Amplifying division: Electoral misinformation and political intolerance in Brazil

Summary

This paper investigates how belief in electoral misinformation shaped political intolerance during Brazil’s polarized 2022 presidential election. Drawing on a three-wave panel survey timed around the runoff and the subsequent anti-democratic protests, Rossini, Kalogeropoulos, and Mont’Alverne show that Brazilians who accepted false claims about electoral fraud became measurably less willing to extend civil rights to disliked political groups. They further argue that messaging apps — particularly WhatsApp — operate as an indirect pathway to intolerance by elevating misinformation beliefs, distinguishing them from general social media use. The authors reframe electoral misinformation not as a problem of confused citizens but as a corrosive force on democratic norms themselves, one that amplifies pre-existing cleavages like Brazilian antipetismo.

Key Contributions

• First empirical link between electoral misinformation belief and political (in)tolerance, extending the standard focus on knowledge and institutional trust.
• Differentiates messaging apps from broader social media: the former indirectly fuel intolerance through misinformation, the latter do not.
• A rare longitudinal panel design in a major non-Western democracy during an actively contested election.
• Reframes the democratic stakes of misinformation toward tolerance erosion and democratic backsliding.
• Argues fact-checking is necessary but insufficient as a response.

Methods

A three-wave nationally representative panel of Brazilian internet users (W1 N=1600 post first round; W2 N=1328 post second round; W3 N=1034 during anti-democratic protests), fielded by Ipec with demographic quotas. Political intolerance was measured via a GSS-style “least-liked group” approach across rights (voting, protesting, civil service, public speech). Misinformation belief was measured in W2 only (for ethical reasons) using eight fact-checked false statements with true and placebo controls. Analysis used autoregressive OLS models predicting W2 and W3 intolerance controlling for prior intolerance, plus structural equation models (lavaan, bootstrapped CIs) to test mediation from messaging/social media news use through misinformation belief to intolerance.

Findings

• Belief in electoral misinformation predicted lower tolerance in W2 (β = −.09, p<.001) and W3 (β = −.10, p<.01).
• Messaging app news use had no direct effect on intolerance, but an indirect effect via misinformation belief (β = −.009, p<.05).
• Social media news use had no comparable indirect effect; if anything, it was marginally associated with greater tolerance in W3.
• Engagement with online disagreement was associated with decreased tolerance between W1 and W2, contrary to classical deliberative expectations.
• 46% of respondents endorsed zero false claims; the mean was 1.58 out of 8, suggesting concentrated rather than universal exposure.

Connections

This paper sits naturally alongside other work on Brazilian information environments and election-period misinformation, particularly Cazzamatta2026-lo, Ventura2026-yc, Ventura2025-sw, and Kansaon2025-id on Brazil-specific dynamics, and Emilio2026-ik on WhatsApp-mediated politics. Its turn from knowledge effects to attitudinal/democratic-norm effects connects to Hameleers2026-mc, van-der-Linden2026-jt, and Humprecht2025-ml, while its finding that online disagreement can erode tolerance complements Mosleh2024-op and the platform-effects debate in Gonzalez-Bailon2024-rq and Budak2024-ef.

Podcast

A research-radio episode discusses this paper: Listen