How the Mainstream Media Help to Spread Disinformation about Covid-19

Introduction In this article, we hypothesise how mainstream media coverage can promote the spread of disinformation about Covid-19. Mainstream are often discussed as opposed to (Glasser; Benkler et al.). While phenomenon is related intentional production and misleading false information influence public opinion (Fallis; al.), news expected be based on facts investigation focussed values such authenticity, accountability, autonomy (Hayes However, journalists might contribute when they skip some stage processing reproduce or (Himma-Kadakas). Besides, even purpose correct disinformation, its dissemination by amplifying it (Tsfati This could particularly problematic in context social media, users just read headlines while scrolling through their timelines (Newman al.; Ofcom). Thus, share from legitimate The pandemic creates a delicate context, pressured produce more and, therefore, susceptible errors. research, hypothesis that reinforce discourses, though actual piece may frame story differently. research questions guide are: URLs with discourses other links shared into same Facebook groups? Are support narratives? As case study, look at Brazilian discussion disease country has been highly polarised politically framed, government agents scientists disputing truth (Araújo Oliveira; Recuero Soares; Particularly, ecosystem seems play an important role these disputes, President Jair Bolsonaro his supporters use key channel virus (Lisboa Soares We data groups collected CrowdTangle combination network analysis content analyse posts. Theoretical Background Disinformation central Covid-19 “infodemic”, created overabundance pandemic, which makes hard for people find reliable guidance exacerbates outbreak (Tangcharoensathien consider distorted, manipulated, intentionally mislead someone used strengthen radical political ideologies (Benkler Around world, actors framed debate (Allcott al., Gruzd Mai; Soares). On contexts polarisation between two different views present narratives fact dispute attention (Soares suitable environment thrive al.) discussions associated idea “bubbles”, tend only aligned group's ideological views. Consequently, turn bubbles (Pariser). cases, within one group not vice versa. Pariser argues exposed exclusively agree. shown Pariser’s concept limitations (Bruns), most variety sources (Guess Nevertheless, lead diets consumption That is, would have contact types information, but choose certain over others because alignment (Bruns). Therefore, understand action who give preference circulate (through retweets, likes, comments, shares) supports views, including (Recuero ephemeral structures (created users’ actions particular discussion) permeable boundaries (users outside) media. type bubble tool create unique discourse does mean “disinformation bubble” do access content, It means acts discredit overlap “alternative” (Larsson). addition, disseminate inaccurate focus narratives, especially large number exposure websites heavily concentrated few Tsfati likely larger will bubbles. Based discussion, aim Methods study Facebook. propose hypotheses, follows: H1: When way reinforces narrative, go bubble”. H2: narratives. selected three studies events both high Brazil. chose them part produced was high. 24 March 2020, made pronouncement live television. week before pronouncement, governors decided follow World Health Organisation (WHO) protocols closed non-essential business. speech, criticised distancing measures. reproduced claims personalities, entrepreneurs also said harm economy. 