Extraordinary changes in communications technology over the past two decades have generated a fragmentation of the political news environment in North America. A landscape dominated by print and broadcast television news was challenged first by cable television, which often catered to narrow partisan audiences (Ladd Reference Ladd2011), and then the Internet, which lowered the cost of news production for alternative news outlets not as interested in investing in credibility as traditional news organizations (Munger Reference Munger2020). The added competition from freely available online news sources, which often cater to niche, partisan audiences, has put pressure on traditional news outlets, especially newspapers and magazines.
As a result, the partisan news ecosystem in the United States is now more than just Fox News and MSNBC with the rise of online sources like Breitbart, the Blaze, Raw Story, and Daily Kos. This change is not unique to the United States. We have seen some growth of these outlets in Canada as well, most prominently on the right, with the likes of True North News and Ezra Levant’s Rebel News. News consumers have far greater choice now about what news to consume as mainstream news struggles and alternative, often partisan, news grows at their expense.
A media ecosystem increasingly dominated by partisan media outlets untethered to traditional journalistic norms is one that will produce more polarizing, inflammatory content. Partisan media, after all, traffic in outrage (Mutz Reference Mutz2015; Sobieraj and Berry Reference Sobieraj and Jeffrey M2011). Mass polarization could be an important consequence of these developments. Experimental evidence has shown that exposure to partisan news can lead to more extreme policy beliefs (Levendusky Reference Levendusky2013). There is also compelling real-world evidence of these effects, especially as they pertain to Fox News. Fox News viewers typically take on more conservative positions, like hostility towards immigration (Gil de Zúñiga et al. Reference Gil de Zúñiga, Correa and Valenzuela2012; Hoewe et al. Reference Hoewe, Cramer Brownell and Wiemer2021) and climate denialism (Carmichael et al. Reference Carmichael, Brulle and Huxster2017; Feldman et al. Reference Feldman, Maibach, Roser-Renouf and Leiserowitz2012). Other work has been able to causally identify relationships. A field experiment that randomly induced Fox News viewers to switch to CNN saw considerable moderation in their attitudes (Broockman and Kalla Reference Broockman and Kalla2025). Others have successfully exploited as-if randomized variation in Fox News’s channel position as a proxy for exposure to connect it to higher Republican vote shares (Martin and Yurukoglu Reference Martin and Yurukoglu2017) and lower COVID-19 vaccination rates (Pinna et al. Reference Pinna, Picard and Goessmann2022).
Not everyone agrees about the polarizing potential of partisan news. Objections to this theory take on two forms. First, there is disagreement about whether partisan media truly polarize conditional on exposure in real world settings. Already polarized individuals may gravitate towards partisan news, known as selective exposure (Stroud Reference Stroud2011), so the effects of these news sources could be limited by the fact their consumers hold attitudes close to a ceiling. Experimental evidence shows that, in practice, less polarized individuals tend to opt out of partisan news exposure, mitigating its polarizing potential (Arceneaux and Johnson Reference Arceneaux and Johnson2013). Studies linking partisan media exposure to polarization (and other outcomes) using trace data have typically exhibited null results (Guess et al. Reference Guess, Barberá and Munzertd2021; Peterson et al. Reference Peterson, Goel and Iyengar2021; Wojcieszak et al. Reference Wojcieszak, Sjifra de Leeuw, Lee, Ke M and Weeks2023).
