Reducing Online Misinformation

Welcome back. Several weeks ago, I took my car to the dealer for service, which was estimated to take an hour. When I arrived, there was one young woman in the waiting room. She left as I was reading, and soon, a middle-aged couple and one other customer joined me.

Though I walked a bit every 15 minutes or so, I started getting antsy after an hour. While I was wandering the corridor and showroom, another customer joined the waiting group, a non-stop talker I could hear from the corridor: Fauci can’t make up his mind; now he says 3 feet not 6 feet is enough…I don’t know why they’re picking the Floyd jury now when the trial isn’t until November...We have to do something; China has more aircraft carriers than we do.

As I debated whether to correct this fountain of misinformation (Dr. Fauci was reporting the latest research finding. The trial will start, as it did, when the jury was selected. China has far fewer aircraft carriers than the U.S., and as yet, none is nuclear-powered), I was advised that my car was ready. 

Fabricated social media posts have lured millions into sharing lies (graphic by Dave Cutler from https://www.pnas.org/content/114/48/12631).

Would my corrections, offered tactfully of course, have made any difference? A recent study suggests they could have had an effect.

Why People Share Misinformation
A team of researchers then affiliated with Canada’s University of Regina, Massachusetts Institute of Technology and the U.K.’s University of Exeter conducted a series of experiments to address why people share misinformation on social media. They tested three theories: confusion about what is accurate, preference for factors such as political partisanship over accuracy, and inattention to accuracy.

Their initial experiment tested 1,002 MTurk American participants on what people deem accurate and what they would share on social media. The participants were presented the headline, lead sentence and image for 36 news stories taken from social media. Half of the headlines were false, half true, half favorable to Democrats, half to Republicans. Participants were randomly assigned to either judge each headline’s accuracy or indicate whether they would share each headline.

True headlines were rated as accurate 56% more often than false headlines, but partisan alignment had a significantly larger effect on sharing intentions (19%) than whether the headline was true or false (6%). Nevertheless, participants overwhelmingly said accuracy was more important than partisanship when deciding whether to share on social media. 

When 1,002 American participants judged the accuracy of 36 headlines and whether they would share them given their political alignment with the headlines, accuracy judgements were much more discerning than sharing intentions despite the participants’ desire to share only accurate content (from https://www.nature.com/articles/s41586-021-03344-2).

The disconnect between accuracy and sharing formed the basis for additional experiments.

Effect of Prompting Accuracy
In one experiment, for example, participants were asked the likelihood they would share 24 headlines. To induce thoughts of accuracy, the treatment condition began by first asking participants to rate the accuracy of a single nonpartisan headline; that step was omitted for the control condition.

Participants who rated the accuracy of the single introductory headline were significantly less likely to share false headlines yet just as likely to share true headlines as the control participants that did not rate the introductory headline. The difference in sharing intentions for true versus false headlines was two times larger, with 727 MTurk American participants.

In another experiment, participants were asked to assess the accuracy of each headline before deciding whether they would share it. Of sharing intentions for false headlines, inattention to accuracy explained 51%, confusion about what was accurate explained 33%, and preferring partisanship over accuracy explained 16%, confirming that inattention has a central role in the sharing of misinformation.

Wrap Up
To test the findings on social media, the researchers sent an unsolicited message to 5,379 Twitter users who had recently shared links to websites that regularly produce misleading and hyperpartisan content. The message asked the recipients to rate the accuracy of a single nonpolitical headline. Monitoring for 24 hours showed the Twitter users increased the average accuracy of news they shared after receiving the accuracy message.

Overall, the study demonstrated that most people fail to implement their preference for accurate sharing because their attention is focused on factors other than accuracy. The results challenge the notion that people opt for partisanship over accuracy. It would appear that attention-based interventions, easily implemented by social media platforms, would help reduce online misinformation.

Thanks for stopping by.

P.S.
Study of online misinformation in Nature journal: www.nature.com/articles/s41586-021-03344-2
Article on study on EurekAlert! website: www.eurekalert.org/pub_releases/2021-03/miot-arf031521.php