Debunking Works: Massive Real-World Test Shows Climate Myth Rebuttals Shift Minds

When someone types a climate myth into Google and clicks through to a rebuttal on the website Skeptical Science, does the debunking actually change their mind? For the first time, the team behind the world’s most comprehensive collection of climate misinformation rebuttals has answered that question with hard numbers, running a real-world experiment on their own live website over nearly four years. The results, published in the journal Geoscience Communication, offer both reassurance and a warning for anyone fighting the spread of false claims about climate change.

The study, led by John Cook of the Melbourne Centre for Climate Change at the University of Melbourne together with Bärbel Winkler, Collin Maessen, Timo Lubitz, Doug Bostrom and Dana Nuccitelli, surveyed visitors who arrived at Skeptical Science rebuttals directly from Google searches. Between November 2021 and July 2025, the site displayed a pop-up invitation to 858,016 visitors. Of the 13,432 people who consented and completed an initial survey, 6,261 — about 46 percent — went on to read a rebuttal to the end and complete a follow-up question, providing the researchers with an unusually large and ecologically valid dataset.

The experimental design was elegantly simple. Each participant was shown a single statement about climate change — either a factual claim or a piece of misinformation, randomly assigned — and asked to rate their agreement on a six-point scale from strongly agree to strongly disagree. They then read the rebuttal they had clicked through to see. If they scrolled to the end, a second pop-up asked them to rate the same statement again. The researchers scored answers so that higher values always meant more accurate perceptions, allowing a direct before-and-after comparison of belief in both facts and myths.

The first striking finding concerned who visits the site at all. Nearly half of participants — 46.3 percent — already showed full agreement with climate facts or full disagreement with climate myths before reading anything, and 66.5 percent agreed with climate facts overall. The distribution of pre-existing beliefs was sharply bi-modal, with peaks at the extreme ends of the spectrum and only a minority of undecided visitors in the middle. This suggests the website serves two distinct audiences: people already convinced of climate science who are looking for ammunition to counter misinformation in their own conversations, and people who genuinely doubt the science but are open enough to search for answers.

The authors argue this pattern challenges a dismissive interpretation of such audiences. Critics have suggested that convinced visitors merely engage in what one researcher called chanting to the choir. But Cook and colleagues favour a more constructive framing: the site is teaching the choir to sing, equipping people with material to push back against misinformation. That matters because fear of confrontation drives what researchers call climate silence — many people avoid discussing climate change with friends and family altogether, producing a self-reinforcing spiral of silence. Conversely, discussing the issue raises awareness and triggers a positive feedback loop of further discussion.

The core result was encouraging but nuanced. Overall accuracy improved slightly after reading the rebuttals, though the combined change was not statistically significant. Looking at the two statement types separately revealed why: agreement with myth statements dropped significantly, which is exactly what debunking is supposed to achieve, but agreement with factual statements also dropped significantly — an unwelcome and counterproductive side effect. The two effects partially cancelled each other out. The researchers measured discernment, defined as the gap between belief in facts and belief in myths, and found it did increase overall, but the decline in fact acceptance remains a genuine concern for an organisation whose mission is to raise climate literacy.

Perhaps the most hopeful finding was who changed most. A strong negative correlation emerged between initial accuracy and the magnitude of belief change: the less accurate someone’s starting position, the greater their improvement after reading a rebuttal. Among participants who began with inaccurate views, 7.2 percent switched to accurate responses after a single reading. In other words, the rebuttals were most effective precisely on the people who needed them most — those most dismissive of climate science — suggesting that even brief, single-exposure corrections can reach the hardest-to-shift segment of the audience.

To understand why some rebuttals worked better than others, the team analysed the twenty rebuttals with the most data and then qualitatively examined the three best and three worst performers. The pattern was clear. Effective rebuttals, such as the one debunking the claim that climate impacts are not bad, communicated a clear replacement fact — that the harms of warming far outweigh the benefits — simply and repeatedly. Ineffective ones, such as the basic and intermediate versions of the claim that carbon dioxide was higher in the past, failed to state the key fact that a cooler ancient sun offset higher carbon dioxide levels, and failed to name the single-cause fallacy the myth commits. Psychological research recommends a fact-myth-fallacy-fact structure, and the analysis showed the strongest rebuttals hewed closest to it.

The findings arrive at a moment when the landscape of climate misinformation is shifting. Recent research shows contrarian claims increasingly target climate solutions rather than climate science itself — so-called discourses of delay — and solutions misinformation has proven especially polarising among political conservatives. Skeptical Science has responded by adding rebuttals of renewable energy myths, adapted from work with Columbia Law School’s Sabin Center, but the effectiveness of such corrections remains understudied. The team is now redesigning the website to integrate explicit fallacy explanations into its content infrastructure, hoping to lift fact acceptance while continuing to erode myth acceptance.

The study has limitations the authors acknowledge candidly. It measured only a single outcome — agreement with one statement — captured just one exposure to a correction whose effects are known to decay over time, and relied on a self-selected sample of website visitors rather than a representative population sample. Yet that same self-selection gave the field test real-world external validity that laboratory experiments cannot match. The researchers call for controlled experiments to test repeated exposure and isolate the contributions of replacement facts and fallacy explanations. For now, the message is clear: debunking works, especially on the most sceptical readers, but even the best corrections must be engineered with care, because a poorly constructed rebuttal can inadvertently erode belief in the very facts it sets out to defend.

Subject of Research: Effectiveness of online rebuttals in reducing climate change misperceptions among website visitors

Article Title: Quantifying the impact of Skeptical Science rebuttals in reducing climate misperceptions

Article References: Cook, J., Winkler, B., Maessen, C. J. H. M., Lubitz, T., Bostrom, D., & Nuccitelli, D. (2026). Quantifying the impact of Skeptical Science rebuttals in reducing climate misperceptions. Geoscience Communication, 9(2), 145-160. https://doi.org/10.5194/gc-9-145-2026

Image Credits: AI Generated

DOI: 10.5194/gc-9-145-2026

Keywords: climate misinformation, Skeptical Science, debunking, science communication, misinformation correction, climate myths, discernment, inoculation theory, public perceptions, Geoscience Communication, fact-checking, climate literacy