The work, conducted jointly with Arnout van de Rijt (European University Institute) and Jonas Stein (University of Groningen), includes a series of experimental studies published openly in Scientific Reports and PNAS Nexus.
Structural interventions
One paper focuses on structural interventions—changes to the architecture of online networks. In large-scale online experiments with U.S. participants, the researchers created independent digital ecosystems in which users could share true and false messages about society, science, and politics. By altering the network composition—from integrated (50:50 liberals and conservatives) to segregated echo chambers—they demonstrated that partisan sorting systematically undermines information accuracy, whereas increasing ideological diversity slows the spread of misinformation. Agent-based simulations confirmed that this result generalizes across network topologies and larger populations.
Individual-level interventions
The second study examines individual-level interventions, asking why people misjudge politicized information: are they being insincere, or do they genuinely believe false claims? In experiments where participants rated the accuracy of unfamiliar statements—with and without financial incentives for accuracy—the researchers found that ideological biases persisted even under strong motivation to be correct. Many participants sincerely believed misinformation to be true, suggesting that accountability-based policies alone are unlikely to solve the problem.
Highlighted
Taken together, the findings highlight that the misinformation challenge is not just behavioral but structural. As Keuschnigg argues,
“The fight against misinformation is more likely won through interventions that mix networks and reshape newsfeeds rather than those that merely target well-intended individuals.”