Second, and more relevant for this note, is that there is considerable evidence that exposure to partisan media is quite limited. This observation is long standing, with Neilson data showing that the share of Americans watching at least 10 minutes of partisan cable news per day was, at most, 10-15%, while most cable news watchers split their viewing time between left- and right-wing outlets: Fox News viewers watching CNN and vice versa, for instance (Prior Reference Prior2013, pg. 113). Although the rise of the Internet and online partisan media could potentially complicate this picture, Guess (Reference Guess2021) finds that audiences for online partisan news are small (though intense) and as a result there is considerable overlap (50-65%) in the media diets of Republicans and Democrats. Muise et al. (Reference Muise, Homa Hosseinmardi, Mobius, Rothschild and Duncan J2022) find that only 4% of Americans are “partisan-segregated” in their online news consumption, a figure that was closely matched by Fletcher et al. (Reference Fletcher, Robertson and Kleis Nielsen2021). Heseltine et al. (Reference Heseltine, von Hohenberg, Menchen-Trevino, Gackowski and Wojcieszak2024) find that two-thirds of online political news exposure is to centrist websites. To some degree, it is a subjective evaluation as to whether the prevalence of partisan media exposure or echo chambers is large or not. Broockman and Kalla (Reference Broockman and Kalla2024), for instance, argue that 15% of Americans watching 8 hours of partisan cable news per month is substantial, but that is a point on which reasonable people could disagree. There is a consensus, however, that only a minority of citizens consume partisan news, and an even smaller minority do so regularly. Consequently, media diets are more balanced than we would expect, and echo chambers are small. The degree to which this is true in Canada remains to be seen.
Measuring the scale and scope of the partisan media echo chamber phenomenon is no easy task. Self-reported survey data suffer serious flaws. People tend to overreport their news consumption, especially for partisan-aligned outlets (Prior Reference Prior2013), and the ability of people to accurately recall their news consumption is poor as well (Dilliplane et al. Reference Dilliplane, Goldman and Mutz2013; Guess Reference Guess2015). This makes it difficult to credibly measure the size of partisan media audiences, let alone understand its determinants and consequences. The principal alternative is to use behavioural data that allow researchers to directly observe the links clicked on by survey respondents, which includes links to news media outlets (see Guess Reference Guess2021). These data are expensive and hard to come by, however. As a result, we do not have a lot of quantitative evidence on the nature of online news audiences outside of the United States (see Stier et al. Reference Stier, Kirkizh, Froio and Schroeder2020 for an important exception), let alone the determinants or consequences of these news consumption patterns.
In this note, I aim to quantify the size of partisan news audiences in Canada, both in absolute terms and relative to mainstream news audiences, using behavioural data collected by the 2019 Digital Democracy Project. This dataset provides unique tracking of the online behaviour of over 700 participants in a four-week span during the 2019 election and includes over 2.2 million website visits. This tracking data is paired to a survey for a subset of participants. I use these data to 1) describe the absolute and relative size of the online partisan media audience; 2) show how these audiences vary by partisan identity; and 3) describe correlations of partisan media use with factors like political sophistication and attitudinal strength.
Canada is a useful test case for the partisan media echo chamber hypothesis. Canada has a similar profit-driven media system as the United States, albeit with a public broadcaster and a parallel French-language media ecosystem based in Quebec. It fits the mold of a liberal, Western or hybrid media system depending on the classification scheme (Brüggemann et al. Reference Brüggemann, Sven Engesser, Humprecht and Castro2014; Hallin and Mancini Reference Hallin and Mancini2004; Humprecht et al. Reference Humprecht, Laia Castro Herrero, Brüggemann and Engesser2022). Canada’s print sector has been in a similarly steep decline (Lindgren and Corbett Reference Lindgren and Corbett2018; Lindgren et al. Reference Lindgren, Corbett and Hodson2017). Its political parties are polarizing in ideological terms (Cochrane Reference Cochrane2015), while partisan supporters in the electorate have followed suit (Merkley Reference Merkley2023). Perhaps as a result, we see growing affective polarization in Canada between partisans of the left-leaning parties, like the Liberals and NDP, and the principal party of the right, the Conservatives (Johnston Reference Johnston2023). Further, trust in mainstream news is declining in Canada (Bridgman et al. Reference Bridgman, Alexei Abrahams, Thomas Galipeau, Haaya Naushan, Sara Parker, Jennie Phillips, John Loewen and Owen2023). There is a potential market for partisan-slanted news.
On the supply side, Canadians have access to the highly fragmented U.S. media ecosystem. Canadians, especially in English Canada, are exposed to a tremendous amount of news from the United States due to shared language and culture, close-proximity, and population imbalance. Social media further facilitates this access, particularly to online partisan news sites. Some recent work has shown that most COVID-19 misinformation shared on Twitter in Canada originated in the U.S., while people who consumed U.S. news and used social media were more misinformed about COVID-19 (Bridgman et al. Reference Bridgman, Eric Merkley, John Loewen, Owen and Ruths2021). Canadians might not need domestic outlets to get their fix for partisan news. They might turn to U.S. sources as well.
