AoIR 2021

The Conversation on Facebook: Patterns of Dissemination in Australia and Anglophone Canada

Axel Bruns, Michelle Riedlinger, and Jean Burgess

• 12-16 Oct. 2021 – long version of a paper presented at the Association of Internet Researchers conference, online
• presented as part of a panel on The Conversation, Ten Years On: Assessing The Impact of a Unique Scholarly Publishing Initiative

Video of the Presentation

Presentation Slides

Paper Abstract

Introduction

The Conversation’s stated aim is “to provide access to quality explanatory journalism essential for healthy democracy” (The Conversation, 2021), but the provision of access alone is insufficient to this purpose: the content also needs to engage a broad and diverse public. The extent to which Conversation articles circulate beyond the website itself, and among which kinds of communities, is therefore an important set of questions. In addition to an assessment of on-site readership patterns for The Conversation’s content, it is therefore also critical to investigate the extent to which Conversation articles circulate on leading social media platforms, and whether such circulation reaches a diverse audience or occurs predominantly amongst user communities that may already have a considerable affinity for scholarly perspectives.

This approach is especially critical since we know that a considerable component of the potential audience for news and information content no longer actively seeks out such materials, but encounters them more serendipitously via their social media networks (Newman et al., 2016: 99-101). If articles from The Conversation are to reach such larger and less active audiences, then, they will need to circulate widely on the major social media platforms in each country, and be shared by groups and individuals representing a broad range of interests and ideologies. Conversely, if they are posted only rarely, and receive only limited amplification via in-platform on-sharing (Wischnewski et al., 2021), or if such posting is limited to distinct communities of users (e.g. scholars and students), Conversation articles are likely to miss their mark.

Approach and Method

Therefore, this paper investigates the sharing patterns for content from The Conversation for its Australian and (Anglophone) Canadian editions – respectively the oldest and one of the youngest Conversation editions. We focus here especially on the circulation of Conversation content on Australian and Canadian Facebook pages, as Facebook remains the most prominent social media platform in both countries, and as available information on the geographic location of page administrators provides a useful proxy for identifying Australian and Canadian pages.

Further, we assess this circulation of Conversation articles in both countries alongside that of other mainstream news media content. Inspired by the work of Faris et al. (2017), and its adaptation and extension by Giglietto et al. (2018), who explored news engagement during the 2016 US and 2018 Italian elections, respectively, we use this broader sharing data to determine the relative affinity of Conversation content with the material published by other news organisations, and use background information about the specific audience demographics of those outlets to develop a perspective on the audiences that are likely to have encountered Conversation articles in their Facebook feeds.

For this research, we draw on the Facebook-owned data provider CrowdTangle as our data source. CrowdTangle offers data on the posting activities of public pages, public groups, and verified profiles on Facebook only; further, it classifies the country of origin of Facebook pages based on the geographic location of the majority of their administrators. While this introduces considerable limitations (our focus on public pages only means that we are unable to assess the sharing of Conversation URLs in public groups or in non-public Facebook spaces) and is likely to produce some false positives or false negatives (some pages may have outsourced their administration to teams located outside of Australia or Canada), in the absence of any other reliable access to Facebook data this approach nonetheless offers valuable insights.

Thus, for this analysis we have gathered all posts from public Facebook pages with a majority of administrators based in Australia and Canada, respectively, posted during the 2020 calendar year and containing a link to The Conversation or any of the top 40 Australian or top 70 Anglophone Canadian news domains (the smaller number of Australian news domains demonstrates the significant market concentration in Australia, which is dominated by a small number of major commercial news conglomerates). This resulted in datasets of 1,385,192 unique posts from Australian Facebook pages, containing 410,014 unique Australian news URLs (including 22,760 links to 9,048 unique Conversation URLs), and 1,262,318 unique posts from Canadian Facebook pages, containing 516,656 unique Canadian news URLs (including 3,412 links to 2,119 unique Conversation URLs).

