DDP Research Memo #7: The Partisan Playground

Executive Summary

One of the goals of the Digital Democracy Project has been to track the digital media environment in order to identify potential threats to Canadian democracy. Our previous reports have shown that social media does not likely contribute to political polarization in Canada, and that Canadians generally trust the traditional news media, with even strong partisans more likely to engage with mainstream outlets on social media than media sites that cater to the fringes. Moreover, the more partisan and fragmented discourse found online is not necessarily reflected in the attitudes and beliefs of the mass population. Despite this positive news, our final report of this election period—which draws on data collected over the past three months—suggests that the information environment continues to have key vulnerabilities. Partisans, who are the most active citizens during election periods, tend to favour sharing and consuming content conducive to their existing worldviews and engage less with other perspectives.

Key Findings:

1. Social media activity around Canadian politics has surged since the election period began. Political activity is up about 800% on Twitter and 250% on public Facebook posts.

2. As partisanship increases, so does participation in politics, with the most politically active Canadians also being the most partisan.

3. Even when exposed to news coverage from a variety of perspectives, audiences are still more likely to choose content that supports their political views.

4. Exposure to politicized messaging tends to harden political views. Canadians tend to take stronger positions on key electoral issues when they are presented with statements in line with their views, but also when they are exposed to both sides of an argument.
5. There is a strong link between political efficacy—people’s feelings that they are able to take part in political life and that politicians will represent their interests—and their political participation. Those who lack this belief are more likely to be non-partisans, and less likely to share their political views or get involved in other political activities.

Survey and Social Media Findings

We have seen a consistent, massive increase in political activity on social media in the final weeks leading up to the election, particularly after the campaign period officially began on Sept. 11. The number of tweets and retweets using Canadian political hashtags has jumped eight-fold since the beginning of August. The number of users tweeting on these hashtags has grown by a similar amount. As of Oct. 13, there were 40,000 to 80,000 tweets per day on the eight most common Canadian political hashtags (e.g. #elxn43, #cdnpoli), shared by 15,000 unique users a day. On Facebook, there has been a 250% increase in the number of posts on public Canadian politics groups and pages since Sept. 11. This increase is driven partially by partisans who were active prior to the election, but also by newer or previously less-engaged voices that began talking about the election during the writ period.

Our survey found that self-identified partisans are far more likely to engage in political activities—including sharing their political opinions—compared to their non-partisan peers. Less than half of non-partisans (48%) said they had shared their political opinions in the past year, compared to 66% of left-partisans and 61% of right-partisans. Our social media data appears to bear that out. We looked at approximately 55,000 Twitter users who followed at least five rank-and file election candidates from the main six parties (excluding party leaders) and classified them as likely partisans of the party that includes most of the candidates they followed. These users posted more than four times as often as non-partisans on Canadian political hashtags.

One recurring question about the influence of social media platforms on democratic dialogue is whether algorithms serve users more of the content that reflects their views, making it less likely that they will encounter alternative perspectives. However, our survey found that even when presented with news items from a variety of perspectives—which would theoretically counter this effect—respondents were still more likely to pick the ones that aligned with their partisan beliefs.

Our Twitter data also suggested some degree of selective news consumption by partisans. We looked at the top two media links shared by Liberal, NDP, Conservative and Bloc Québécois likely partisans from Aug. 1 to Oct. 13. While Research Memo 3 showed that partisans across the board share posts from similar media organizations, we find here that the individual stories they share from those outlets vary drastically. The mostshared stories for each party are in line with that party’s positions, and each is shared by an overwhelming majority of supporters and disproportionately few political opponents. So while partisans may be consuming news from similar outlets, they could still get a different picture of political events through their social media feeds.

Further complicating matters, our survey experiment suggests that exposure to any political messaging at all serves to intensify peoples’ opinions, hardens their views, and makes them increasingly likely to participate in political activity. If partisans are drawn to information that supports their views, and exposure to information that supports their views reinforces them more strongly, it could contribute to an escalating cycle of partisanship, whereby those who consume the most political information become the least moderate and potentially less likely to consider alternative views. This provides an explanation for one of our findings from Research Memo 1: partisans select into their own facts aided by exposure to mainstream and social media news consumption.

Partisans are also more likely than non-partisans to show political efficacy: confidence in their ability to participate in democratic life (internal efficacy) and confidence that politicians are responsive to their values and interests (external efficacy). Our survey found that non-partisans have notably lower levels of internal efficacy (0.45 out of 1) compared to Liberal (0.55), Conservative (0.57) and NDP partisans (0.55); they are also less externally efficacious (0.38) than Liberal (0.52), Conservative (0.42) and NDP partisans (0.43).

This matters because people who don’t feel confident about taking part in politics, and don’t believe politicians are working for them, are more likely to opt out of more expressive forms of democratic participation altogether. We found a robust link between political efficacy and political participation. Survey respondents with higher levels of both internal and external efficacy were more likely to have taken part in political activities over the past year. When it came to sharing political opinions, 56% of those with low external efficacy said they had done it in the past year, compared to 62% for those with high external efficacy. The difference was more stark for internal efficacy: 53% of respondents with low internal efficacy shared their political opinions, compared to 69% with high internal efficacy.

This week’s findings paint a picture of a media environment where partisans are more willing to share their views and choose media that reinforces their beliefs, while less-partisan Canadians are less inclined to share their views and take part in political life. The worst-case outcome is that the online political discourse becomes dominated by committed partisans further entrenching their positions, while more moderate voices opt out.

Are there ways to make sure more voices are heard in the political conversation, and not just those with a vested interest in a certain outcome? Looking beyond this election campaign, this will be one of the important questions facing technology companies, policymakers and researchers.

Methodology

Our survey data team conducted an online panel survey of 1,545 Canadian citizens 18 years and older using the online sample provider Qualtrics. The sample was gathered from Oct. 4-13. Data was weighted within each region of Canada by gender and age to ensure it adequately represented the Canadian public. Survey respondents were asked questions related to basic demographics, as well as their partisan, ideological and issue preferences. We present 90% confidence intervals for each of our figures below.

Since June, our online data team has been collecting data on the online political participation of Canadians across a number of platforms. On Twitter, we collected approximately 20 million tweets and retweets from Aug. 1 to Oct. 13 from four categories of users: candidates, journalists, third parties, and the public at large. We have also been collecting information on retweets, follows and mentions. On Facebook, we used CrowdTangle to track more than 2,200 public groups and pages that focus on Canadian politics, including pages for all candidates, official news outlets, and many discussion groups and pages. While this report draws on data from Twitter and Facebook, findings from data collected on Reddit, YouTube, Tumblr, Instagram and 4Chan will be shared in post-election reporting.