Data
I use data that I collected as part of the 2019 Digital Democracy Project. We tracked online media use over a four-week period during the 2019 Canadian federal election for a sample of respondents from Qualtrics’ online panel. The study initially aimed to collect four weeks of tracking data—staggered over that two-month span—for 2,000 respondents, matched to surveys conducted before and after the tracking period. The panel provider was not able to deliver on the project, resulting in fewer participants (N=728) and an even smaller sample where tracking data could be matched to surveys (N=273). Nevertheless, the resulting dataset involves over 2,213,410 website visits (excluding duplicates) with metadata indicating the type of link (e.g., news), the amount of time spent on a page, the date and time of the visit, and the device used, among other things. We can directly observe the news and social media applications our respondents were using over the tracking period.Footnote 1
I matched the web domains of a list of prominent Canadian, American, and international news sources (including partisan media) to a website domain variable included in the data set.Footnote 2 I manually excluded false positives from the matching procedure. I take an expansive view of news exposure, preserving use of news sources for forms of soft news (e.g., entertainment, lifestyle, and sports news), but I exclude non-news uses of these web domains. For example, while the web domain nbcnews.com implies some form of news exposure when it is visited, the same cannot be said for the cbc.ca, which includes pages related to its television and cultural programming—most notably its NHL Hockey Night in Canada stream.Footnote 3 I use the domain paths to exclude such visits. I exclude duplicate visits by removing duplicate URL visits within the same minute. Results are substantively similar when duplicate visits are not removed from the data set (see Clemm von Hohenberg et al. Reference Clemm von Hohenberg, Stier, Cardenal, Guess, Menchen-Trevino and Wojcieszak2024). I preserve only respondents who visited an average of at least one website per day over the tracking period.
80% of respondents visited what was classified as a news website over the four-week period. However, these visits only constituted 2.3% of overall traffic, which amounted to roughly 51,679 website visits. This is an exceedingly liberal estimate of news consumption, let alone political news consumption, because it includes websites like the Weather Network and those dedicated to sports and lifestyle news as well. The distribution of web traffic is also very far from equitable. 1% of respondents are responsible for 23% of news traffic (N=7). 10% of respondents are responsible for 69% of traffic (N=73). These figures are even more striking when focused on my list of prominent American, Canadian, and international sources. 1% of respondents were responsible for 34% of visits to these websites, while 10% of respondents were responsible for 78% of visits. Canadians are not voracious news consumers, far from it, and patterns of news consumption are highly inequitable.
Results
Figure 1 provides the share of respondents who visited domestic and international (left) and American news sites (right) over the four-week tracking period. CBC, CTV, and Global lead the way, but other online news aggregators like Yahoo! are also quite important.Footnote 4 The major newspapers, the National Post, Globe and Mail and Toronto Star, cluster in the middle, alongside international outlets like the BBC, Huffington Post (which at this time had a Canadian branch), the Daily Mail, and the Guardian. Local newspapers, which include major regional papers like the Waterloo Region Record, Calgary Herald, and the Saskatoon StarPheonix were read at some point by 20% of respondents.
Share of respondents using domestic, international (left) and American (right) online news sources over the four-week tracking period. TVA and TV5 are Quebec-based, French language broadcasters. The CBC category includes visits to Radio-Canada. Local newspaper includes: Vancouver Sun, The Province, Victoria Times-Colonist, Calgary Herald, Calgary Sun, Edmonton Sun, Edmonton Journal, Saskatoon Leader-Post, Regina Star-Pheonix, Winnipeg Sun, Winnipeg Free Press, Ottawa Sun, Ottawa Citizen, Waterloo Region Record, Toronto Sun, London Free Press, The Guardian (PEI), Telegraph-Journal (NB), Chronicle Herald (NS), The Telegram (NL), Le Devoir, La Presse, Journal de Montreal, Journal de Quebec, and the Montreal Gazette.