Preliminary Analysis and Further Outlook

A work-in-progress visualisation of the Australian data at the domain level as a hybrid network containing both Facebook pages and the news domains they have linked to (fig. 1) already reveals some notable structural properties. Here, The Conversation has greater affinity with public broadcasters ABC and SBS, and a cluster of broadly progressive outlets like The Guardian (which operates a distinct Australian edition), Crikey, or The New Daily, while it appears less close to a cluster comprised of the outlets of the conservative NewsCorp stable, including The Australian, Sky News Australia, and the Daily Telegraph, and the Australian edition of UK tabloid Daily Mail.

Fig. 1: bipartite network between Facebook pages (grey) and the Australian news domains they link to (blue; Conversation highlighted in red)

In further analysis, we plan to convert this bipartite network to two monopartite networks (between domains and between pages only, respectively), in order to better assess the relative affinity between the domains being shared, and between the pages sharing them. This will also take into account available information about how these pages define themselves, through their descriptions and categorisation. We will also conduct in-depth manual and computational coding and analysis to identify the most prominent themes of the Conversation articles being shared in each country, and in the country-specific network clusters that emerge from our network analysis, as well as of the Facebook post texts accompanying these shared links, and examine the patterns of engagement with these posts on Facebook.

A first exploration of the Australian data has already revealed that of the ten most frequently posted Conversation URLs in Australian pages, eight address issues relating to climate change and the environment, one addresses Australia’s history with slavery, and only one centrally focusses on the pandemic. While devastating bushfires were a major story in Australia in the first months of 2020, this is somewhat surprising given the overwhelming attention paid to COVID-19 in the remainder of the year, and the role scientific work plays in understanding and addressing it. However, it is possible that insights relating to coronavirus are spread across a larger number of Conversation articles which in aggregate reached a large Facebook audience even if none were shared especially widely, and that our further analysis will identify these in the data. In Anglophone Canada, meanwhile, the two top articles cover the COVID-19 crisis, while others in the top ten predominantly address issues related to Indigenous discrimination and gender equality.

From our further mixed-methods analysis, we expect additional distinct patterns to emerge. These should indicate, for both country editions of The Conversation, whether and to what extent they have managed to address a broad and diverse audience of Facebook communities, and whether there remain specific demographically or ideologically defined communities who have yet to engage with The Conversation and its efforts to communicate scientific research to a wider audience. We anticipate that the Australian edition of The Conversation will have made greater inroads towards its goals over the ten years of its existence, but also that the different social, media, and political environment in Canada will have directly affected the emerging positioning of the Canadian edition in its domestic media landscape.

Acknowledgment

Data from CrowdTangle, a public insights tool owned and operated by Facebook.

References

The Conversation. (2021). Who We Are. https://theconversation.com/au/who-we-are

Faris, R., Roberts, H., Etling, B., Bourassa, N., Zuckerman, E., & Benkler, Y. (2017). Partisanship, Propaganda, and Disinformation: Online Media and the 2016 U.S. Presidential Election (Berkman Klein Center Research Publication No. 2017–6). Social Science Research Network. https://papers.ssrn.com/abstract=3019414

Giglietto, F., Ianelli, L., Rossi, L., Valeriani, A., Righetti, N., Carabini, F., Marino, G., Usai, S., & Zurovac, E. (2018). Elezioni—2018: Mapping Italian News. LaRiCA at Università di Urbino Carlo Bo. https://elezioni2018.news/blog/report/23

Newman, N., Fletcher, R., Levy, D. A. L., & Nielsen, R. K. (2016). Reuters Institute Digital News Report 2016. Reuters Institute for the Study of Journalism, University of Oxford. http://reutersinstitute.politics.ox.ac.uk/sites/default/files/Digital-News-Report-2016.pdf

Wischnewski, M., Keller, T., & Bruns, A. (2021). Shareworthiness and Motivated Reasoning in Hyper-Partisan News Sharing Behavior on Twitter. Digital Journalism. https://doi.org/10.1080/21670811.2021.1903960