Ideological biases in social sharing of online information about climate change

from which content was shared show higher similarity between weeks in Table 3c. User popu- lations show limited persistence between weeks, summarised by Table 3a. The overall pattern is that, from week to week, a limited proportion of the user population and set of source domains persist, with much lower persistence in the sets of articles that are shared. This is an interesting finding with respect to the high consistency in the two-cluster network structure that is reliably observed every week. The lack of persistence in shared URLs may be explained by the general volatility of news media, where articles typically have a short lifetime (e.g. 2-3 days visibility in online sharing [13]). Moderate persistence of users and sources between weeks perhaps suggests an active core group who are present each week, with a wider group who appear less frequently. A marked difference can be observed in the typical similarity scores for Week 4 and for other weeks. Lower similarity scores were observed for users, URLs and domains from Week 4 appearing in other weeks, while conversely, higher similarity was observed for users, URLs and domains from the other six weeks appearing in Week 4. The number of new users, URLs and domains in Week 4 also shows a stark contrast with other weeks: 70.1% of users, 82.0% of URLs and 47.9% of domains are unique to Week 4 and not present in any other week. This is strong evidence that the events of Week 4 appear to have spurred an influx of both social media participants and digital media sources (users: mean 49.3%, min. 45.6%, max. 51.2%,

Table 3. Similarity scores for users, URLs and domains between each of the seven weeks calculated using Eq (3). Similarity is directional. The similarity given in cell (4, 1) is the proportion of the users/URLs/domains in Week 4 also in seen Week 1. Cool shades indicate values smaller than the mean while warm shades indicate values greater than the mean. Week 1 2 3 4 5 6 7 1 0.29 0.33 0.39 0.25 0.25 0.22 2 0.24 0.30 0.35 0.22 0.24 0.21 min.(S) = 0.084 3 0.19 0.21 0.38 0.20 0.19 0.18 mean.(S) = 0.236 4 0.08 0.09 0.15 0.11 0.10 0.09 max.(S) = 0.389 5 0.18 0.20 0.25 0.36 0.24 0.21 stdev.(S) = 0.078 6 0.19 0.22 0.26 0.35 0.26 0.25 7 0.20 0.23 0.28 0.36 0.27 0.30 (a) Asymmetric overlap of users. Week 1 2 3 4 5 6 7 1 0.13 0.08 0.11 0.06 0.05 0.04 2 0.10 0.14 0.10 0.05 0.05 0.04 min.(S) = 0.022 3 0.05 0.10 0.16 0.05 0.04 0.03 mean.(S) = 0.078 4 0.02 0.03 0.06 0.06 0.03 0.02 max.(S) = 0.193 5 0.04 0.05 0.06 0.19 0.16 0.07 stdev.(S) = 0.047 6 0.04 0.05 0.06 0.10 0.17 0.16 7 0.04 0.04 0.05 0.09 0.09 0.19 (b) Asymmetric overlap of URLs. Week 1 2 3 4 5 6 7 1 0.55 0.59 0.72 0.58 0.54 0.51 2 0.44 0.58 0.75 0.48 0.55 0.48 min.(S) = 0.155 3 0.34 0.41 0.76 0.42 0.42 0.39 mean.(S) = 0.496 4 0.15 0.20 0.29 0.23 0.22 0.18 max.(S) = 0.767 5 0.41 0.43 0.53 0.76 0.52 0.46 stdev.(S) = 0.158 6 0.41 0.52 0.56 0.76 0.56 0.53 7 0.46 0.54 0.62 0.77 0.59 0.63 (c) Asymmetric overlap of domains. https://doi.org/10.1371/journal.pone.0250656.t003

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 17 / 25