Domestic partisan media received virtually no readership. Only Rebel Media and the National Observer received any traffic at all over the monthly tracking period, reaching around 1% of the sample in each case (N=7-10). 0.4% of the sample is responsible for almost 70% of the visits to these two websites (N=3). No one in the sample relied on domestic partisan media for more than half of their news. Domestic partisan media has little foothold in Canada’s news media ecosystem.
However, social media provides a gateway to the American partisan media ecosystem as well (Bridgman et al. Reference Bridgman, Eric Merkley, John Loewen, Owen and Ruths2021).Footnote 5 The right panel of Figure 1 provides a list of major national American news outlets that encompass broadcast (e.g., ABC), cable news (e.g., CNN), print (e.g., New York Times), and online partisan sources (e.g., Breitbart). American news consumption is not particularly high. CNN (14%) is the most used website, followed by the Washington Post (8%) and the New York Times (8%). Fox News follows behind at 5%. All other listed outlets fall below 5% over the four-week period.
Aside from Fox News, U.S. partisan news sites are hardly visited. Of 28 websites covered here, Rolling Stone was the most viewed, at 2% of the sample. 10 of the 28 websites received no visitors at all. When counting all domestic and American partisan news, including Fox on the right, and MSNBC on the left, 14% of the sample visited a partisan news site over the four-week period, but just over 50% of visits come from 0.4% of the sample (N=3), and 79% from 1% (N=7). Only 1% of respondents got more than half of their news from partisan news sources of any type, among those that consumed any news at all (N=6).
The three big broadcasters (i.e., CBC, CTV, and Global) have almost 3 times (14 times) the traffic as the 14 right-wing (left-wing) partisan news sites identified here. Their overall reach is approximately 6 times that of right-wing and left-wing partisan news sites. Partisan news consumption is highly concentrated in a very small minority of individuals. An overwhelming majority of both news traffic and viewership is directed to mainstream news outlets, and that is before accounting for the reach of these outlets through print readership and television viewership.
How do these findings compare to self-reported media use measures? I use a series of nationally representative surveys conducted by the Media Ecosystem Observatory (MEO) from 2020 to 2021 to provide a benchmark for comparison.Footnote 6 Respondents were asked to report their exposure to domestic partisan news sources over the past week in these surveys. In a subset of these surveys, conducted between June and August 2020, they were also asked to report exposure to U.S. outlets, which included partisan outlets like Fox News and MSNBC. About 15% of respondents reported being exposed to an aligned partisan media outlet in the past week, compared to the 10% of our tracked respondents who visited such sites over a four-week period. Self-reported exposure to partisan news is likely exaggerated.
It is possible that visits to partisan media, while concentrated among a very small segment of the population, are intensely visited (Guess Reference Guess2021). The top panels of Figure 2 provide the average number of visits to news sources over the tracking period among those who visited a given site at least once. The left panels focus on domestic and international sources, while the right panels feature U.S. outlets. For domestic and international sources, news aggregators like MSN (127 visits), Yahoo! (48) and Google News (28) lead the list, while Rebel News (12) falls in the middle of the pack and the National Observer trails considerably (2). We see the same general pattern for average duration of exposure, which are displayed in the bottom panels of Figure 2. Broadcasters (perhaps owing to video content) and news aggregators lead the pack, while Rebel News falls towards the middle and the National Observer trails at the bottom of the list. Traffic to domestic partisan media sources among those who visited these sites over the tracking period is not notably more intense than for other news sources.
Average number of visits to (top) and average visit duration in minutes (bottom) for domestic, international (left) and American (right) online news sources over the four-week tracking period, conditional on respondents visiting the site. TVA is a Quebec-based, French language broadcaster. The CBC category includes visits to Radio-Canada. Local newspaper includes: Vancouver Sun, The Province, Victoria Times-Colonist, Calgary Herald, Calgary Sun, Edmonton Sun, Edmonton Journal, Saskatoon Leader-Post, Regina Star-Pheonix, Winnipeg Sun, Winnipeg Free Press, Ottawa Sun, Ottawa Citizen, Waterloo Region Record, Toronto Sun, London Free Press, The Guardian (PEI), Telegraph-Journal (NB), Chronicle Herald (NS), The Telegram (NL), Le Devoir, La Presse, Journal de Montreal, Journal de Quebec, and the Montreal Gazette.
