The 2009 Pulitzer Prize for National Reporting was awarded to PolitiFact, a then-upstart fact-checking initiative, for its coverage of a bruising presidential campaign.
Hear more from Anya van Wagtendonk about this story:
The presentation of journalism’s highest honor to what was then a novel idea — using “probing reporters and the power of the World Wide Web to examine … political claims,” as the award described — marked a turning point for the nascent fact-checking industry.
More than a decade later, fact-checking is a global phenomenon, with 300 websites in 80 countries. Many partner with legacy media outlets and the largest tech platforms on Earth.
“After [the Pulitzer] there was tremendous interest [in fact-checking] not just across the United States, but around the world,” said Mark Stencel, co-director of the Reporters’ Lab at Duke University, where he researches political fact-checking. “And we suddenly had this explosion of fact-checking that has grown into the journalism movement that it is today.”
Despite the growth in their ranks, fact-checkers are overwhelmed. Their numbers are dwarfed by the volume of misinformation spreading across the internet, often onto cable TV and into the real world — more effectively than ever before.
Over the last decade, researchers have tested how well fact-checking works at changing minds and behaviors. They’ve discovered that fact-checking does make us better informed, but whether or not we change our behavior is a different story. It turns out misinformation is not just about what we know is true or false.
“The science is clear that we don’t proceed from facts to opinions,” said Brendan Nyhan, a political scientist at Dartmouth College who has extensively studied online misinformation. “In the case of promoting vaccination, for example, more information is not necessarily the solution.”
Fact-checking is also dwarfed by the sheer scale of misinformation and disinformation campaigns. What’s more, dutiful fact-checks are not always reaching the right audience. In an increasingly fragmented media landscape, this weakens the opportunity for a diverse polity to believe they share truth.
As misinformation dominates the public debate over many of the most important issues of our time, including covid-19 and vaccination, it’s worth looking at to what extent fact-checking is the right tool to meet the challenges of the moment — and what else is needed.
Fact-checking can change minds
The good news, researchers largely agree, is that fact-checking helps inform us. Research has found that people who are presented with a fact-check are more likely to correctly answer informational questions about that subject.
Fact-checks “reduce false beliefs, full stop,” said Ethan Porter, who leads the Misinformation/Disinformation Lab at George Washington University’s Institute for Data, Democracy and Politics. “Compared to someone who has seen the misinformation without a fact-check … people who see just a fact-check are more accurate as a result.”
In a September study, Porter and Thomas Wood, a political scientist at the Ohio State University, found that fact-checking categorically improves people’s belief in true information, regardless of a person’s geography and political affiliation, and irrespective of the subject matter.
Published in the Proceedings of the National Academy of Sciences, the study looked at fact-checks in Argentina, Nigeria, South Africa and the U.K. in the fall of 2020, covering topics including covid-19, economics and politics. According to that study, fact-checking decreased people’s reception to misinformation, and misinformation did not significantly sway participants overall.
By and large, people don’t want to spread misinformation, researchers led by David Rand and Gordon Pennycook found in a March 2021 study published in Nature. Their research found that more than three-quarters of people believe in sharing accurate information, and most are good at distinguishing news from partisan or misleading information.
Fact-checking can expire
But those findings are just one piece of the puzzle. The positive effects of fact-checking may fade over time. Misinformation may often spread farther than fact-checks — and even if fact-checking could keep up, it still doesn’t reach the public equally.
“Scientifically, we’re quite confident that at least when people see accurate, factual information, it does improve the accuracy of the factual beliefs,” said Nyhan. “The effects aren’t necessarily massive. And, importantly, they don’t always last.”
According to the Brookings Institution, a fact-check’s effectiveness also depends on who is encountering it. While all people are slightly better informed after encountering a fact check, if a person is unsure about their belief, rather than certain about a misbelief, they are more likely to walk away from a fact-check with better information.
And the “hostile media effect” — wherein people on opposing sides of an issue find the same information to be counter to or biased against their respective worldviews — holds true for fact-checks as well. Consequently, some people can view fact-checks that challenge their existing opinions as partisan.
In the face of these stakes, this tool’s shortcomings are serious, too. One of the biggest criticisms of third-party fact-checking is that it’s simply not reaching the people who most need it. The audience for stand-alone fact-checks is skewed, with self-identified liberals and registered Democrats more likely to interact with and buy explanations from fact-checkers. And perhaps most distressingly, even if people have the correct facts in front of them, it doesn’t necessarily influence their behavior.
“One of the big concerns that all of us in the fact-checking world have and share is a concern that the people who are reading this journalism are not necessarily the people who need it most — that the message, the corrections and amplifications and clarifications that come out of these journalistic fact-checks … doesn’t always reach the public that needs it most,” Stencel said.
Consumption of fact-checking also skews toward people who are already more politically engaged, “the most likely to have strong predispositions in the first place, and therefore to know which candidates they support,” Nyhan added.
“When it comes to actually having contact with that kind of deep-in-the-weeds political content, that’s just not something normal people do,” he said.
Misinformation isn’t just about what is true and what is false
Perhaps the most beguiling aspect of misinformation is the extent to which truth isn’t always the key to behavior. In fact, research suggests that debunking a false claim might not even be the answer to changing how we think and act on important issues.
Information is instead up against a bevy of psychological and personal reasons why people make the choices they do, like community and cultural norms, and religious and identitarian beliefs.
“It’s not obvious that the best way to get people vaccinated is to focus on debunking misperceptions about the efficacy of vaccines,” Nyhan said. “People have lots of reasons for the attitudes they hold. They have lots of reasons for the behaviors they engage in. Knocking out one particular factual claim doesn’t necessarily change their minds.”
Shannon McGregor and Daniel Kreiss, journalism researchers at the University of North Carolina at Chapel Hill, argue that “social identities” — like race, class and gender — and how those “map onto partisan identities” have a much stronger influence on people’s political decision-making than information, accurate or otherwise.
“Instead of being swayed by a particular narrative (false or otherwise), people generally discern their own political identities, and those of others, to make choices at the polls,” they wrote in a recent Slate article.
This bears out in the demographics of who has been receiving covid-19 vaccines in the year that they have been widely available. Rates in the U.S. have lagged, and, according to the Kaiser Family Fund COVID-19 Vaccine Monitor, political identity is a stronger indicator of vaccine status than other demographic factors.
One paper on the subject sums up this attitude in a provocative title: “They Might Be a Liar But They’re My Liar.”
Social media can exacerbate people’s reliance on intuition and emotion when choosing what kind of information to share with peers, Rand and Pennycook argue — even if people do not seek to share incorrect information.
“There’s a big disconnect between what people believe and what they share,” the authors argue, because people are likely to share posts based on “social concerns,” like “how much engagement our posts will get, how much our friends will enjoy them, or what they communicate about our identities, and so on.”
People are also apt to see fact-checks that counter their existing beliefs as biased. According to a Poynter study, just under half of Americans think that fact-checkers are biased and do not trust corrections to be balanced.
This belief corresponds some with party affiliation: 70 percent of Republicans believe fact-checkers are not one-sided, compared with 29 percent of Democrats.
Fact-checking operations were set up in a specific media landscape. New players have accelerated the challenge.
According to Stencel, examples of newspapers explicitly debunking false information can be found in the early 20th century. The universe of fact-checking that modern news consumers may encounter — such as FactCheck.org, PolitiFact and the Washington Post’s Fact Checker, which ranks political statements on a scale of Pinocchios — emerged in the last two decades.
The sites were an outgrowth of a response to a culture of nastiness that took hold during the 1988 presidential campaign. During that time, journalists and scholars, including Kathleen Hall Jamieson, now of the Annenberg School of Communications at the University of Pennsylvania, argued that “there’s a journalistic obligation here to adjudicate matters that are consequential,” in Hall Jamieson’s words.
In 2003, White House officials and major newspapers were trumpeting information about a weapons program in Iraq that turned out not to exist, to justify a military intervention there that year. That same year, Hall Jamieson co-founded FactCheck.org.
