[EMBARGOED] The 2020 election saw fewer people clicking on misinformation websites, Stanford study finds

(Press-News.org) In the run-up to the 2020 election, people appear to have become savvier in spotting misinformation online: clicks onto unreliable websites have declined, according to a new Stanford study published April 13 in the journal Nature Human Behaviour. According to prior research, some 44.3 percent of Americans visited websites during the 2016 U.S. election that repeatedly made false or misleading information. 

During the 2020 election, Stanford scholars saw that number drop by nearly half to 26.2 percent.

While these findings are promising, the scholars are cautious in interpreting the study’s results. Exposure even among fewer people can still have serious consequences, they noted in the paper. Extrapolating their results, the scholars estimated that nearly 68 million Americans made a total of 1.5 billion visits to untrustworthy websites during the 2020 election.

“Although we saw a serious reduction in the overall number of people exposed to misinformation on the web, misinformation remains a serious problem in the information ecosystem for some populations, especially older adults and diverse communities,” said Jeff Hancock, a professor of communication in Stanford’s School of Humanities and Sciences and senior author of the study.

The scholars found that those who did visit websites touting false claims tended to be older and lean more to the right of the political spectrum, a finding consistent with 2016 data. They did however visit fewer untrustworthy websites and spend less time on them than they did in 2016.

How the study worked The study builds on previous research conducted by Andrew Guess at Princeton University. In 2016, Guess compiled a list of some 490 untrustworthy websites that included pages that prominent researchers in the disinformation research community had previously identified, including Stanford economist Matthew Gentzkow.

Here, Hancock, along with Stanford PhD students Ryan Moore and Ross Dahike, augmented the list with an additional 1,240 unreliable domains from NewsGuard, an organization that rates the credibility of news and information websites. Their rankings are done manually by experienced journalists and editors who rigorously review and rank websites on a variety of criteria, including whether they repeatedly publish false content, issue corrections on errors in their reporting, and distinguish between news and opinion.

The Stanford researchers then recruited a representative sample of 1,151 American adults through the polling firm, YouGov. Participants completed an online survey and installed a browser plugin that allowed the researchers to passively track web activity between Oct. 2, 2020, and Nov. 9, 2020. In sum, they gathered some 7.5 million website visits on users’ desktop and mobile devices.

So, who reads false news online and how did they find it? The scholars found that in 2020, 5.6 percent of visits to untrustworthy websites were referred to by Facebook – in 2016, it was 15.1 percent. The scholars credit this decrease to efforts the social media platform took to mitigate the issue of false news on the website.

“The drop in visits referred by Facebook may reflect investment in trust and safety efforts to decrease the prevalence of misinformation on their platform, such as flagging, content moderation, and user education, which they and other platforms weren’t doing as much of in 2016,” said Moore.

While the scholars found that the average number of times a person visited a misinformation website decreased from an average of 32 visits in 2016 to 23 visits in 2020, there are a few individuals who still consumed misinformation online at extremely high clicks. “There are some people still consuming hundreds of misinformation websites,” said Dahlke. “We need more research to understand the effect of this type of exposure on people’s beliefs and actions.”

The scholars also found that older adults were twice as likely to visit a misinformation website compared to those aged 18-29 years old. While a smaller percentage of Americans 65 and older were exposed in 2020 (56.2 percent) than in 2016 (37.4 percent), they continue to consume misinformation at much higher rates than younger adults.

“Older adults continue to be targeted by misinformation purveyors because that generation tends to be wealthier and more civically engaged than other generations, making them prime targets for bad actors trying to make money or change election outcomes,” said Hancock.

Misinformation evolves, mutates Misinformation is pernicious, it morphs and mutates quickly, the scholars said.

“While one could interpret our findings as evidence that the problem of online misinformation is improving in some way, they could also be interpreted as evidence that the nature of the problem is changing,” the scholars write in the paper.

The scholars only studied web browsing activity, and misinformation could have been displaced to other social media platforms or encrypted messaging services, such as WhatsApp or Signal. Moreover, a click is not the only metric of fake news consumption; people could have still consumed untrustworthy information passively online through a meme or even just skimming a headline when scrolling through news feeds. All of these factors make it a difficult topic to study.

Looking ahead to the 2024 election Hancock, Moore, and Dahike are already thinking about what their findings might reveal about how misinformation will spread in the next general election in 2024.

They anticipate that older adults will continue to be vulnerable to fake news, an issue that Hancock’s Social Media Lab has separately been working to address with support from the Stanford Impact Lab program. In 2020, Hancock and Moore collaborated with the nonprofit news organization Poynter to create a digital media literacy intervention to help seniors identify misinformation online.

They are also concerned about the role of misinformation in under-resourced areas, such as non-English speaking communities, as highlighted in a recent paper Hancock and Moore co-authored with Stanford PhD student Angela Y. Lee on the topic.

-30-

END