We see more evidence that exposure intensity is higher for U.S. partisan media sources, at least compared to mainstream U.S. sources. Visits to the Drudge Report lead the list by a considerable margin (142, top-right panel), but this figure is based on only 4 respondents. Fox (45) and CNN otherwise lead the pack (38). Other partisan sources occupy the top of the list such as Raw Story (38), Infowars (18) and Daily Kos (7), but we need to place these figures in context. Partisan sources trail news aggregators considerably, like MSN (127) and are comparable to broadcasters like the CBC (18), CP24 (17), and newspapers like La Presse (40).
The same is true when observing duration of exposure over the tracking period. U.S. partisan media tend to occupy the top of the American list, like Fox (128 minutes), Salon (39), and the Daily Caller (37), but are still in line with, or even trail, prominent mainstream domestic sources, like the CBC (91), Global (53), and the Globe and Mail (37). With a few exceptions, partisan news outlets are not more intensely used by respondents than mainstream domestic sources. In fact, duration of exposure over the tracking period is higher, on average, across the mainstream outlets I evaluate (45 minutes, excluding news aggregators) compared to partisan media outlets (32 minutes), which is true for average number of visits as well (42 vs. 8).
How does exposure to news sources vary by partisanship? We should expect sharp differences in source exposure, with Conservatives preferring right-wing outlets and Liberal and NDP supporters preferring left-wing news sites if political attitudes are the primary driver of partisan media use. I focus my attention here on respondents for whom I have matched survey data and specifically on the difference between Liberal and NDP partisans and voters (left) on the one hand, and Conservatives on the other (Right).Footnote 7 I have a sample of 131 Liberal and NDP supporters, along with 90 Conservatives.
Table 1 displays the partisan differences in online readership of partisan news.Footnote 8 We see very little differentiation. The biggest difference is for Fox News, where just over 6% of Conservatives visited the site in a 4-week period, but almost 4% of Liberal and NDP supporters did the same. 2% of Conservatives visited Rebel News, compared to 0.8% of Liberal and NDP supporters. In most cases, partisan division reflects the slant of the outlet, but there are some exceptions. For example, no Conservatives visited Breitbart, which was visited by 0.7% of Liberal and NDP supporters. Unexpected partisan gaps are denoted in the table with italics.Footnote 9
Partisan differences in online partisan news exposure
Note: difference is expressed as an absolute value. “Wrong-signed” partisan differences denoted with italics.
The average share of a Conservative Party supporter’s news media diet dedicated to right-wing partisan sources is only 4%. This figure for Liberal and NDP partisans, in regards to their consumption of left-wing partisan news, is an even lower 1.4%. Only a single Conservative respondent received more than half of their news from partisan media outlets. The same was true for Liberal and NDP supporters for equivalent left-wing sources. In short, there is considerable overlap in the news media diet of left and right in Canada and partisan media has limited relevance for all but a tiny sliver of Canadians.
Again, it is possible that partisan differences will emerge mostly in the intensity of one’s partisan media use. Few people may be exposed to partisan media, but those that are may be very intense users, and that intensity may exhibit more obvious partisan differences than exposure as displayed in Table 1. Perhaps Liberals do visit Fox News, but they are unlikely to do so frequently, or for a considerable amount of time, compared to Conservatives.
Table 2 displays partisan differences in average number of visits (top panel) and average duration of exposure (bottom panel) conditional on respondents having visited the website at all. We see some expected partisan differentiation in news source visits, but these differences are rather modest, aside from a handful of sources (Raw Story, Drudge, and the Daily Caller). And again, there are unexpected partisan gaps. Most notably, Liberal and NDP supporters visited Fox News more frequently than Conservatives (i.e., 64 vs. 52 visits).
Partisan differences in number of visits and duration of exposure conditional on source being visited
Note: difference is expressed as an absolute value. “Wrong-signed” partisan differences denoted with italics.
The duration results in the bottom panel more clearly conform to expectations. Conservatives visited right-wing sites longer than Liberal and NDP supporters and vice versa for left-wing sites. There is one notable and surprising exception, Rebel News, though we must keep in mind the small sample size. The other exceptions are rather trivial. Although Liberal and NDP supporters spent more time on Info Wars than Conservatives, for example, both partisan groups kept their visits extremely brief on average.
