Ideological biases in social sharing of online information about climate change

σ = 2.10%; URLs: mean 70.3%, min. 66.6%, max. 73.2%, σ = 6.54%; domains: mean 29.8%, min. 27.3%, max. 31.5%, σ = 4.21%). While web domains exhibit the most stability between weeks, URLs exhibit the least stability and users fall between these two extremes. To confirm that the persistent two-cluster network structures seen in Fig 7 are polarised along the same left-right political and environmentalist-sceptic ideological axes as that seen for Week 4 (see Fig 5), the domain bias codings were applied to the networks created for Weeks 1- 3 and 5-7 (S6 and S7 Figs in S1 File). Furthermore, average biases were calculated for all com- munities in each week in Fig 8. Trends in polarisation over time are shown as network aver- ages for political and climate change biases, alongside the community averages, in Fig 8. In most weeks, the whole-network average shows mild left-wing and environmentalist bias. Week 5 shows (for the only time across the seven-week period) a large neutral community. Supporting the visual evidence seen in S6 and S7 Figs in S1 File, and the community-level bias scores in Fig 8, these findings show that the polarised network structure observed in Week 4 is persistent and is not an artefact of the increase in activity in the climate change conversation despite the turnover in users, source domains and shared articles. Discussion This paper presents an analysis of digital media sharing behaviour around the contested issue of climate change. Our analysis looks at the network structure formed by users sharing web

articles related to climate change, combining network analysis with computational and human text analyses to identify and characterise communities of users and the articles/sources they Fig 8. Levels of political and climate change bias over the course of the seven week study period. These are measured as the mean coded bias of domains weighted by total shares (see Ideological coding of source domains). Bias in each of the five largest communities are represented by the scatter points in each week, and the bias across the whole network is given by the lines. The colour of the community points is consistent with other figures. In most weeks, the average network bias is left of centre and more environmentalist than sceptic. This trend continues to the individual communities, with the majority being left-wing and environmentalist. https://doi.org/10.1371/journal.pone.0250656.g008 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 19 / 25