Many people say facts do not change anyone's mind if they do not like them. We read studies from 23 countries on six continents, about vaccines, health, climate and politics.
False.
A test in 10 countries, from Brazil to Nigeria, checked a fact-check about vaccine myths. It made beliefs more accurate in 8 countries and worse in none. Most other studies on politics, health and climate found the same. But in Brazil, fact-checks did not work during the 2018 election or on Zika myths.
In short. No. In 10 countries, a fact-check about vaccine myths made beliefs more accurate in 8 and worse in none. People do prefer information that agrees with them, and a fact-check often gets weaker over time.
None. In 8 of the 10 countries, beliefs became clearly more accurate. In India and Indonesia, the change was too small to be sure. SRC-PVW2023
Each figure with its source, period and what it measures.
0 of 10countries
10,600 people · Brazil, France, Germany, India, Indonesia, Mexico, Nigeria, Peru, South Africa, US · 2021
0 of 52topics
Five US studies · more than 10,100 people
0.59points
28 tests · Argentina, Nigeria, South Africa, UK
46% vs 27%
US survey test · 1,000 people · 2012
91 studies
Review of many studies · just under 8,000 people
16 of 16countries
34,286 people on six continents · COVID headlines
Each belief checked on its own.
Correcting people makes them believe the false claim even more
No. Not in any of 10 countries tested on vaccine myths. Not on any of 52 political topics tested in the US either.
✕FalsePeople prefer information that agrees with them
Yes. On average across 91 studies, people chose agreeing information more often. In five countries, people picked news that agreed with them about 55% of the time.
✓TrueFact-checks only work in rich Western countries
No. They worked in Nigeria, South Africa, Brazil, Mexico and Peru. But in Brazil's 2018 election, fact-checks did not reduce belief in fake stories.
✕FalseIf people accept a fact, they change what they do
Not always. After a flu vaccine fact-check, the most worried people believed the myth less but were less willing to get vaccinated.
✕FalseWhen asked, most people want news that agrees with them
No. In the Reuters Institute's 2026 survey, 45% preferred news from sources with no particular point of view. 22% preferred sources that share their views.
✕FalseMostly yes. Ethan Porter, Yamil Velez and Thomas Wood tested vaccine fact-checks with 10,600 people in 10 countries. Beliefs became clearly more accurate in 8 countries and worse in none.
Another test ran in Argentina, Nigeria, South Africa and the UK. There, fact-checks lowered belief in false claims by 0.59 points on a 5-point scale. In Australia, voters on the left and the right both believed false statements by politicians less after a fact-check.
But it does not work everywhere. In Brazil's 2018 election, fact-checks did not reduce belief in six fake stories, even among people with no party. In Brazil, a fact-check about Zika did not reduce the myths and made some people doubt true facts too.
Do people also doubt good news about their own country? We checked one case: Is Finland really the happiest country in the world?
Yes, and often more easily. A review of 65 studies found that correcting health myths works better than correcting political or marketing claims. The same review found that real false claims are harder to correct than ones made up for a test.
In the UK, a sheet of vaccine myths and facts lowered belief in the autism myth right after reading. On average, it did not make people believe the myth more. A week later, the difference was too small to be sure.
A smaller 2017 study of 120 students in Scotland and Italy found something different. Belief in the autism myth went back up in the week after reading. The study does not say if it ended higher than before reading. The larger UK study tried to repeat this and could not.
In the US, people learned that 97% of climate scientists agree. Their view of how much scientists agree changed most among conservatives and doubters.
Not always. In the US, a fact-check said the flu vaccine cannot give you the flu. Fewer people believed the myth. But look at the most worried people. The researchers estimated that 27% of those who read the fact-check would say they were likely to get vaccinated. For those who did not read it, the estimate was 46%.
The 10-country test found no sign that the fact-checks made people want a COVID vaccine more or less.
In a test in France in 2017, people read Marine Le Pen's misleading claims about refugees. Their wish to vote for her rose by 5 percentage points, with or without a fact-check. The fact-check corrected their facts, but not their voting plans.
