It is tempting to think that misinformation is a problem for other people: the gullible, the uneducated, the other party. The research is less flattering. Intelligent, educated people believe and share false things too, and the reasons are worth understanding, because they apply to everyone reading this sentence.
Biased or lazy?
One influential view holds that people believe what fits their group. The legal scholar and psychologist Dan Kahan calls this identity-protective cognition: processing evidence so as to protect your standing in a group you care about. In one well-known study, people who were good with numbers correctly interpreted data said to be about a skin cream. But when the same numbers were presented as evidence about gun control, many misread them when the right answer on gun control clashed with their politics. On this view, more skill can make people better at defending their side's beliefs, not better at finding the truth.
A second view, developed by the psychologists Gordon Pennycook and David Rand, emphasizes something more ordinary. In a 2019 paper titled "Lazy, Not Biased," they found that reflective thinkers spotted false headlines even when it hurt their side: people who scored higher on a test of careful reasoning were better at telling true headlines from false ones, including false headlines that favored their own politics. Much belief in false news, they argued, comes from not stopping to think. Scrolling quickly, people react to how a headline feels rather than asking whether it is accurate.
The two views are not exclusive. Identity matters most on charged issues where beliefs mark membership; inattention matters across the board. Both point to the same practical lesson: the moments of greatest risk are when a claim flatters your side and when you are moving fast.
The accuracy nudge
If inattention is part of the problem, a small prompt might help. In a study published in Nature in 2021, Pennycook, Rand and colleagues found that rating one headline's accuracy improved later sharing: people who were first asked to judge whether a single headline was true went on to share more reliable news. In a field experiment, the team sent more than 5,000 Twitter users who had shared links from low-quality sites a private message asking their opinion on the accuracy of one nonpolitical headline. Afterward, those users shared more news from higher-quality sources. Most people, the researchers concluded, do care about accuracy. They just forget to think about it in the moment. The effects are modest, and some replications have found them smaller, but the habit it suggests is cheap: before sharing, ask yourself whether the claim is true.
The backfire effect that mostly wasn't
For years, many people believed that correcting misinformation could make it stronger. A 2010 study by Brendan Nyhan and Jason Reifler found that some participants shown corrections about weapons of mass destruction in Iraq became more convinced of the false claim, a reaction called the backfire effect. The idea spread widely and discouraged many would-be correctors. But when Thomas Wood and Ethan Porter tested corrections on 52 different issues with more than 10,000 people, they found that backfire was rare, and corrections moved people toward facts, even on charged political topics. Nyhan himself has since written that backfire effects are uncommon. Correcting false claims is worth doing.
Corrections do not fully erase
Still, corrections have limits. In a classic 1994 experiment, Hollis Johnson and Colleen Seifert had people read a series of reports about a warehouse fire. An early report blamed paint and gas cylinders stored carelessly in a closet; a later report said the closet had in fact been empty. Even people who remembered the correction kept using the original explanation when asked why the fire burned so fiercely. This is the continued influence effect: misinformation shapes reasoning even after it is corrected and the correction is accepted. A bare retraction leaves a gap in the story, and the mind prefers a complete false story to an incomplete true one.
The practical answer is replacing the false explanation with a true one, not merely denying it. The linguist George Lakoff has popularized a related technique often called the "truth sandwich": lead with the fact, mention the falsehood briefly and say why it misleads, then return to the fact, so the false claim is not the part people remember.
Worked example
Maya, an engineer who prides herself on skepticism, shares a headline saying that a politician she dislikes has been arrested. It is false. Which forces were at work? The headline flattered her side, so identity-protective cognition may have played a part; she was scrolling on a crowded bus, so inattention almost certainly did. A friend replies with a correction. If Maya follows the research, she will not assume a correction will only backfire on her followers; she will post it. She will also do more than delete: she will say what actually happened, that the politician was questioned as a witness, not arrested, so her followers have a true story to replace the false one. Next time, a two-second accuracy check before sharing could have prevented the whole sequence.