
**Rheault, L., Rayment E., & Musulan A.** [\(2019\). Politicians in the line of fire: Incivility and the treatment of women on social media.](https://journals.sagepub.com/doi/10.1177/2053168018816228) Research & Politics, 6(1), 1-7. https://doi.org/10.1177/2053168018816228
**[Ribeiro, F. N., Saha, K., Babaei, M., Henrique, L., Messias, J., Benevenuto, F., Goga, O., Gummadi, K. P., & Redmiles, E. M.](https://dl.acm.org/doi/10.1145/3287560.3287580)** (2019). On microtargeting socially divisive ads. Proceedings of the Conference on Fairness, Accountability, and Transparency - FAT\* 19. Association for Computing Machinery https://doi.org/10.1145/3287560.3287580
**Riker, W. H., Calvert, R. L., Mueller, J. E., & Wilson, R. K.** (1996). The strategy of rhetoric: Campaigning for the American Constitution. Yale University Press.
**Rogers, R.** (2013). Digital Methods. MIT Press.
**Ryan, T.J.** [\(2017\). No compromise: Political consequences of moralized attitudes. American Journal of Political Science, 61, 409–423.](https://onlinelibrary.wiley.com/doi/abs/10.1111/ajps.12248) https://doi.org/10.1111/ajps.12248
**Ryan, T.J.** [\(2019\) Actions versus consequences in political arguments: Insights from moral psychology. Journal of Politics 81\(2\),](https://www.journals.uchicago.edu/doi/10.1086/701494) 426–440. https://doi.org/10.1086/701494
**Ryerson Leadership** [\(2019\). Rebuilding the Public Square. Ryerson Leadership Lab https://www.ryersonleadlab.com/rebuilding-the](https://www.ryersonleadlab.com/rebuilding-the-public-square)public-square.
**Sagi, E., & Dehghani, M.** [\(2014\). Measuring moral rhetoric in text. Social Science Computer Review, 32\(2\), 132–144. https://doi.](https://journals.sagepub.com/doi/10.1177/0894439313506837) org/10.1177/0894439313506837
**Scheufele, D. A., & Tewksbury, D.** [\(2007\). Framing, agenda setting, and priming: The evolution of three media effects models. Journal of](https://academic.oup.com/joc/article-abstract/57/1/9/4102624?redirectedFrom=fulltext) Communication, 57(1), 9–20. https://doi.org/10.1111/j.0021-9916.2007.00326.x
**Severson, A.W., & Coleman, E.A.** [\(2015\). Moral frames and climate change policy attitudes.](https://onlinelibrary.wiley.com/doi/abs/10.1111/ssqu.12159) Social Science Quarterly, 96, 1277–1290. https://doi.org/10.1111/ssqu.12159
**Shifman, L.** (2014). Memes in digital culture. The MIT Press.
**Silva, L., Mondal, M., Correa, D., Benevenuto, F., & Weber, I.** (2016). Analyzing the Targets of Hate in Online Social Media. AAAI Press.
**Silverman, C., Lytvynenko J. & Chown Oved, M.** (2019, October 29). The Buffalo Chronicle ran Facebook ads promoting dubious Trudeau [stories. BuzzFeed News. https://www.buzzfeednews.com/article/craigsilverman/facebook-let-the-buffalo-chronicle-run-ads-for-trudeau](https://www.buzzfeednews.com/article/craigsilverman/facebook-let-the-buffalo-chronicle-run-ads-for-trudeau)
**Slaughter, G.** (2019, October 12). Truth tracker: Conservative ads falsely say Liberals will legalize all drugs. CTV News. https://election. [ctvnews.ca/truth-tracker-conservative-ads-falsely-say-liberals-will-legalize-all-drugs-1.4636194](https://election.ctvnews.ca/truth-tracker-conservative-ads-falsely-say-liberals-will-legalize-all-drugs-1.4636194)
**Smith, D. E.** [\(2016\). Governing data and data for governance: The everyday practice of Indigenous sovereignty. Indigenous Data](http://press-files.anu.edu.au/downloads/press/n2140/pdf/ch07.pdf) Sovereignty: Toward an Agenda, 117–138. https://doi.org/10.22459/CAEPR38.11.2016.07
**Southern, R., & Harmer, E.** [2019. Twitter, incivility and 'everyday' gendered othering: An analysis of tweets sent to UK members of](https://journals.sagepub.com/doi/10.1177/0894439319865519) parliament. Social Science Computer Review, 1–17 https://doi.org/10.1177/0894439319865519
**Spenkuch, J. L., & Toniatti, D.** [\(2018\). Political advertising and election results. The Quarterly Journal of Economics, 133\(4\), 1981–2036.](https://academic.oup.com/qje/article/133/4/1981/4993157) https://doi.org/10.1093/qje/qjy010
**Stahel, L., & Schoen, C.** [\(2019\). Female journalists under attack? Explaining gender differences in reactions to audiences' attacks. New](https://journals.sagepub.com/doi/10.1177/1461444819885333) Media & Society. https://doi.org/10.1177/1461444819885333
**Steenkamp, M., & Hyde-Clarke, N.** [\(2014\). The use of Facebook for political commentary in South Africa. Telematics and Informatics,](https://www.sciencedirect.com/science/article/abs/pii/S0736585312000755?via%3Dihub) 31(1), 91–97. https://doi.org/10.1016/j.tele.2012.10.002
**Stier, S., Bleier, A., Lietz, H., & Strohmaier, M.