In Australia, voters liked a politician less when 4 of 5 checked statements were false. They did not like the politician less when half were false.
How often do politicians say false things in the first place? We checked that too: Do politicians lie to people?
It is the idea that a fact-check makes people believe a false claim even more. A 2010 study by Brendan Nyhan and Jason Reifler made the idea famous.
In the first test, 130 students at one US university took part, and 33 of them were conservatives. Among the conservatives who read a fact-check, 64% agreed that Iraq had weapons of mass destruction before the US invasion. Among those who did not read it, 32% agreed.
Among conservative students, belief also went up after a fact-check about tax cuts. When the researchers ran the Iraq test again, it did not happen.
Five later studies with more than 10,100 people did not find it once. In 2021, Nyhan himself wrote that it is extremely rare.
Not by much. Peter Ditto and his team looked at 51 studies with more than 18,000 people. They found almost the same bias on the left and the right.
Two other researchers, Jonathan Baron and John Jost, disagree with how this was measured. They say the studies were first made to answer other questions, so they were more likely to find no difference. They also say it may not be a mistake. If you have good reasons for a belief, it can make sense to doubt new information that goes against it.
Ditto's team answered that the studies covered many main topics in US politics. They say both sides had a similar chance to show bias. They also say they chose the studies where the 'it can make sense' explanation fits least.
These 51 studies are all from the US. A test in Europe and the US found that people picked news that agreed with them about 55% of the time.
Every chart is drawn from the downloadable data.
How often the fact-check clearly moved people closer to the facts, out of the topics tested. Liberals 85%, moderates 96%, conservatives 83%. Wood and Porter 2019.
Sources: SRC-WP2019 · Chart values must match the source register.
| Study | Where | Size | What it found |
|---|---|---|---|
| Porter, Velez & Wood 2023 | 10 countries on 5 continents | 10,600 people | Vaccine fact-checks: beliefs more accurate in 8 countries, worse in none |
| Wood & Porter 2019 | US | More than 10,100 people, 52 topics | Belief never went up after a fact-check |
| Porter & Wood 2021 | Argentina, Nigeria, South Africa, UK | 28 tests | Belief in false claims was 0.59 points lower on a 5-point scale |
| Aird et al. 2018 | Australia | 370 people | Left and right voters both believed false claims less after a fact-check |
| Barrera et al. 2020 | France | 2,480 people | Facts corrected, but the wish to vote for Le Pen stayed 5 points higher |
| Batista Pereira et al. 2022 | Brazil | 2,236 people | Fact-checks did not reduce belief in fake stories during the election |
| Carey et al. 2020 | Brazil | About 1,100 to 1,300 people per test | Zika fact-check failed and raised doubt about true facts. Yellow fever fact-check lowered belief in 2 of 3 myths |
| Nyhan & Reifler 2015 | US | 1,000 people | Flu myth belief fell, but the most worried were less willing to vaccinate |
| Ecker et al. 2023 | UK | 380 people | On average, no version made people believe the myth more |
| Pluviano et al. 2017 | Scotland and Italy | 120 students | Belief in the autism myth rose in the week after reading a myths-and-facts sheet |
| van der Linden et al. 2019 | US | 6,301 people | Learning that 97% of climate scientists agree moved doubters' view of that agreement most |
| Steppat et al. 2022 | Switzerland, Denmark, Italy, Poland, US | 4,349 people | People picked news that agreed with them about 55% of the time |
| Arechar et al. 2023 | 16 countries on 6 continents | 34,286 people | No fact-check: people judged true headlines as more accurate than false ones in every country |
| Hart et al. 2009 | Review of many studies | 91 studies, just under 8,000 people | People prefer information that agrees with them (moderate effect) |
| Ditto et al. 2019 | Review of US studies | 51 studies, more than 18,000 people | Almost the same bias on the left and the right |
| Nyhan & Reifler 2010 | US students | 130 students in the first Iraq test (33 conservatives) | Among conservatives, belief went up after a fact-check in 2 tests (Iraq, tax cuts) |
Sources: SRC-PVW2023SRC-ARECHAR2023SRC-WP2019SRC-PW2021SRC-AIRD2018SRC-BARRERA2020SRC-PEREIRA2022SRC-CAREY2020SRC-NR2015SRC-ECKER2023SRC-VDL2019SRC-STEPPAT2022SRC-HART2009SRC-DITTO2019SRC-NR2010SRC-PLUVIANO2017
Most of these studies were done in the US and Europe. A review of 155 tests found that over 80% came from rich countries. We found no fact-check test from the Middle East, only one where people judged true and false headlines. Most studies also ask people right after the fact-check. This verdict would change if big studies in many countries found that people often reject fact-checks they do not like.
