Skip to main content

University of Michigan School of Information


Data Science/Computational Social Science Seminar: Mohsen Mosleh

“DS/CSS. Data Science/Computational Social Science Seminar Series. Mohsen Mosleh. University of Exeter Business School, UK. Connection and correction on social media. Thursday, September 22. Noon - 1 pm EDT. Online. Registration required.”

Noon - 1:00 p.m.

Connection and correction on social media

Register to attend DS/CSS online events at

In this talk, I will explore the dynamics of social tie formation and debunking of false content on social media. There is a great deal of observational evidence of homophily on social media, where users are more likely to be connected to like-minded others who share similar worldviews (contributing to concerns about “echo chambers”). However, this observational evidence does not offer credible evidence of a causal effect of shared partisanship on tie formation. I will begin by presenting a field experiment where we made human-looking bots that identified as Democrats or Republicans and then followed partisans on Twitter. We find that users were roughly three times more likely to reciprocally follow-back our experimental human-looking bot accounts whose partisanship matched their own (Mosleh et al 2021 PNAS). In a follow-up field experiment, we also find that users are much more likely to block counter-partisans who follow them. I will then shift gears to explore how these connection dynamics interact with the debunking of misinformation. In a Twitter field experiment, we have human-looking bots deliver fact-checks to users who had shared false news. We find that shared partisanship does not moderate engagement with the fact-checks, but that forming a minimal social connection by the bot following the user and liking some of their recent tweets increases engagement with fact-checks (particularly with co-partisans) (Mosleh et al 2022 PsyArxiv). Finally, I will explore the downstream effects of the social fact-checks delivered by our bots, finding that being corrected led users to significantly reduce the quality of shared news they subsequently shared (Mosleh et al 2021 CHI).

Mohsen Mosleh

Speaker bio:
Mohsen Mosleh is a senior lecturer (assistant professor) at University of Exeter Business School, a fellow at Alan Turing Institute, and a research affiliate at MIT. Mosleh was a postdoctoral fellow in cognitive science at the MIT Sloan School of Management and the Department of Psychology at Yale University. Mosleh's research interests lie at the intersection of data science and cognitive science. In particular, he studies how information and misinformation spread on social media.