8 June WHO official “seems rare asymptomatic person transmits [Covid-19] onward secondary individual”. Part claim out misperception pre-symptomatic persons (early stages illness, first symptoms) transmit all. 9 November Brazil’s national sanitary watchdog Anvisa reported had halted clinical CoronaVac vaccine, developed Chinese company Sinovac. being partnership São Paulo’s Butantan Institute became subject Governor Paulo, João Dória. halt trial "another victory Bolsonaro". trail after "severe adverse event". rapidly reverberated decision. Later, revealed incident death nothing vaccine. Before our final dataset includes together, explored each event. keywords collect posts monitored CrowdTangle, owned tracks publicly available platform. timeframe days event prevent collection unrelated cases. containing URLs. Table 1 summarises collected. 1: Data Dates 24-26 2020 8-10 9-11 Keywords “Covid-19” “coronavirus” “isolation” “economy” “asymptomatic” “vaccine” “Anvisa” “CoronaVac” Number 4780 2060 3273 From original dataset, 60 period (n=180). then filtered those were outlets (n=74). (Krippendorff) observe (two independent coders, Krippendorff’s Alpha = 0.76). Facebook, headline appears users. considered headlined reinforced flagged coders (n=21 – examples provided 3 Results section). 2 provides breakdown analysis. 2: Content Event Headlines Economy quarantine 7 112 Asymptomatic 22 163 Vaccine 28 120 Total 74 21 395 supported low (n=395), conducted another search dataset. sample classified “balanced” Out 10 (this time, without any timeframe) (n=1346 posts). “control group” neither (n=1416 identify comprises 20 2762 (Wasserman Faust) map links. bipartite network, nodes (1) (2) URLs; edges represent post URL applied modularity metric (Blondel clusters. allows us “communities” similar map. helped if there “bubble” shares (H1). To supporting narrative groups, alignments cluster (H2). Textometrica (Lindgreen Palm) word clouds frequent words names (at least five mentions) connections. Finally, analysed disinformation. 1346 posts, 373 included message (the 973 link). see 0.723). There disagreements categorisation 27 reviewed classification reach agreement. Bubbles graph (Figure 1), red blue links, black groups. Our finding rarely 1623 174 (10.7%) discourse, link; 712 (43.8%) disinformation; 739 (45.5%) information. Figure Network confirmed tendency “bubbles” 2). purple seven discourse. green result partially identified, went separate “bubble”, did shows boost creation (Bakshy other. source all analysed. discourses. framing discussions, explore below. Political Discourse groups’ affiliation strongly 3). word. Other prevalent Brazil, patriots (both nationalist discourse), right-wing, conservative, military (three conservative dictatorship ruled Brazil 1964 1985), President, support, Alliance [for Brazil] name party). Some active “Alliance Brazil”, “Bolsonaro 2022 [next presidential election]”, “Bolsonaro’s nation 2022”, “I am right-wing pride”. 3: Purple cloud 4). connected “against” “out”, many anti-Bolsonaro. Furthermore, left-wing, Workers’ Party (centre-left party), Lula Dilma Rousseff ex-presidents) show general. local (related locations Rio de Janeiro, Grande Sul, Minas Gerais, others), (news, newspaper, radio, portal). “We 70 per cent [anti-Bolsonaro movement]”, “Union Left”, “Lula president”, “Anti-Bolsonaro” cluster. 4: Green Then, total, found 81.8% messages frequency higher (86.2%) considering (based metric). lower (64%) showed Bolsonaro; health authorities WHO, Paulo Dória, China, “leftists” (all opposition Bolsonaro); claimed measures unnecessary; vaccines dangerous. provide (we most-shared illustrate). H2 narrative; bubble. Examples Headline Post "Unemployment crisis much worse than coronavirus", says Go President. Unemployment kills. More virus... hunger, depression, despair everything UNEMPLOYMENT, THE DEPUTIES CHAMBER, SENATE AND SUPREME COURT KILL MORE THAN COVID19 patients coronavirus, QUARANTINE IS FAKE #StayHome, lie century! THIS GOES TO PUPPETS OF COMMUNIST PARTIES FUNERARY MEDIA halts Coronavac vaccine "serious event" [The event] serious, so killed covid And Doria [Governor adversary Bolsonaro] wants force you take shit making potential (Himma-Kadakas; particular, credibility opposing end up fuelling infodemic sharing reverberating actors. Conclusion looked compared Two guided study: identified bubble, pro-Bolsonaro alignment. contributed ought extra care producing news, headlines, visible stories limitations. (n=20) outcomes. collection. tool. entire References Allcott, Hunt, al. “Polarization Public Health: Partisan Differences Social Distancing during Coronavirus Pandemic.” National Bureau Economic Research, Working Paper No. 26946 (2020). 