The Process of Responding to COVID-19 Misinformation in a Social Media Feed

Abstract

Misinformation can undermine public health recommendations. Our team evaluated a 9-week social media campaign promoting COVID-19 prevention to mothers (n=303) of teen daughters in January-March 2021. We implemented an epidemiological model for monitoring, diagnosing, and responding quickly to misinformation from mothers. Overall, 54 comments out of 1617 total comments (3.3%) from 20 mothers (6.6% of sample) contained misinformation. Misinformation was presented in direct statements and indirectly as hypothetical questions, source derogation, and personal stories, and attributed to others. Misinformation occurred most (n=40; 74%) in comments on vaccination posts. The community manager responded to 48 (89%) misinformation comments by acknowledging the comment and rebutting misinformation. No mothers who provided misinformation left the Facebook groups and a few commented again (n=10) or reacted (n=3) to responses. Only a small number of comments conveyed misinformation. Our quick-response epidemiological protocol appeared to prevent debate and drop-out and exposed these mothers to credible information.

Context

Misinformation – erroneous, false, or misleading information that departs from scientifically-valid health information¹ – has been a public health problem before and during the COVID-19 pandemic,¹ especially when coupled with narrow media consumption habits, lack of health and media literacy skills, and tendencies to hold to pre-existing beliefs or express group identity.² Misinformation can confuse and influence people to reject public health advice so efforts to correct misinformation are needed.³ Scales et al.⁴ proposed an epidemiological model to counter misinformation using constant monitoring, diagnosis, and quick response. They recommended that responses start with acknowledging participants and their uncertainty, confusion, or motivation, to convey empathy, prevent defensiveness, and/or maintain trust among participants,⁴ before debunking the misinformation using various techniques (e.g., fact-checking, coherence appeals, reframing, safety/efficacy information, resistance-to-persuasion tactics, and stories with statistical content^5-8). In 2021, as COVID-19 vaccines were initially distributed, the authors conducted a 9-week social media campaign on COVID-19 prevention.⁹ Here, we describe experiences deploying a protocol to respond to misinformation in comments by users.

Program

The study enrolled mothers with teenage daughters (n=303), who had participated in an earlier trial on preventing indoor tanning. Mothers were recruited to the earlier trial using community-based methods and the Qualtrics internet panel and given its focus, mothers lived in one of 34 states not completely banning indoor tanning by minors. Mothers were recontacted and enrolled in a separate randomized single-factor experimental design. Following baseline assessment, mothers received a 9-week social media campaign on non-pharmaceutical interventions (NPIs; e.g, social distancing and masks), COVID-19 vaccination, media literacy, and family communication. Ethical approval was obtained from the Western Institutional Review Board on September 15, 2020 (Decision No 1-872442-1).

Implementation

Posts were delivered in three Facebook private groups from January to March, 2021 – 45 experimental posts plus 12 posts on study procedures or to attract engagement per group. Text for posts were written by the researchers and were nearly identical across groups, except that mothers were randomly assigned to receive links to different information sources in the posts – government agencies (n=100), near-peer parents (n=99), or news media (n=104) – in the three groups (a detailed description was published previously⁹). Randomization should have balanced exposure to these information sources outside of the study. Mothers were posttested at 3-weeks, 6-weeks, and 9-weeks post-randomization on NPIs, vaccination intentions, media literacy, family communication, and theoretic mediators (results reported elsewhere⁹). Most mothers (98.3%) stayed in the Facebook groups for all nine weeks; 80.5% completed all assessments.

A community manager administered the Facebook groups by scheduling posts (5 per week), monitoring reactions/comments by logging in at least twice per day, “liking” mothers’ comments to promote engagement, and responding to questions, comments, and technical problems. The manager was a mother and used her own photo and personal experiences to increase relatability. Designated community managers or social media managers for social media feeds are common in private businesses and are useful in social media feeds from public health departments.¹⁰ The manager identified misinformation in mothers’ comments, defined as containing information that was in opposition to recommendations from Centers for Disease Control and Prevention, National Institutes of Health, and World Health Organization . One post asked mothers what concerns about COVID-19 prevention they had heard from others to identify misinformation. The research team created responses to misinformation and posted them within 24 hours.