All told, partisan differences in partisan media use are not as stark as we might expect. But the table above masks an important asymmetry. Conservatives visited the average right-wing news site 10 times for 58 minutes (conditional on the source receiving a visit over the four-week period), compared to 2 times for 2 minutes for left-wing sources. In contrast, Liberal and NDP supporters visited the average left-wing site 12 times for 25 minutes, compared to 10 times for 23 minutes for right-wing sites. Conservative respondents appear to be more sensitive to the slant of partisan news.Footnote 10
Correlates of Partisan News Consumption
We can also use these data to evaluate the correlates of partisan news. I make no claims to causality here. Rather, I aim to describe the characteristics of those that consumed partisan media over the tracking period. I run a series of bivariate regression models predicting partisan media use with four categories of correlates: political intensity, political attitudes, political sophistication, and media consumption.Footnote 11
Political intensity measures include partisan strength and ideological extremity. People with strong partisan attachments or extreme ideological commitments may prefer attitude-reinforcing content from partisan news because of selective exposure (Iyengar and Hahn Reference Iyengar and Hahn2009), or partisan media may itself be a cause of these strong attitudes (Levendusky Reference Levendusky2013). For political attitudes, I use populism and conspiratorial thinking. The former relates to the tendency to view politics as a conflict between a virtuous common citizenry and a corrupt elite (Mudde Reference Mudde2004), while the latter refers to the propensity of some to imagine political events are shaped by conspiracies of sinister political and economic elites. These two dimensions have recently been combined into a construct called anti-establishment orientation (Uscinski et al. Reference Uscinski, Adam M, Michelle I, Casey A, John R, Everett, Wuchty, Premaratne and Manohar N2021). I also use left-right ideology because previous work has shown partisan media consumption in the U.S. to be more common on the political right (Guess Reference Guess2021), perhaps owing to their greater distrust of mainstream news, which I also use as a final covariate in this category.
For political sophistication I use political interest and political knowledge. People who are invested in politics may be more likely to take the time and energy to seek out alternative news sources, while partisan outlets may carry more of the political content they crave. And finally, I use domestic media exposure and social media use. I have no expectation that either variable causes or is caused by partisan media, but these correlations will shed light on whether these media substitute for, or are complimentary to, partisan media use.
In Figure 3 we see surprisingly little connection between attitude strength and partisan media use. Neither partisan strength (top-left panel), nor ideological extremity, is correlated with partisan media use (top-centre). Partisan media use is not significantly higher for populists (bottom-left), the conspiratorial minded (bottom-centre) and those that distrust the media (bottom-right), nor is it higher on the political right (top-right).Footnote 12
Predicted partisan media use by partisan strength (top-right), ideological extremity (top-centre), political ideology (top-right), populism (bottom-left), conspiratorial thinking (bottom-centre), and media distrust (bottom-right). 95% confidence intervals.
Figure 4, on the other hand, shows that partisan media use rises between 0.21 and 0.23 points across levels of political interest and knowledge (top-right, p=0.010 & centre-left, p=0.013). We see associations with domestic (0.31, p<0.001, bottom-centre) and social media use (0.13, p=0.013, bottom-right).Footnote 13 There is more evidence that partisan media use in Canada is associated with general political awareness and sophistication than intense or extreme political attitudes. We also see general symmetry in partisan media use between the left and right.Footnote 14
Predicted partisan media use by political knowledge (top-right), political interest (top-right), domestic news consumption (bottom-left), and social media use (bottom-right). 95% confidence intervals.