Three cards, each made from this page's data. Settle the argument in the group chat.
Every number on the cards comes from this page's data. Please credit the source and link here so people can check it.
Need a figure for your own country or a different year? The data is downloadable below, and questions are welcome.
Wording you can copy. Each line says where the number comes from, so your readers can check it.
In a test with 10,600 people in 10 countries, a fact-check about COVID vaccine myths made beliefs more accurate in 8 countries and worse in none.
Porter, Velez & Wood, Royal Society Open Science, 2023
In five US studies covering 52 topics and more than 10,100 people, researchers found no case where a fact-check made people believe a false claim more.
Wood & Porter, Political Behavior, 2019
Niskasaari, L. (2026). Do people only believe facts they like? What the data says. FactsOut. Retrieved 12 Nov 2026.
Editor · Questions about the data and corrections are welcome.
Reuse and licensing: CC BY 4.0: free to reuse numbers, tables and the card with credit. If you publish online, please link to this page so readers can find the method and sources.
Where the figures come from and what they can and cannot tell us.
We tested the claim that people only believe facts they like, and five related beliefs.
Fact-check tests on vaccines, health, climate and politics in 14 countries, plus studies of judging headlines and choosing news. 23 countries in all. Also reviews that combine many studies on bias, and one large news survey.
We used published studies that other researchers checked before publication (peer-reviewed), plus the criticism of the main review and the reply to it. On 12 Nov 2026 we searched for studies that support the claim and for studies that go against it.
We read every number in the original study or on the publisher's page. We read most sources in full and a few only as summaries. Each source record says which.
The studies cover 23 countries on six continents, but most were done in the US and Europe. For the Middle East we found only a test of judging headlines, not a fact-check test. Most studies ask people right after the fact-check.
We did no calculations of our own on this page. Numbers are shown as the studies report them. For readers who want the statistics: the preference for agreeing information is d = 0.36 (Hart et al. 2009), and the bias is r = 0.235 on the left and r = 0.255 on the right (Ditto et al. 2019).
Few studies follow people for more than a few weeks. What people answer in a survey is not always what they do. Fact-checks worked less well in some studies in Brazil, India and Indonesia. Other researchers disagree with the main US review on bias.
No funding, no ads and no affiliate links on this page. Drafted with AI help. Before publishing, every number is checked against its original source.
Two separate checks on 12 Nov 2026 compared every number with its original source. Later that day we added studies from more countries and on health and climate, and two more separate checks compared those numbers with their sources. A final check went through the sentences corrected in those checks.
Correcting COVID-19 vaccine misinformation in 10 countries. Royal Society Open Science 10(3), 221097
Read in full (PMC10014243). Checked twice on 12 Nov 2026.
Open original sourceUnderstanding and combatting misinformation across 16 countries on six continents. Nature Human Behaviour 7, 1502–1513
Read in full (author PDF). Tests judging headlines, not fact-checks. Checked twice on 12 Nov 2026.
Open original sourceDoes truth matter to voters? The effects of correcting political misinformation in an Australian sample. Royal Society Open Science 5(12), 180593
Read in full (PMC6304148). Checked twice on 12 Nov 2026.
