Ideological biases in social sharing of online information about climate change

longevity of reproduced messages [6]. Recent work has shown that media effects vary depend- ing on existing political biases [7]. Understanding the media landscape around climate change is of key importance in mapping public engagement with the issue and support for political actions to confront it. Assessing which people are exposed to what information is fundamental to any study of the effects of media on public understanding and opinion. The disruptive impact of online media has transformed the media environment, radically altering the diversity of content people encounter as well as the exposure process itself. Individuals are faced with a wide range of media options (both social and traditional), new patterns of exposure (selected by the end user or driven by their social network) and increased production of user-generated content [8]. Previous work in this area has focused on the effects of incidental exposure on media aware- ness (e.g. [9]) and the diversity presented by online recommender systems (e.g. [10]). Such efforts are hampered by the diversity of online platforms, the rapid pace of their development and the obfuscation of the algorithmic processes they follow, and as a result no universal understanding of exposure effects is possible. Whether an individual is consuming the news online from a legacy media organisation, or producing and consuming information on social media, the fundamental dynamic of communication exposure and influence is that of network formation [11], based on creation of new relationships between users and media content by a

variety of means (such as web-browsing and social information-sharing). Online media expo- sure creates a network that links sources and consumers of content (nodes) via their interac- tions (edges), requiring a network perspective for its proper understanding. This study aims to describe patterns of sharing online media content about climate change. While a complete record of users’ exposure to digital content is only possible via accurate tracking of web browsing histories, media engagement can be inferred from the content users share on social media. Sharing a web article requires action by the user, and causes it to appear on the social media feeds of friends and followers, as well as contributing to aggregate trends, often advertised by social media platforms. This is used to indicate a significantly higher level of engagement than exposure. Social sharing of content instantiates a promotion mechanism, increasing the visibility of any content across a user’s social network and likely indicating that the sharer agrees with, or approves of, the content. These factors, along with the increasing use of social media as a source of news [12], mean that study of digital media sharing can provide insights into how information is propagated online, including important contemporary issues like climate change. In particular, Weaver et al. [13] shows that network analysis of such prop- agation patterns can reveal meaningful social structures of news engagement and consumption around political events.

Here we operationalise our study of online information-sharing around climate change by examining link-sharing on Twitter. User posts (tweets) referencing climate change and con- taining links to web content (URLs, which are often rendered into news blurbs or images in the Twitter client) were collected from the Twitter social media platform via its public API. This dataset was used to construct bipartite networks linking users to the digital media they shared. Analysis of network topology finds strong community structure, supplemented by comparison of source domains and textual content of articles shared within each community. Exploration of detected communities identifies strong ideological polarisation within the news-sharing network where users are segregated by divergent opinions, rather than an ongo- ing process, an alternative definition which is important in other contexts [14]. Overall the results indicate highly polarised and politicised engagement with web content around climate change, with largely segregated and ideologically biased communities receiving information from different media sources. We demonstrate that the observed correlation between political views and climate change beliefs (e.g. [3]) extends to online information sources and their PLOS ONE Ideological biases in social sharing of online information about climate change PLOS ONE | https://doi.org/10.1371/journal.pone.0250656 April 23, 2021 2 / 25 Funding: TC is funded by an EPSRC Research Studentship (grant number EP/M506527/1, https:// epsrc.ukri.org/). IW and HW acknowledge funding

from ESRC (grant number ES/N012283/1, https:// esrc.ukri.org/) and NERC (grant number NE/ P017436/1, https://nerc.ukri.org/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist.