Checking on the facts

Fact-checking may be American journalism’s most influential export.

What began in the United States in the early 2000s has now spread to more than 50 counties.  Some 113 independent fact-checkers operate today.

While accuracy is a foundational element of modern journalism, the fact-checking movement focuses almost solely on evaluating the veracity of newsworthy claims made by politicians or others. Departing from passive reporting and horse race analysis, fact-checkers instead hope to render public judgements of what’s true, and what isn’t. Anchored by three big players (FactCheck.org, Politifact, and Washington Post’s Fact Checker), the practice has come to play an increasingly prominent role in political coverage in the United States.

In Europe, most fact-checkers operate as independent ventures, often rejecting the label of journalism outright. Outside the U.S., where newspapers dominate, only 44 percent of the world’s active fact-checkers are affiliated with media organizations.

While pioneering American fact-check outfits such as the Washington Post’s cast long shadows — one of France’s first features likewise adopted Les Pinocchios as its cultural peg — many of the format’s recent innovations have been spurred abroad.

Spain’s Maldito Bulo (“damn hoax”), for example, was launched specifically to debunk viral hoaxes, and has weighed in on misleading images from both sides of recent the Catalan demonstrations.  By superimposing corrections directly over hoax images, the group spreads fact-checks by the same channels as the misinformation they address.

In Colombia, the news site La Silla Vacía uses WhatsApp to stanch the spread of false claims in their native habitat. WhatsApp — the world’s most popular messaging service — has become a hotbed of misinformation in many countries. Whereas Facebook and Twitter discussions are largely public, WhatsApp chats are compartmentalized, its groups are limited to 256 users, and all information is encrypted. This makes tackling rumors spread though the app especially difficult. Rather than rely on misinformed readers seeking out their website, La Silla Vacía makes use of their most-trusted friends and relatives to share the correction directly, via private chat.

While new outfits push the boundaries, ties to legacy media have their own benefits.  Most of Europe’s established media outlets offer at least occasional fact-checking, from Le Monde in France to Germany’s Der Spiegel.  These outlets provide reach that can dwarf the upstarts’. Most notably, up to 2 million viewers each week see the fact-checking segments on El Objetivo con Ana Pastor, aired by Spanish TV network La Sexta.

American fact-checkers have yet to tap into targeted social media distribution or big TV audiences.  But moves in the European Union to combat fake news might up-end Continental fact-checking’s credibility, according to Alexios Mantzarlis, director of the International Fact-Checking Network. This would reproduce the problem that has long-plagued the enterprise in America.

Feelings about the EU remain one of the biggest political cleavages in most member states.

If EU institutions or pro-EU think tanks establish fact-checking features to address far-right misinformation, or begin legislating the truth (“ban fake news”), as Pres. Macron has signaled he might, they risk transferring this polarization to readers’ feelings about fact-checking more broadly, Mantzarlis argues.

Addressing trust

Conservatives’ distrust remains U.S. fact-checkers’ biggest problem.  Several have taken steps to address this shortfall in recent months, but often stumbled.

Politifact sent staffers last fall on a listening tour of Red America, visiting Tulsa, Oklahoma, Mobile, Alabama, and Charleston, West Virginia. They also took on fact-checking political claims from these locales in an attempt to build a rapport.

In February, though, Politifact announced it had hired two politicians — former Florida Reps. David Jolly (a Republican) and Alan Grayson (a Democrat) as “reader advocates,” likely intending to fight off the stink of liberal bias.  Almost immediately, however, readers expressed disbelief: Grayson had been in a physical altercation with a reporter in 2016.  He was fired within hours. Politifact had failed embarrassingly in its basic fact-finding mission.

In another move intended to win over conservatives, Facebook added the Weekly Standard to its fact-checking collaboration initiative in December.  The initiative began in the wake of the 2016 U.S. election, allowing a small set of third-party fact-checkers — ABC News, AP, FactCheck.org, Politifact and Snopes — to help flag fake news.

The Weekly Standard became the first right-leaning (and first explicitly partisan) news outlet in the group.  Citing the Weekly Standard’s history of questionable content, left-leaning groups protested. “I’m really disheartened and disturbed by this,” said Angelo Carusone, president of Media Matters for America.  “They have described themselves as an opinion magazine.”

To many, admitting the Weekly Standard’s fledgling weeks-old fact-check division smacked of appeasement.

Changing minds

Facebook’s initiative has also been the site of renewed debate over just what impact fact-checks might have.

Gordon Pennycook and David Rand, psychologists at Yale, released a study in September showing that tagging inaccurate news stories likely won’t work.  A “disputed” tag improved readers’ accuracy judgement of news stories by just 3.7 percent.  The research also suggested that flagging stories may have an ironic effect.  Some readers may come to see any story not tagged as more likely to be true, despite that fact that the vast majority of news stories are never reviewed at all.

In December Facebook appeared to respond, shifting away from disputed tags and instead displaying “Related Stories” below identified false news stories to “help give people more context about the story,” according to Tessa Lyons, Facebook’s Product Manager.

Facebook may generally want to avoid any role as arbiter of truth. Even when offloading the role to third-parties, the company has faced claims of censorship, regardless of to what degree the First Amendment applies to a private company running a free service.

But the reasoning Facebook gave for the shift – that strong warnings like red flags may “entrench deeply held beliefs” – also calls back to mind the now-famous “backfire effect.”

The legacy of the backfire effect is worth revisiting. In a study that began making the rounds a decade ago, a pair of political scientists showed that fact-checks can, in extreme circumstances, increase unsupported beliefs among groups who hold them dear. In particular, Dartmouth’s Brendan Nyhan and the University of Exeter’s Jason Reifler showed that debunking the presence of Weapons of Mass Destruction in Iraq doubled the belief among conservative readers.

News of the study was carried in the Washington Post, Slate, the Wall Street Journal, This American Life, and hundreds of outlets around the world.  The study has since been cited 794 times in the academic literature.  For many, the sense that fact-checking was a doomed enterprise sank in.

But the coverage may have been oversimplified.  Far from the rule, backfires appear to be the exception.

A new pair of political scientists, Ethan Porter at George Washington University and Thomas Wood at Ohio State University, conducted a follow-up study with 10,000 subjects, and thus greater statistical power, assessing how well readers respond to fact-checks across dozens of issues.  They found no instances of backfire. Instead they find readers generally update their beliefs in the correct direction, but those whose party is impugned move less.

Similarly, a Nyhan, Reifler, Porter and Wood collaboration conducted during the 2016 election showed Trump supporters would accept corrections to his misleading statements, but the fact-checks had no impact on attitudes toward Trump or intention to vote for him.

Even at their best, though, it is important to note that the power of facts is limited.

As Ullrich Ecker, a psychologist at the University of Western Australia, told Slate recently, “whether or not they backfire, that’s up for debate.  But look, if it’s ineffective, that’s pretty much the same story as if there’s a small backfire effect.”

Fact-checking may be more successful at influencing elites than readers, cutting false claims off at the source in some instances. Nyhan and Reifler conducted a field study in 2012 focusing on about 1,000 state legislators in the U.S.  Those informed that they would be monitored by fact-checkers, with their misleading claims embarrassingly dissected before the public, ended up having a significantly better record on misleading statements in subsequent months than a control group.

Outlook

As the fact-checking movement grows and matures, its future is unclear.  Even as new outlets innovate, fact-checkers will continue to face issues revolving around reach, trust, and influence.

According to Lucas Graves, a journalism scholar at the University of Wisconsin, however, “fact-checkers should be understood above all as journalistic reformers.”  Rather undertaking an impossible mission fixed on eradicating all political lies, they instead set out to change how reporters and editors approach their jobs.

But while fact-checking should be conducted regardless of the ultimate outcome, its practitioners also should be willing to adapt to have an impact. Experimentation in format, medium, delivery mechanism, ratings scale, and any other aspect should be on the table.

Practitioners must also accept that new solutions may also generate new problems.  After abandoning the “disputed” tag, Facebook has announced its intention to crowd-source ratings of outlet credibility. In other words, Facebook plans to ask users whether they are familiar with a given news source, and they trust that source. Those that are widely trusted would receive preferential placement in users’ feeds, while those that don’t would essentially be buried.

Again Pennycook and Rand have swooped in to test how well this might work in practice. They find that crowd-sourced trustworthiness ratings are much less effective if they exclude ratings from readers unfamiliar with a given site, as Facebook plans to do.  Instead, Pennycook and Rand’s study, released in February, suggests “a lack of familiarity is an important cue for untrustworthiness.”

The lesson, it seems, is that new obstacles will always arise. Shrugging this off as the result of a new “post-truth age,” is not the answer, though – When exactly was the Age of Truth, again? Instead, fact-checking will be best served by continuing to blend optimism and pragmatism in its commitment to holding the powerful accountable.