Each account's politics was more closely linked to what it shared than to the source
On U.S. Twitter, accounts' estimated political orientation aligned more with what they shared than with the reliability of the sites.
The key idea
COVID-19 links showed political differences, but the moderate majority and fact-based information mattered in the overall picture.
The study, in plain language
The team analyzed more than 260 million tweets about COVID-19 posted between January 21 and July 31, 2020, and focused on 48 million tweets from more than two million users whose estimated location was the United States. They assigned each account an inferred political orientation and gave linked sites reliability ratings from an external database.
What each account shared more closely resembled its political orientation than the reliability rating of the sites. Conservative accounts in this sample had greater estimated exposure to and sharing of material classified as misinformation, but amplification was concentrated among highly partisan accounts on both sides.
Moderate accounts tended to filter out this material and share more information classified as fact-based; overall, that information received more attention. Circulation showed political differences, but a moderate majority also shaped the overall picture. One caveat: the rating applies to the site, not each claim.
What they found
- Accounts' estimated political orientation aligned more with what they shared than with the reliability of linked sites.
- Conservative accounts had greater estimated exposure to less reliable material. Highly partisan accounts on both sides amplified it.
- Moderate accounts tended to filter out less reliable material, while fact-based information received more attention overall.
How they studied it
The researchers observationally analyzed more than 260 million tweets about COVID-19 posted between January 21 and July 31, 2020, focusing on 48 million tweets from more than two million users estimated to be in the United States. They combined inferred account politics, external site-reliability categories, and retweet networks to estimate who saw and shared each link.
What the study cannot show
- The sample does not represent the entire population or all platforms.
- The analysis describes associations, not why each person shared a post.
Something to think about
A link coming from a site rated as unreliable is not enough to establish that a specific claim is false. Do you look at the site, the claim, or who shares it?
The original study
Partisan asymmetries in exposure to misinformation
Source under CC BY 4.0. English adaptation and explanation by PaperFlow, with AI assistance. The text has been reorganized for a general audience; it is not the complete paper and does not imply endorsement by its authors.
One study alone does not represent a field's consensus. Educational content does not replace professional advice. Report a possible error.
