Publishers/sources (Disinformation)

Recent research has mainly used two approaches to identify publishers or sources of disinformation: First, alternative media are identified as potential disinformation. Second, disinformation via fact-checking websites. Samples created using those can partly overlap. However, the differ in terms validity and comprehensiveness population. Sampling outlets is theory-driven allows for cross-national comparison. researchers face challenge misinforming content published by outlets. In contrast, fact-checked facilitates identification a given population; however, fact-checker often have publication bias focusing on small range (elite) actors (e.g. individual blogs, hyper partisan news outlets, politicians). both it important describe, compare and, if possible, assign already existing categories order enable temporal spatial Approaches sources/publishers: Besides operationalization specific variables analyzed field disinformation, sampling procedure presents crucial element operationalize itself. Following approach detecting through its (Li, 2020), analyzes (Bachl, 2018; Boberg, Quandt, Schatto-Eckrodt, & Frischlich, 2020; Heft et al., 2020) identifies various domains sites (Allcott Gentzkow, 2017; Grinberg (2019); Guess, Nyhan Reifler, 2018). Those explained following. Alternative sources/publishers The following summarizes current relevant (following Bachl, Boberg 2020). Snowball define universe may consists steps: Sample previous Consultation search engines articles Departing from prototype, websites provide information about digital metrics (Alexa.com Similarweb.com). For example, Similarweb.com shows three lists per outlet: “Top Referring Sites” (which sending traffic this site), “Also visited websites” (overlap with users other websites), “Competitors Similar (similarity defined company) Definition Journalistic (for excluding blogs forums) current, non-fictional regular Self-description so-called “about us” section mission statement, which underlines relational perspective being an mainstream media. This description example include keywords such alternative, independent, unbiased, critical line statements like “presenting real/true views/facts” “covering what hides/leaves out”. Use predefined dimensions (Frischlich, Klapproth, Brinkschulte, Holt, Ustad Figenschou, 2019) Sources/publishers U.S., Guess al. (2018) “Fake domains” (focusing pro-Trump pro-Clinton content) more that were coded “fake news” fact-checkers (derived Allcott 2017). (2019) classes sources” differentiated severity frequency false (see Table 1). These part total six website labels. additionally into reasonable journalism, low quality satire not applicable. coders reached percentual agreement 60% labeling categories, 80% distinction fake non-fake categories. 1. Three Label Specification Identification Black Based studies: at least declared sites. preexisting constructed fact-checkers, journalists academics 2018) Almost exclusively fabricated stories Red Major frequent falsehoods site's political agenda. Prejudiced: Site focus upon one group regards race / religion ethnicity sexual orientation. little regard truth, but necessarily advance certain By snopes.com questionable claims; then manually between red orange Falsehoods clearly reflected flawed editorial process Orange Moderate occasional Sensationalism: exaggerations extent article becomes misleading inaccurate. Occasionally prejudiced articles: times contain regarding orientation Openly states site be inaccurate, news, cannot trusted factual news. Conspiratorial: explanations events involves unwarranted suspicion government cover ups supernatural agents. Negligent deceptive less systemically Supplementary materials: https://science.sciencemag.org/content/sci/suppl/2019/01/23/363.6425.374.DC1/aau2706_Grinberg_SM.pdf (S5 S6) Coding scheme source labels: https://zenodo.org/record/2651401#.XxGtJJgzaUl (LazerLab-twitter-fake-news-replication-2c941b8\domains\domain_coding\data) References Allcott, H., M. (2017). Social 2016 election. Journal Economic Perspectives, 31(2), 211–236. (2018). (Alternative) AfD-centered Facebook discussions. Studies Communication | Media, 7(2), 256–270. Bakir, V., McStay, A. Fake economy emotions. Digital Journalism, 6(2), 154–175. S., T., L. (2020, April 6). Pandemic populism: pages corona crisis -- A computational analysis. Retrieved http://arxiv.org/pdf/2004.02566v3 Farkas, J., Schou, Neumayer, C. Cloaked pages: Exploring Islamist propaganda social New Media Society, 20(5), 1850–1867. L., F. (2020). Between – Co-orientation right-wing populist Grimme, Preuss, W. Takes, Waldherr (Eds.), Lecture Notes Computer Science. Disinformation open online (Vol. 12021, pp. 150–167). Cham: Springer International Publishing. Grinberg, N., Joseph, K., Friedland, Swire-Thompson, B., Lazer, D. (2019). Twitter during U.S. Presidential Science (New York, N.Y.), 363(6425), 374–378. A., Nagler, Tucker, J. Less than you think: Prevalence predictors dissemination Facebook. Advances, 5(1). https://doi.org/10.1126/sciadv.aau4586 Nyhan, Selective exposure misinformation: Evidence consumption US presidential campaign. European Research Council, 9(3), 1–14. Heft, Mayerhöffer, E., Reinhardt, Knüpfer, Beyond Breitbart: Comparing right?wing infrastructures Western democracies. Policy Internet, 12(1), 20–45. Key 7(7), 860–869. Nelson, Taneja, H. small, disloyal audience: role audience availability consumption. 20(10), 3720–3737.