6 Jan. 2021 <https://doi.org/10.3386/w26946>. Araújo, Ronaldo Ferreira, Thaiane Moreira Oliveira. “Desinformação e Mensagens Sobre Hidroxicloroquina no Twitter: Da Pressão Política à Disputa Científica.” Atoz Novas Práticas em Informação Conhecimento 9.2 <http://dx.doi.org/10.5380/atoz.v9i2.75929>. Bakshy, Eytan, “Exposure Ideologically Diverse News Opinion Facebook.” Science 348.6239 (2015). <https://science.sciencemag.org/content/348/6239/1130>. Benkler, Yochai, Propaganda: Manipulation, Disinformation, Radicalization American Politics. New York: Oxford University Press, 2018. Blondel, Vincent D., “Fast Unfolding Communities Large Networks.” Physics.soc-ph (2008). <http://lanl.arxiv.org/abs/0803.0476>. Bruns, Axel. Filter Real?. Cambridge: Polity 2019. Team. CrowdTangle. Menlo Park, Calif.: 2020. <https://apps.crowdtangle.com/search/>. Fallis, Don. “What Is Disinformation?” Library Trends 63.3 (2015): 401-426. Glasser, Susan B. “Covering Politics ‘Post-Truth’ America.” Brookings Institution Dec. 2016. Feb. <https://www.brookings.edu/essay/covering-politics-in-a-post-truth-america/>. Gruzd, Anatoliy, Philip Mai. “Going Viral: How Single Tweet Spawned COVID-19 Conspiracy Theory Twitter.” Big & Society, 7.2 <https://doi.org/10.1177/2053951720938405>. Guess, Andrew, Avoiding Echo Chamber Chambers: Why Selective Exposure Like-Minded Less Prevalent You Think. Miami: John S. James L. Knight Foundation, Hayes, Arthur S., “Shifting Roles, Enduring Values: Credible Journalist Digital Age.” Journal Mass Media Ethics 22.4 (2007): 262-279. Feb.2021 <https://doi.org/10.1080/08900520701583545>. Himma-Kadakas, Marju. “Alternative Facts Fake Entering Journalistic Production Cycle”. Cosmopolitan Civil Societies: An Interdisciplinary (2017). <https://doi.org/10.5130/ccs.v9i2.5469>. Kripendorff, Klaus. Analysis: Its Methodology. California: Sage Publications, 2013. Larsson, Anders Olof. “News Use Amplification Norwegian National, Regional Hyperpartisan Journalism Communication Quarterly 96 (2019). <https://doi.org/10.1177/1077699019831439>. Lindgreen, Simon, Fredrik Palm. Service Package (2011). <http://textometrica.humlab.umu.se>. Lisboa, Lucas A., “A Disseminação da Desinformação Promovida por Líderes Estatais na Pandemia COVID-19.” Proceedings Workshop Implicações Computação Sociedade (WICS), Porto Alegre: Brasileira Computação, <https://doi.org/10.5753/wics.2020.11042>. Newman, Nic, Reuters Report Oxford: University, Ofcom. “Scrolling News: Changing Face Online Consumption.” 23 <https://www.ofcom.org.uk/__data/assets/pdf_file/0022/115915/Scrolling-News.pdf>. Pariser, Eli. Bubble. Penguin, 2011. Recuero, Raquel, Felipe Soares. “O Discurso Desinformativo sobre Cura Estudo Caso.” E-Compós <https://doi.org/10.30962/ec.2127>. “Polarization, Hyperpartisanship, Circulates Contracampo (2021, press). <https://doi.org/10.1590/SciELOPreprints.1154>. Soares, Bonow, “Disputas discursivas desinformação Instagram o uso hidroxicloroquina como tratamento para Covid-19.” 43º Congresso Brasileiro Ciências Comunicação, Salvador: Intercom, <http://www.intercom.org.br/sis/eventos/2020/resumos/R15-0550-1.pdf>. Tangcharoensathien, Viroj, “Framework Managing Infodemic: Crowdsourced Technical Consultation.” J Med Internet Res 22.6 <https://doi.org/10.2196/19659>. Tsfati, Yariv, “Causes Consequences Dissemination Literature Review Synthesis.” Annals International Association 44.2 (2020): 157-173. <https://doi.org/10.1080/23808985.2020.1759443>. Wasserman, Stanley, Katherine Faust. Applications. Cambridge UP, 1994.