Addressing COVID-19 Misinformation on Social Media Preemptively and Responsively

Emily K. Vraga; Leticia Bode

Disclosures

Emerging Infectious Diseases. 2021;27(2):396-403. 

In This Article

Discussion

Efforts to address misinformation on social media have taken on special urgency with the emergence of COVID-19. Mitigating the risks associated with COVID-19 requires sustained public action, so misinformation that promotes false preventives or cures can hinder necessary behaviors to reduce the spread of the disease. In this study, we tested whether sharing graphics from WHO designed to address COVID-19 misinformation can reduce misperceptions. Our results suggest that although these graphics do not affect all misperceptions, reductions in misperceptions that do occur persist over time.

Notably, exposure to the WHO graphic in any form reduced immediate misperceptions about the science of a false preventive for COVID-19 (that a hot bath can raise body temperature), and this reduction was maintained for at least 1 week for 3 of the 4 correction conditions. This finding suggests that understanding of the science behind why hot baths do not prevent COVID-19 prevention does not deteriorate rapidly.

Although these effects on reducing science-related misperceptions show the promise of the WHO graphics as myth busters on social media, we did not see a parallel reduction in the related misperceptions regarding prevention efficacy (that a hot bath will prevent COVID-19 infection). We offer several post hoc explanations for these findings. First, we suspect that a floor effect may partially explain these null effects; even in the control condition in wave 1, participants were largely well informed, rating the argument that a hot bath can prevent COVID-19 infection as at least probably false (55.8% had an average score ≤2 or less on a scale of 1, definitely false, to 5, definitely true). In contrast, only 17.5% believed that the claim that a hot bath can raise body temperature was probably false, offering more leverage to change beliefs. Second, motivated reasoning may make persons more resistant to updating beliefs as issues around COVID-19 and the WHO become more politicized in the United States;[27] this motivated reasoning is likely less operant for the science of why such prevention is not effective. Third, persons may have thought that the science regarding hot baths and their effects on body temperature is better established given longstanding research,[11,12] boosting confidence in the validity of the correction. Given high levels of scientific as well as public uncertainty regarding COVID-19,[28] the public may have been less convinced regarding the scientific evidence that a hot bath does not prevent COVID-19.

Finally, the fact that a hot bath does not raise body temperature may not be the only (or even the most prominent) reason that persons may believe that taking a hot bath decreases the risk of COVID-19 infection. A supplemental analysis (Appendix 5 Table 1, https://wwwnc.cdc.gov/EID/article/27/2/20-3139-App5.pdf) provides some evidence for this explanation. In the pure control condition, the correlation between misperceptions that a hot bath raises body temperature and a hot bath can prevent COVID-19 is not significant (Pearson's correlation coefficient r = 0.06; p = 0.16). In the misinformation-only condition, the correlation is not significantly stronger than in the control condition (p = 0.27). However, for both WHO correction conditions, the correlation is significantly stronger than both the pure control and misinformation conditions (p<0.05). This preliminary evidence suggests that the correction, especially when shared by WHO, helps participants mentally link the science claim and the prevention claim; however, this explanation accounts for, at most, 18% of variance in COVID-19 prevention beliefs. Therefore, the explanation for why hot baths do not prevent COVID-19 is not the only factor in persons' beliefs about prevention efficacy.

These effects were consistent whether the graphic was shared by WHO itself or by another user. We suspect the similar effects between users and WHO, in contrast to earlier research suggesting experts were more effective than users,[22,23] may result from the prominent labeling of WHO within the graphic itself, boosting the credibility of the post. Therefore, mobilizing users to share WHO's graphics may produce similar effects in reducing misperceptions.

We found limited evidence that preemptive corrections differ in their effectiveness from reactive corrections. Preemptive and responsive corrections are equally effective when considering whether hot baths affect body temperature, both immediately and over time. Likewise, both are unsuccessful in affecting misperceptions about the efficacy of hot baths to prevent COVID-19 infection immediately after exposure to the correction. If preemptive corrections are effective in reducing misperceptions for (some) myths, persons need not wait until seeing someone share misinformation but can share the posts created by official expert organizations to address misperceptions in society at large. Thus, more attention is needed to find ways to motivate persons to share these types of corrections on their feeds.