That site’s initial purpose was to offer time-strapped journalists access to primary sources that they could consult for their own stories. It wasn’t intended as a public resource, Hall Jamieson said. But the organization was cited in the 2004 vice presidential debate, leading to so much traffic that the website crashed.
From there, similar sites popped up, including the Pulitzer-winning PolitiFact, which operates local partnerships and ranks statements on a Truth-O-Meter. Reuters, the Associated Press and Agence France-Presse operate fact-checking arms, and projects have cropped up worldwide.
Since 2014, Stencel and the Duke Reporters’ Lab have maintained a global fact-checking census of nonpartisan projects that check public figures’ claims, debunk misinformation and/or monitor political promises. These projects must be transparent about methodology and funding, and target political figures from all parties.
According to those criteria, nearly 350 such projects operate in at least 102 countries worldwide, with more than 60 based in the United States.
Many of these global operations developed in response to a particular media landscape — one that has shifted dramatically with the advent of social media.
In response to widespread criticism about the ease with which mistruths spread on their platforms, social media companies have launched fact-checking programs in recent years. Facebook touts partnerships with third-party fact-checkers worldwide to vet content shared over its platform. False or misleading content may be labeled and become less likely to show up on a feed.
Twitter recently partnered with AP and Reuters to promote reliable content, allows users to report misinformation independently, and is developing a tool, Birdwatch, to amend crowdsourced fact-checks to posts.
Many experts question the scale of these approaches, however, and their efficacy in reaching non-English speaking audiences. Fact-checking on Facebook does not reach the level of individual posts or infiltrate private groups devoted to sharing misinformation, for example.
“You now have the capacity to microtarget [misinformation] to very limited parts of the audience, and the fact-checkers can’t even find the stuff, so by the time the fact-checkers find it, it’s had its impact,” Hall Jamieson said. “So the partnerships between the fact-checking community and Facebook, for example, are actually really important, because it increases the likelihood the fact-checkers can find it before it really starts to go viral.”
Research also indicates that there can be a backlash to platforms labeling misinformation as such. While people may be less likely to share content flagged as false, according to the “implied truth effect,” they may also view the lack of such a label as an affirmative indication of truth.
And some experts have criticized platforms, particularly Facebook, for not doing enough to counteract the harm on their platforms.
Claire Wardle, who cofounded the nonprofit First Draft, which researches mis- and disinformation, has criticized the opacity of social media user data.
In her keynote address at a global fact-checking conference held by the Poynter Institute in October, Wardle noted that many misinformation researchers and fact-checkers conduct their work in partnership with social platforms but do not own their data.
And in a November column in the Washington Post, Wood and Porter — the GWU and OSU researchers — said that Facebook can do more to put fact-checks in front of readers. For example, they argue that Facebook should ensure that, if a reader sees a post later deemed false or misleading, the reader should see the follow-up fact-check.
“If that company truly cares about the cause of a well-informed public, it should bear the cost of bringing fact checks in front of the people who would benefit most from seeing them — the people who have been exposed to misinformation,” they wrote.
The fact-checkers themselves are aware of these challenges and considering ways to make the tool more accessible and appealing.
“Maybe there’s other formats to use, or maybe there are ways to generate fact-checks that appeal to, that are structured in a way that engage a suspicious reader,” Stencel said. “Almost every fact-check will walk you through the methodology of the fact-check. But maybe if you don’t declare the outcome at the beginning of it, you can get someone to walk through the evidence and maybe go, ‘OK, that makes sense to me.’”
More broadly, they’re responding to the times. One open question is how artificial intelligence technology could be harnessed to scale fact-checking. There is also robust discussion of “prebunking”: essentially inoculating readers against potential misinformation before it spreads.
And still, the global fact-checking industry continues to grow. Facebook and Twitter are expanding their reliance on this tool to quell misinformation on their platforms.
But none of these innovations solves the broader crisis facing media as a whole, like low levels of public trust and extreme polarization.
Without a more systemic approach to the issue, fact-checkers are ultimately left to play Whac-A-Mole with new claims. But the claims will keep coming.