Evaluation

Mothers were middle-aged (mean=42.8 years), majority white (87.1%), college-educated (55.7%), and affluent (68.8% annual incomes>$60,000), but diverse politically (24.4% conservative, 50.2% middle-of-the-road, 25.4% liberal). Of the 135 posts total (n=45 per group; on average, mothers viewed over 35 posts per group⁹), 17 (13%) posts received comments containing misinformation (n=54 [3.3%] out of 1617 comments total) from 20 unique mothers (6.6% of sample; range=1-7 misinformation comments per mother). Misinformation took a number of forms (see examples in Table 1). In some comments, mothers directly presented false information or expressed their own negative opinions or opposition towards COVID-19 prevention. In other comments, mothers presented misinformation indirectly. This included labeling misinformation as a rumor or attributing it to others, presenting misinformation in a hypothetical or sarcastic question, disputing evidence supporting prevention advice but not the advice itself, derogating credibility of government officials and pharmaceutical companies, claiming not to know who to trust, praying prevention is effective, and sharing a personal experience that implied prevention is unnecessary. Finally, mothers’ comments agreeing with misinformation posted by another mother and comments that did not include misinformation but expressed concerns about overstating risk for COVID-19, equity of COVID-19 prevention, or effects of COVID-19 prevention on an existing health condition (e.g., diabetes) were also classified as misinformation.

Table 1:

Examples of Misinformation in Mothers’ Comments and Responses in the Social Media Feed on COVID-19

Example of Misinformation in Mothers’ Comments
Direct Presentation of Misinformation
“I’m not getting it [COVID-19 vaccine] & neither is my daughter. I don’t know what’s in it & what it will do to us. So I’m going to wait & see what happens.”
“I can’t get the vaccine and wouldn’t even if I was eligible. It’s a man-made virus and that doesn’t sit well with me.”
Indirect Presentation of Misinformation
“I wear a mask and have since I could get my hands on them in public but didn’t Fauci et al say this was unnecessary months ago? Rules keep changing on a whim. Is this evidence based? Who knows?
“I live in a farming county. We keep our immune systems healthy by eating right, exercising, getting plenty of rest, sunshine and fresh air. The older folks at risk started taking zinc, quecertin and elderberry tonic for added immune support.”
“I worry about issues for myself with diabetes and gastroparesis. My daughter also has PCOS, diabetes. I have major concerns for either of us to have the vaccine. We both had Covid in October 2020, it’s no laughing matter.”
Examples of Responses to Misinformation from Community Manager
Great question. Vaccines are not 100% effective. Some vaccines prevent infection, others prevent disease and some do both. So far, the Moderna and Pfizer vaccines prevent illness and death. Scientists are studying whether people who have been vaccinated can still spread COVID-19 which is why it is still recommended to wear a mask and social distance until we know more.
There has been a lot of chatter on the internet about this. There is no evidence that the vaccines affects fertility. The vaccine cannot give you the virus and it does not affect your genetics. We’ll continue to learn more as trials go on and more people are vaccinated.
Thank you for sharing your concerns. While it is unlikely vaccines will be mandated nationally, some individual institutions may require vaccines. However, this is not really a new practice. For example, hospitals mandate some vaccines, such as the influenza and Hepatitis B vaccines, for their staff.
We agree - there isn’t a way to prepare for loss and you know how best to talk with your daughter. We are hoping to offer tips on how to help families deal with feelings of grief, whether that is from the loss of a person or of experiences/routines, if they arise during the pandemic.