Open original sourceFacts, alternative facts, and fact checking in times of post-truth politics. Journal of Public Economics 182, 104123
Read in full (author repository PDF). Checked twice on 12 Nov 2026.
Open original sourceFake news, fact checking, and partisanship: The resilience of rumors in the 2018 Brazilian elections. Journal of Politics 84(4), 2188–2201
Read in full (author PDF). Counter-evidence. Checked twice on 12 Nov 2026.
Open original sourceThe effects of corrective information about disease epidemics and outbreaks: Evidence from Zika and yellow fever in Brazil. Science Advances 6(5), eaaw7449
Read in full (PMC6989147). Counter-evidence. Checked twice on 12 Nov 2026.
Open original sourceDoes correcting myths about the flu vaccine work? Vaccine 33(3), 459–464
Read in full (author PDF). Checked twice on 12 Nov 2026.
Open original sourceMisinformation lingers in memory: Failure of three pro-vaccination strategies. PLoS One 12(7), e0181640
Read in full (PMC5547702). Checked twice on 12 Nov 2026. Counter-evidence. The 2023 UK study by Ecker et al. tried to repeat it and could not.
Open original sourceCorrecting vaccine misinformation: A failure to replicate familiarity or fear-driven backfire effects. PLoS One 18(4), e0281140
Read in full (PMC10096191). Checked twice on 12 Nov 2026.
Open original sourceThe gateway belief model: A large-scale replication. Journal of Environmental Psychology (2019)
Read in full (accepted manuscript). Checked twice on 12 Nov 2026.
Open original sourceSelective exposure in different political information environments: How media fragmentation and polarization shape congruent news use. European Journal of Communication 37(1), 82–102
Read in full (PMC8958559). Checked twice on 12 Nov 2026.
Open original sourceHow to unring the bell: A meta-analytic approach to correction of misinformation. Communication Monographs 85(3), 423–441
Abstract only. Checked twice on 12 Nov 2026.
Open original sourceInterventions to counter misinformation: Lessons from the Global North and applications to the Global South. Current Opinion in Psychology 55, 101732
Read in full (PDF). Checked twice on 12 Nov 2026.
Open original sourceThe elusive backfire effect: Mass attitudes' steadfast factual adherence. Political Behavior 41(1), 135–163
Read in full. Checked twice on 12 Nov 2026.
Open original sourceThe global effectiveness of fact-checking. PNAS 118(37), e2104235118
Read in full. Checked twice on 12 Nov 2026.
Open original sourceWhy the backfire effect does not explain the durability of political misperceptions. PNAS 118(15), e1912440117
Read in full in the second check.
Open original sourceFeeling validated versus being correct: A meta-analysis of selective exposure to information. Psychological Bulletin 135(4), 555–588
Read in full. Checked twice on 12 Nov 2026.
Open original sourceAt least bias is bipartisan: A meta-analytic comparison of partisan bias in liberals and conservatives. Perspectives on Psychological Science 14(2), 273–291
Abstract of the published version. Checked twice on 12 Nov 2026.
Open original sourceFalse equivalence: Are liberals and conservatives in the United States equally biased? Perspectives on Psychological Science 14(2), 292–303
Abstract.
Open original sourcePartisan bias and its discontents. Perspectives on Psychological Science 14(2), 304–316
Read in full in the second check.
Open original sourceWhen corrections fail: The persistence of political misperceptions. Political Behavior 32(2), 303–330
Read in full, footnote 21. Checked twice on 12 Nov 2026.
Open original sourceCan public demand for impartial news survive platforms and polarisation? Digital News Report 2026
45% and 22% from the article. Not comparable with the 2020 figures because the markets differ.
Open original sourceNew claim page. Same day: plain language, and studies from 23 countries and on health and climate added after Lauri's request.
CC BY 4.0: free to reuse numbers, tables and the card with credit. If you publish online, please link to this page so readers can find the method and sources.
Niskasaari, L. (2026). Do people only believe facts they like? What the data says. FactsOut. Retrieved 12 Nov 2026.
