Beliefs about COVID-19 in Canada, the U.K., and the U.S.A.: A novel test of political polarization and motivated reasoning

Contributions

• Demonstrates that COVID-19 belief polarization is geographically contingent on the strength of partisan cues in the media environment, not inherent to the issue itself
• Shows cognitive sophistication negatively predicts misperceptions but does not reduce political polarization
• Provides evidence that media trust (conservative vs. liberal news outlets) is a source of belief divergence in polarized environments
• Tests identity-protective cognition theory: people reason accurately when conclusions support their identity, but use reasoning to justify ideology when needed
• Cross-country design permits causal inference about the role of political environment vs. generic ideology in shaping COVID-19 beliefs

Method

Two studies with quota-sampled surveys during the COVID-19 pandemic (Study 1: March 24, 2020; Study 2: December 9–15, 2020).

Study 1 (early pandemic): N=1,339 (Canada 644, UK 642, USA 689), measuring political ideology, COVID-19 risk perceptions, misperceptions about COVID-19 (k=21 falsehoods), behavioral intentions, cognitive sophistication (science knowledge, cognitive reflection, numeracy, bullshit receptivity), and trust in news sources.

Study 2 (later pandemic): N=1,338 (UK 641, USA 697), with updated measures including vaccination intentions and mitigation behaviors, using the same cognitive sophistication scales (shortened science knowledge to 8 items).

Cognitive sophistication operationalized as an omnibus measure combining: science knowledge (true/false test), Cognitive Reflection Test (CRT), numeracy (Berlin and Lipkus tests), and receptivity to bullshit (pseudo-profound statements). Treated as four equally weighted subscales to avoid arbitrary aggregation.

Results

Political ideology and polarization: Political conservatism strongly correlated with COVID-19 misperceptions in all three countries (Study 1: r=.31 USA, .14 UK, .27 Canada). Polarization was markedly stronger in the U.S. than elsewhere. By Study 2 (December), polarization of beliefs increased dramatically in the U.S. relative to UK, driven by greater partisan engagement.

Cognitive sophistication: Overall cognitive sophistication was a strong negative predictor of misperceptions across all countries (r's=−.34 to −.46), with effect sizes larger than ideology in effect-size comparisons. However, cognitive sophistication showed no reliable interaction with ideology in predicting beliefs — it did not buffer against polarization or reduce the association between ideology and misperceptions.

Media trust and polarization: Trust in conservative media correlated with higher misperceptions and lower COVID-19 risk perceptions. In the U.S., when ideology and media trust are both in the model, liberal media trust (not ideology) becomes the significant predictor, suggesting that differential exposure to conflicting information environments drives polarization rather than ideology per se.

Identity-protective cognition: The interaction between ideology and cognitive sophistication was non-significant across outcomes in all countries, contradicting the identity-protective cognition prediction that sophisticated reasoners would use their reasoning to justify ideologically motivated beliefs. Instead, cognitive sophistication benefits all ideologies equally.

Notes

Strengths: The cross-country design is the paper's key methodological contribution — by holding the crisis constant (COVID-19) while varying the political environment (polarized USA vs. less polarized UK/Canada), the authors isolate the effect of media/partisan environment on belief polarization. The longitudinal comparison (March to December) captures the growing polarization over time. The cognitive sophistication findings are especially important: they refute the pop-psychology claim that "smarter people believe more misinformation" while supporting the finding that reasoning ability facilitates accurate belief formation when ideological incentives are neutral.

Limitations: The cross-sectional design within each country prevents causal inference about the direction of media trust and beliefs. The political ideology measure in Canada did not directly assess party membership and thus may be a weak proxy for partisan identity. The paper notes that by March 2020, polarization may not yet have hardened as it did by December, potentially explaining why cognitive sophistication did not interact with ideology.

Implications: Suggests that misinformation interventions targeting "critical thinking" may be necessary but insufficient in highly polarized environments where ideological incentives are strong. The finding that media trust (not raw reasoning ability) predicts divergence implies that interventions should address information environment fragmentation and trust in sources, not just reasoning capacity.