** (2018). Election campaigning on social media: Politicians, audiences, and the [mediation of political communication on Facebook and Twitter. Political Communication, 35\(1\), 50–74. https://doi.org/10.1080/1058460](https://www.tandfonline.com/doi/full/10.1080/10584609.2017.1334728) 9.2017.1334728
**Subrahmanyam, K., Smahel, D., & Greenfield, P.** (2006). Connecting developmental constructions to the internet: Identity [presentation and sexual exploration in online teen chat rooms. Developmental Psychology, 42\(3\), 395–406. http://dx.doi.org.ezproxy.](https://cas.ryerson.ca/login?service=https%3a%2f%2flogin.ezproxy.lib.ryerson.ca%2flogin%3fqurl%3dezp.2aHR0cDovL2R4LmRvaS5vcmcvMTAuMTAzNy8wMDEyLTE2NDkuNDIuMy4zOTU-) lib.ryerson.ca/10.1037/0012-1649.42.3.395
**Sung, Morgan.** [\(2019, December 25\). "2019 was the year 'cancel culture' took on a gorgeously messy life of its own." Mashable.](https://mashable.com/article/cancel-culture-2019/) https://mashable.com/article/cancel-culture-2019/
**Tenove, C., Tworek, H. J. S., & McKelvey, F.** (2018). Poisoning democracy: How Canada can address harmful speech online. Public [Policy Forum. https://ppforum.ca/wp-content/uploads/2018/11/PoisoningDemocracy-PPF-1.pdf](https://ppforum.ca/wp-content/uploads/2018/11/PoisoningDemocracy-PPF-1.pdf)
**Trudeau accusor lands a seven-figure NDA to keep quiet about West Grey departure.** (2019, October 10). The Buffalo Chronicle.
**Trudeau is rumored to be in talks with an accusor to suppress an explosive sex scandal** (2019, October 7). The Buffalo [Chronicle. https://buffalochronicle.com/2019/10/07/trudeau-is-rumored-to-be-in-talks-with-an-accusor-to-suppress-an-explosive-sex](https://buffalochronicle.com/2019/10/07/trudeau-is-rumored-to-be-in-talks-with-an-accusor-to-suppress-an-explosive-sex-scandal-that-may-force-him-from-office/)scandal-that-may-force-him-from-office/
**Valentino, N. A., Hutchings, V. L., Banks, A. J., & Davis, A. K.** (2008). Is a worried citizen a good citizen? Emotions, [political information seeking, and learning via the Internet. Political Psychology, 29\(2\), 247–273. https://doi.org/10.1111/j.1467-](https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1467-9221.2008.00625.x) 9221.2008.00625.x
**Valenzuela, S., Piña, M., & Ramírez J.** (2017). Behavioral effects of framing on social media users: How conflict, economic, human [interest, and morality frames drive news sharing. Journal of Communication, 67\(5\), 803-826. https://doi.org/10.1111/jcom.12325](https://academic.oup.com/joc/article-abstract/67/5/803/4642206?redirectedFrom=fulltext)
**van Dijck, T. A.** (1998). Ideology: A multidisciplinary approach. Sage.
**Ward, S. J., & Mcloughlin, L.** (2017, April 26). Turds, traitors and tossers: The abuse of UK MPs via Twitter. ECPR Joint [Sessions 2017. ECPR Joint Sessions of Workshops, University of Nottingham. https://ecpr.eu/Events/PaperDetails.](https://ecpr.eu/Events/PaperDetails.aspx?PaperID=34203&EventID=104) aspx?PaperID=34203&EventID=104
**Wardle, C., & Derakhshan H.** [\(2017, October\). Information Disorder: An interdisciplinary framework. First Draft. https://firstdraftnews.](https://firstdraftnews.org/latest/coe-report/) org:443/latest/coe-report/
**Wardle, C.** [\(2018\). The need for smarter definitions and practical, timely empirical research on information disorder. Digital Journalism,](https://www.tandfonline.com/doi/full/10.1080/21670811.2018.1502047) 6(8), 951–963. https://doi.org/10.1080/21670811.2018.1502047
**Wasike, B.** [\(2017\). Persuasion in 140 characters: Testing issue framing, persuasion and credibility via Twitter and online news articles](https://www.sciencedirect.com/science/article/pii/S0747563216306720?via%3Dihub) in the gun control debate. Computers in Human Behavior, 66, 179–190. https://doi.org/10.1016/j.chb.2016.09.037
**Weedon, J., Nuland, W., & Stamos, A.** (2017). Information Operations and Facebook (No. 1). Facebook.
**Weeks, B. E.** (2015). Emotions, partisanship, and misperceptions: How anger and anxiety moderate the effect of partisan bias on [susceptibility to political misinformation. Journal of Communication, 65\(4\), 699–719. https://doi.org/10.1111/jcom.12164](https://academic.oup.com/joc/article-abstract/65/4/699/4082338?redirectedFrom=fulltext)
**Zine, J.** (2003). Dealing with September 12: Integrative antiracism and the challenge of anti-Islamophobia education. Orbit (Special Issue: Antiracism Education), 33(3), 14–16.
