In press at @apsrjournal: "Partisan polarization is the primary psychological motivation behind political fake news sharing on Twitter" (psyarxiv.com/v45bk). We find that fake news sharers are not less reflected or literate. They just hate the other party more. 🧵[1/13]
Method: We received permission from 2,300 survey users collected by YouGov to scrape their Twitter accounts. We matched their tweets against lists of "fake news" and "real news" web domains and categorized them according to their political slant. [2/13]
How much is shared? Consistent with prior work, fake news seems a small problem. 3 % of tweets link to a fake news domain & only 1 % of the panelists share 75 % of all fake news. Most fake news stories are pro-Republican. Most real news, in contrast, are pro-Democrats. [3/13]
Who shares? We examined 3 accounts: (1) Ignorance (e.g., reflection, knowledge and digital literacy), (2) disruption (e.g., trolling) and (3) polarization (e.g., hatred of the other party). Measures are coded such that *positive* coeffs are consistent with the account. [4/13]
Results: No support for the role of ignorance in these data; some support for disruption; & strong support for polarized sentiments. BUT: (1) The same factors are associated with *real* news sharing. (2) Outparty hatred is better predictor for sharing of pro-Rep. sources. [5/13]
How to explain the symmetry in fake & real news sharing? For partisans, fake news is simply another (but extreme) source of "useful" info for denigration. Thus, partisanship is a stronger predictor when moving towards these extreme ends of the partisan news spectrum. [6/13]
Similarly, there is a strong overlap in those sharing links to fake news sources and those sharing linkes to strongly partisan but real news sources. [7/13]
How to explain the asymmetry between Dems and Reps in fake news sharing? To this end, we analyzed all shared news headlines for sentiment and whether they were about Dems or Reps. We replicated this for all 500,000 news stories published on the websites in the same period. [8/13]
Our analysis suggets that, for Republicans, fake news sources are simply much more *useful* then for Democrats. Dems can find negative news stories about Reps in mainstream media. To find very negative news about Dems, Reps need to go beyond the mainstream. [10/13]
As discussed, this doesn't necessarily imply that there is strong media bias. It could reflect a "bias in reality". But it does suggest that Dems and Reps are not psychologically different & cares equally about truth (which, however, may not be alot for strong partisans). [11/13]
Overall, this shows that "fake news" is partisan business-as-usual: A search for info to denigrate opponents. While a small problem in itself, "fake news" is thus a canary-in-the-coalmine: Revealing the massively polarized sentiments that currently drives all news sharing [12/13]
An urgent sense of crisis made people disregard fear and trust and say, "tell us what to do & we'll do it", leading to history's largest behavioral change.
From March to May, we surveyed more than 26,000 individuals across 8 countries: 🇺🇸🇩🇰🇮🇹🇬🇧🇭🇺🇸🇪🇩🇪🇫🇷. We found high levels of especially avoidant (distancing) but also preventive (handwashing) across all countries, independently of covid-cases and policies. [3/8]
En ny fase mod #covid19dk venter med genåbningen. Det er afgørende, at vi der fastholder læren fra 2020: Regeringen kan & bør have tillid til borgerne. Men der er tegn på et strategiskifte. Læs min kronik i @berlingske (berlingske.dk/kronikker/prof…) & denne 🧵[1/9]
Natten til 11. marts 2020 delte @Statsmin en artikel, der argumenterede for frivillig adfærd og kommunikation som strategi mod #covid19dk (thelancet.com/journals/lance…). HOPE-projektet har vist, at den strategi virkede ekstremt effektivt i DK (psyarxiv.com/uzwgf/). [2/9]
Fokus på frivillighed er vigtigt, da pandemien er lang og indgribende. Krisen har radikaliseringspotentiale (psyarxiv.com/ykupt/), & ca. 1 mio. danskere er bekymrede for deres rettigheder jf. HOPE-data. Den andel er lav ift. andre lande, måske pga fokus på frivillighed. [3/9]
Overordnet set støtter en høj andel af befolkningen håndteringen (ca. 70 %). Dette er det højest målte påtværs af de lande vi følger. Der ses dog et lille fald siden feb., og særligt har der de seneste måneder været fald i opbakningen til lukning af uddannelser. [2/11]
Optimismen stiger og bekymringen falder. Bekymringen er nu igen primært rettet mod landets økonomi. [3/11]
Der er særdeles god dokumentation for at splid på Christiansborg skaber splid i befolkningen (doi.org/10.1111/ajps.1…). Studier fra USA viser, at splid gør, at kritiske vælgergrupper i mindre grad følger sundhedsrådene og forværrer epidemien (fx sciencedirect.com/science/articl…) [2/11]
Også i DK er der splid mellem vælgergrupperne, når det kommer til corona - og den splid er blevet større efterhånden, som krisen er skredet frem (raw.githubusercontent.com/mariefly/HOPE/…). [3/11]
Adfærden er på niveau fra før januar-restriktionerne, og opmærksomheden er faldende. Det kan afspejle ferie og godt vejr, så man kunne mødes i det fri. Udfordringen er dog, at modellerne antager konstant adfærd - og adfærden har stor betydning (sum.dk/Media/6/3/bila…). [2/5]
At adfærden ikke blot handler om ferie afspejles også af vedvarende ændringer i de psykologiske faktorer bag adfærden. I takt med de faldende smittetal er bekymringen faldet støt. Bekymringen er nu på niveau med slut-november, lige inde smitten accelererede. [3/5]
Tillidskurven er knækket, og adfærden påvirkes nu negativt. Vi står i en kritisk fase af epidemien. Regeringen kan ikke "købe ro" med små lunser genåbning. Der skal skabes mening igennem en langsigtet strategi.
Genåbningsdebatten raser og måske kunne lidt "lunser" af genåbning "skabe ro"? Men det er en fejlanalyse. I foråret øgede genåbningerne polariseringen (raw.githubusercontent.com/mariefly/HOPE/…), og genåbningen af de mindste klasser skabte blot yderligere krav om genåbning. [3/7]