How similar are these correlations to those produced with self-reported exposure to partisan news? Tables S4 and S5 in the Supplementary Materials replicate these correlations using the MEO data set. We see close correspondence in the effects of political knowledge, political interest, news exposure, and social media use. But estimated correlations are higher for partisan strength, conspiratorial thinking, and left-right ideology using the survey self-reports. We must be cautious in interpreting these differences because of the small sample size for the behavioural data, but nonetheless these results suggest that survey self-reports overstate the link between partisan news exposure and characteristics associated with political polarization.Footnote 15
Discussion
Canadians are not voracious news consumers. Partisan news exposure is far less common, and the vast majority of traffic to these websites is driven by a tiny sliver of the population—and this includes U.S. based sources that are easily accessed by Canadians. There is surprisingly little partisan differentiation in exposure to these news sites. The left and right in Canada mostly read online news from the same outlets. Also interesting is the lack of significant association between partisan media use in Canada and right-wing ideology, anti-establishment orientations, and media distrust. Nor is there a significant connection to political attitude strength as measured by factors like partisan strength and ideological extremity. Rather, the people that use partisan media are mostly politically sophisticated individuals who generally consume a lot of news media in general.
These results compliment previous work in the U.S. showing the limited reach of partisan news in the United States (Guess Reference Guess2021) and other cross-national work showing that the connection between partisan media use and political attitudes like populism vary depending on the media system—being stronger in countries with more robust partisan media ecosystems like the U.S. (Stier et al. Reference Stier, Kirkizh, Froio and Schroeder2020). But these findings are also somewhat at tension with literature on partisan media and selective exposure that would expect partisan news use to be higher among strong partisans and the highly ideological since they should have stronger motivation to seek out politically aligned news. We see little of this in Canada. This echoes findings by Stier and his colleagues (Reference Stier, Kirkizh, Froio and Schroeder2020) who argue that the correlates of partisan media use (and other forms of media use) may depend on the media system. What is true in the United States may not hold elsewhere.
There are obvious implications of these findings for understanding political polarization. Canada is a case where partisans are affectively polarizing between Conservatives, on the one side, and Liberal and NDP supporters, on the other (Cochrane Reference Cochrane2015; Johnston Reference Johnston2023). At the same time, these partisan groups have diverged from one another in their beliefs and ideology (Merkley Reference Merkley2023). These findings suggest partisan media is not an important part of this story. That is not to say that nothing about the media environment polarizes Canadians. Mainstream news can polarize indirectly by conveying the polarized stances of political elites to the mass public (Zaller Reference Zaller1992) or by framing political issues in sensationalistic and conflict-laded ways to increase consumer interest (Bennett Reference Bennett2008), thereby making citizens believe society is more polarized than it really is (Levendusky and Malhotra Reference Levendusky and Malhotra2016). We need more research on how different dimensions of media use influence polarization in Canada.
There are important limitations to these data. The sample size, particularly for the matched survey data, is not ideal. We cannot rule out small correlations between partisan media use and my null covariates (e.g., ideological extremity). It is not an entirely representative sample. The individuals participating in the survey tend to be much more “online” than the average person. For instance, 67% of the sample used Facebook at some point over the tracking period, while 52% used Twitter. The sample also skews male, native-born, and English Canadian, though it is otherwise relatively representative across age, education, and vote choice (Table S1). We also cannot use these data to observe what news respondents might see or hear on television or radio, while these data primarily involve web visits on one’s browser, rather than mobile devices.
Nevertheless, there are obvious advantages to directly observing what news respondents read given problems with measures of self-reported media consumption. Indeed, we see some indication that self-reported measures inflate partisan news consumption, especially among those with strong political commitments. And, if anything, these data provide a liberal estimate of partisan news consumption since most of these outlets are online and find their audiences though social media. Partisan news exposure is almost certainly lower in the general population.
We need more research on partisan news audiences outside of the United States—particularly using behavioural data—before we can hope to understand the cross-national consequences of this innovation in the media environment, like polarization and political extremism. At least in Canada, it appears that the causes of affective polarization lie elsewhere.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0008423925000198.
Acknowledgements
Thanks to those who provided feedback at the 2024 Canadian Political Science Association annual meeting and to the 2019 Digital Democracy and 2021 Media Ecosystem Observatory teams, including Peter Loewen, Taylor Owen, Derek Ruths, and Aengus Bridgman, for allowing access to the data used in this note. The DDP surveys were funded by Heritage Canada, Rossy Foundation, the McConnell Foundation, the Lumninate Group, and the Public Policy Forum.
Competing interests declaration
The author declares no competing interests.
Open access
