Effects of Fact-Checking Social Media Vaccine Misinformation on Attitudes toward Vaccines

"What approaches are most effective at targeting vaccine misinformation on social media among users unlikely to visit fact-checking websites or engage with thorough corrections?"

In light of increasing levels of vaccine hesitancy in several countries, researchers are studying the role of exposure to misinformation - particularly as it spreads on social media - as one of the factors that could negatively influence vaccination attitudes and decisions. One approach being studied is the design of platform-based interventions to provide social media users with signals on content and source quality. This paper explores 2 central questions: (i) Can one such signal - fact-checking labels on misinformation - result in more favourable attitudes toward vaccines, and is the effect contingent upon race, education, and/or conspiracy ideation? (ii) Does the fact-checking labels' effect depend on the source to which the label is attributed?

Conducted from August to September 2018, the online experiment used a sample of 1,198 United States (US) adults. The researchers tested 5 fact-checking treatment conditions using 13 distinct sources representing 5 source categories: (i) algorithms (i.e., Deep Learning Algorithms), (ii) health institutions (e.g., the Centers for Disease Control and Prevention, or CDC), (iii) news media (e.g., the American Broadcasting Corporation, or ABC), (iv) fact-checking organisations (e.g., Snopes), and (v) research universities (e.g., Johns Hopkins University). Participants were exposed to mock Twitter page featuring a misinformation claim that a specific vaccine caused harm. One post, for example, read: "According to a US Vaccine Adverse Events Reporting System (VAERS) there were 93,000 adverse reactions to last year's Flu Shot including 1,080 deaths & 8,888 hospitalizations." The treatment condition added a simple fact-checking label below the misinformation message, which consisted of a red warning sign, a falsification message, and a source logo. One read, for example, "This post is falsified. Fact-checked by the Centers For Disease Control. Learn why this is falsified." The control condition included only misinformation.

The researchers analysed differences in vaccine attitudes between the fact-checking label and control conditions. Fact-checking labels attached to misinformation posts made vaccine attitudes more positive compared to the misinformation control condition. The effect of the fact-checking labels on attitudes were similar for participants with higher or lower levels of baseline skepticism. Supplementary analyses found that the labels' effect was not moderated by the type of vaccine misinformation or by political ideology. However, conspiracy ideation moderated the effect of the labels on vaccine attitudes, with a stronger effect for the participants who had a higher level of conspiracy ideation. Race and education level did not moderate the fact-checking labels effect. Fact-checking labels from health institutions and research universities were seen as more "expert" than others, indirectly resulting in more positive attitudes toward vaccines.

Thus, this study suggests the potential for a proposed "versatile and simple factchecking intervention to target individuals who may have expressed or have been exposed to conspiracy theories." Incorporating labels from trusted universities and health institutions on social media platforms could be a promising direction for addressing the vaccine misinformation problem. This points to the potential that closer collaboration between public health and research institutions and social media companies holds to address the misinformation threat.

The data were collected before the COVID-19 pandemic, but the study's results could influence public communications about COVID-19 vaccines, the researchers said.