Misinformation beyond traditional feeds: Evidence from a WhatsApp deactivation experiment in Brazil

Summary

This paper reports a field experiment in Brazil that uses a WhatsApp deactivation design to causally estimate how the encrypted messaging platform shapes users’ political knowledge, beliefs, and exposure to misinformation during an electoral period. The authors argue that WhatsApp — as a private, non-feed-based platform central to Brazilian political communication — plays a distinct role in the information ecosystem that cannot be inferred from existing research on Facebook or Twitter. By inducing a subset of users to stop using the app and comparing them to a control group, the study extends platform-effects research into both a Global South context and a class of platforms that has been largely overlooked.

Key Contributions

• Provides experimental, causal evidence on the political effects of an encrypted messaging app, a category previously underexamined in misinformation research.
• Expands the empirical scope of misinformation studies beyond public, feed-based platforms to private messaging environments.
• Delivers evidence from Brazil, a critical non-U.S. case where WhatsApp is a central conduit of political communication.
• Demonstrates that deactivation designs — previously applied to Facebook — can be adapted to messaging platforms.

Methods

A field experiment conducted in Brazil around an electoral period. Participants were randomly assigned to deactivate WhatsApp for a defined period, while a control group continued normal use. Outcomes were measured through surveys capturing political knowledge, beliefs, attitudes, and self-reported or behaviorally inferred exposure to misinformation, then compared across treated and control conditions.

Findings

• Deactivating WhatsApp produced measurable changes in users’ information environments and political attitudes relative to control users.
• The effects illustrate that private messaging platforms are consequential sites for misinformation exposure, not just supplementary channels to public feeds.
• Specific effect magnitudes are not available from the abstract-level summary.

Connections

This study sits alongside other deactivation and platform-effects experiments such as DeVerna2025-dl, and complements work that decenters U.S. feed-based platforms in misinformation research — including studies of Brazilian information environments like Rossini2026-jn, Emilio2026-ik, and Kansaon2025-id. More broadly, it speaks to debates about how exposure to misinformation on social and messaging platforms shapes political beliefs, as in Gonzalez-Bailon2024-rq and Budak2024-ef.