Lieu, R., Hayes, O. R., & Cook, J. (2025). Testing the impact of fallacies and contrarian claims in climate change misinformation. British Journal of Psychology. https://doi.org/10.1111/bjop.70049
Summary
This experimental study tests how different types of climate misinformation shape perceived credibility, social media engagement intent, and climate beliefs among a representative US sample (N=1311). The authors cross two analytic frameworks — the content-based CARDS taxonomy of contrarian claims and the logic-based FLICC taxonomy of rhetorical fallacies — in a 5×6 design using 30 fabricated Facebook posts. Their central finding is asymmetric: the content of denialist arguments matters substantially while the fallacy structure does not, with attacks on climate solutions emerging as both the most credible-seeming and the most ideologically polarizing category. The paper argues this justifies prioritizing solutions-denial and attacks on scientists for debunking and automated fact-checking efforts.
Key Contributions
- First experiment to cross the CARDS (content) and FLICC (logic) taxonomies, allowing orthogonal comparison of what makes climate misinformation persuasive.
- Empirical identification of solutions-denial (CARDS 4) and attacks on scientists (CARDS 5) as the most polarizing varieties of climate denial.
- Evidence-based triage guidance for human fact-checkers and AI-based detection systems on which myths most warrant intervention.
- An open stimulus set of 30 crossed misinformation posts plus OSF data for replication.
- Further evidence of conservatives’ asymmetric vulnerability to climate misinformation, refined by content category.
Methods
An online experiment recruited ~1311 US adults via Walr into a 5 (CARDS content category) × 6 (FLICC fallacy: misrepresentation, oversimplification, red herring, false equivalence, cherry picking, slothful induction) between-subjects design, plus a control, yielding 31 conditions. Stimuli were 30 fabricated Facebook posts generated with Generatestatus, each instantiating a unique content × fallacy combination drawn from existing misinformation corpora. Outcomes were perceived veracity (5-item scale covering accuracy, trustworthiness, believability, credibility, informativeness), interaction intent (like/comment/share), and three climate beliefs (it’s real, it’s us, it’s bad). Kruskal–Wallis H tests, linear regression with ideology as moderator, and simple-slopes analyses were used.
Findings
- Perceived veracity varied significantly across the 30 individual stimuli (χ²(29)=60.83, p<.001, η²=.026), but no effects on interaction intent or overall climate belief.
- Grouped by FLICC fallacy: no significant differences on any outcome — logical structure alone did not drive perceived credibility.
- Grouped by CARDS category: category 4 (“solutions won’t work”) was rated significantly more trustworthy, accurate, and credible than other categories.
- Highest-rated individual myths: “no consensus on sea level rise,” “polar bears are improving,” “China emits more.” Lowest-rated: “climate change is like religion,” “climate sensitivity is low.”
- Political ideology significantly interacted with CARDS 4 and 5, with steepest ideological slopes indicating these are the most polarizing categories.
- Among conservatives only, exposure to CARDS 4 significantly lowered climate belief relative to control and other categories.
Connections
This paper extends the experimental misinformation-effects tradition into the taxonomy-comparison space; it pairs naturally with Spampatti2026-kx on climate misinformation interventions and with van-der-Linden2026-jt given the latter’s foundational work on inoculation against science denial that explicitly draws on FLICC-style logic-based prebunking. The finding that ideology moderates susceptibility echoes broader exposure asymmetries documented in Gonzalez-Bailon2024-rq and Budak2024-ef. The argument for prioritizing certain misinformation types for automated detection connects to debunking-pipeline concerns raised in Dierickx2026-tw and Cazzamatta2026-lo.
Podcast
A research-radio episode discusses this paper: 🎧 MP3 · Spotify · Apple Podcasts