Study title

Sweden

Time series: Continuous

Media unit: Text

Other

1. On 1 August 2022, we used the software httrack to crawl the CON’s blog, retrieving 2654 posts.2. We extracted 1943 hyperlinks from the retrieved blog posts.3. We identified 268 occurrences of the term “googla” on 177 different blog posts since 2014.4. We identified and tabulated all explicitly suggested keyphrases, i.e., those that follow an imperative verb and are quoted or follow a colon.5. We coded the retrieved keyphrases according to their syntactical composition. Coding was carried out by the first author and validated by the second author. 6. We created a set of 25 keyphrases to use as seeds for further data creation. The set included all ten keyphrases that had been suggested at least four times, and added 15 strategically selected keyphrases used two or three times to increase variation.7. We submitted the compiled keyphrases to the Swedish media database Retriever, yielding 240 results from Swedish print media. Of those, 204 asked readers to “google” the respective keyphrase.8. We submitted the same keyphrases as queries to Google Search and DuckDuckGo, using the search retrieval analysis software RAT (Result Assessment Tool) to obtain the first SERP for each search engine as well as the HTML source code of results (Lewandowski et al., 2022; data available via Sünkler et al., 2023). We submitted (a) the suggested queries; (b) the suggested queries in quotation marks (i.e., verbatim search); and (c) the Swedish imperative form of “google” followed by the suggested keyphrase (no quotation marks). With a maximum of 10 results per query, but often fewer and sometimes no results for queries b and c, we obtained 146 SERPs and 1001 search results.9. Of these, 249 results link to the CON’s blog, and further 236 results mention the CON or its authors—usually signed by the CON or linking to it. Few referred to the blog as engaged in climate obstruction (e.g. by debunking the CON’s claims); conversely, not all climate obstruction content in the data set mentions the CON.10. Based on search results and hyperlinks, we classified 204 unique domains as frequent, i.e. they occurred in at least two SERPs, at least 10 hyperlinks, or at least 1 SERP and 4 hyperlinks. As these counts represent the possibility of finding a specific domain, we included duplicate targets in these counts. We coded these frequent domains regarding their site type and language. Coding was carried out by the first author and validated by the second author.For more details, see the included README file.

Content coding