However, the reactive correction addresses both the prevention efficacy of a hot bath (which is raised by the misinformation post) and the science behind this explanation, which is not addressed in the misinformation post. If the misinformation had also offered an explanation for why a hot bath supposedly reduces COVID-19 risk through raising body temperature, perhaps a reactive correction would be more effective. Although research suggests that false cures and preventives are a major subset of COVID-19 misinformation,[2] these studies do not elaborate on whether the misinformation contains false claims about the science behind the myth. We suspect that providing false explanations is a subset of misinformation claims and therefore chose to have the misinformation post include only the COVID-19 prevention myth to enhance external validity. Best practices for correction suggest that including an alternative explanation and corroborating evidence enhances the power of corrections.[6,7,17] Furthermore, emerging research suggests that correcting a related myth not raised in the misinformation can reduce misperceptions on that related myth, serving as an alternative form of preemptive correction.[29]

We did find 1 case in which a responsive correction from WHO may be more effective than the other corrections: exposure to the WHO responsive condition reduces misperceptions that a hot bath can prevent COVID-19 infection as compared with the control condition 1 week later, although this result must be interpreted with caution given the insignificance of the model overall and the limited amount of variance explained. If this result holds, it could be that the WHO responsive condition is the most memorable, and therefore had the most lasting effect on misperceptions, which future research should test.

We also found that both body temperature and COVID-19 prevention misperceptions were lower in wave 2 than in wave 1 for both the control and misinformation conditions (Appendix 5 Table 2). We suspect that the debriefing that all participants viewed at the end of wave 1 of the study, which included the WHO graphic and explained the myth, functioned as a correction itself (as intended to reduce potential misperceptions). Therefore, it is noteworthy that some correction conditions reduced hot bath misperceptions even further in wave 2 compared with the control, which reinforces the value of multiple corrections.[7,22]

This study's limitations suggest caution in interpreting our findings. First, we relied on a diverse but unrepresentative sample of the US public, most notably skewing educated and male. Future research should explore these effects among a representative sample and samples outside the United States, including countries where the worst of the pandemic has passed and ones that are struggling to contain new outbreaks, to examine how these contexts affect the relationships we observed here. Second, although our study suggests that the WHO graphics have potential given their effects on body temperature misperceptions, low levels of initial belief that hot baths can prevent COVID-19 limited our ability to perceive potential effects on prevention efficacy. Similarly, the post promoting misinformation about hot baths preventing COVID-19 was largely not persuasive in generating misperceptions. Future research should consider efforts to debunk more prominent or plausible COVID-19 myths. Third, we selected a myth with little partisan divide; we cannot speak to whether these graphics would be effective for politically polarized myths.[11] Fourth, the effect sizes explained were relatively small, so corrections should be deployed as part of a larger health communication strategy for promoting accurate COVID-19 information.

Despite these limitations, this study offers several practical and theoretical advancements. First, we found little evidence of a backfire effect in promoting misperceptions of sharing the WHO's infographics on social media. This finding not only fits with increasing evidence about the rarity of backfire effects[30] but is also reassuring that sharing the graphics at least does no harm. Second, we find that preemptively sharing these graphics can be effective. Users and organizations can debunk misinformation circulating in society by sharing high-quality information on social media emphasizing the facts without waiting to see it shared directly in their feeds, which expands the opportunities for observational correction to occur. Third, we found that a WHO bot that directly responds to misinformation may be a particularly effective technique. Partnerships with platforms may enable these automated responses to prominent myths, furthering the reach of expert organizations. Creating easily shared graphics that promote facts in spaces in which misinformation abounds appears promising as part of a broader strategy to enable more efficient and effective corrections on social media.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....