Most misinformation comments were in response to vaccination posts (n=40 comments, 74%), followed by NPIs (n=11, 20%), digital and media literacy (n=2, 4%), and family communication (n=1, 2%) posts. Among vaccination posts, misinformation themes included mothers declining to be vaccinated (n=13), adverse reactions to vaccine (n=6), vaccine was rushed to market (n=5), unknown side effects of vaccine (n=5), distrust of government, pharmaceutical companies, or media (n=5), lack of safety for pre-existing conditions/comorbidities (n=4) or pregnancy (n=3), fear of vaccine mandates (n=2), and other misinformation (n=5) (some comments addressed multiple themes). Misinformation relating to NPIs included conflicting messages regarding masks (n=6), harms of double masking (n=5), and other misinformation (n=5). Mothers in the near-peer parent group posted more misinformation comments (n=27; 50%) than mothers in the government agencies (n=17, 31.5%) and news media (n=10, 18.5%) groups.

The community manager responded to 48 (88.9%) misinformation comments (see examples in Table 1), excluding a few comments in which the same mother restated the same misinformation. Responses began by thanking the mothers for the comment, acknowledging validity of the content, opinion, or emotion they expressed (e.g., “It sounds like there are a variety of experiences so far;” “I get it that it can be confusing”), or stating a personal anecdote from the community manager (e.g., “I just started double masking when I have to go to the doctor’s office. It’s not too bad.”) to build empathy and prevent defensiveness. All responses then rebutted the misinformation with facts, highlighting where there was no evidence for the misinformation, reframing or contextualizing issues, and providing links to health authorities (e.g., CDC; American Diabetes Association; National Medical Association). Responses also identified gaps in current knowledge to build credibility, noted mothers’ responsibilities for children’s health, and encouraged mothers to talk with their medical providers. The community manager often asked a follow-up question in the response to promote dialogue. In a few cases, information countering misinformation were included in future campaign posts, too. After responses were posted, 10 original posters commented again (2 said thank you; 8 restated their position with 4 implying they understood the correction and 4 providing additional reasoning) and 3 reacted (i.e., “like”, “love”). A few mothers commented (n=2), reacted (n=14), or sent a private direct message to the moderator (n=2) related to the response. No mothers who provided misinformation comments left the Facebook groups.

Discussion

The social media campaign, despite its short duration, attracted misinformation. However, only a small number of posts (13%) elicited misinformation and very few participants (6.6%) shared misinformation. Posts on COVID-19 vaccinations were the most frequent targets of misinformation. The history of misinformation surrounding vaccinations before the pandemic may be one explanation.¹¹ Also, vaccines were being initially distributed during the study¹² and widely discussed on social media, perhaps highlighting vaccine misinformation. At times, mothers presented misinformation indirectly, perhaps to forestall negative reactions from others or to avoid appearing to violate community standards in the Facebook groups. Past participation in the first trial on indoor tanning may have made mothers strive to maintain a harmonious tone.

Our response protocol included acknowledging mother’s comments to validate mothers, show respect, and prevent debate, providing facts and evidence-based sources, and describing similar experiences by the community manager to show empathy⁴ and build an emotional connection.¹³ Responding quickly may be important to prevent misinformation from escalating; however, quick response may not always occur in practice. Social media managers may ignore, delete, or hide offending misinformation posts, or instead respond with a general message.¹³ This may prevent dialogue and risks offending the person sharing the misinformation, which may compromise opportunities for discussion. The community manager’s responses did not seem to generate extended debates with the mothers posting misinformation, possibly because they conveyed an understanding of mothers’ positions or their confusion.⁴ Avoiding debates may create a “safe space” in Facebook groups with trust and connection among all users, which may make correction messages more effective. However, correction style or tone had only small effects on debate in other studies.^14,15 Response messages that encourage more overt support for the rebuttal from other group members might reinforce the correction and create social norms around acceptance of public health advice. While users can be ambivalent about or unwilling to respond, being exposed to corrections might increase their willingness to correct misinformation.¹⁵

This study was limited by lack of assessment of participants’ reactions to misinformation and responses, participants’ roles as mothers of teens and limited gender and ethnic diversity, and short duration of the feed.

Implications for Policy and Practice

• Social media feeds on COVID-19 prevention can be targets for misinformation, which should be quickly corrected to prevent misinformation going viral.

• An epidemiological approach of constant monitoring, diagnosis, and quick response by a community manager may help to counter misinformation when using social media to convey information on COVID-19 prevention.

• Using message formats that forestall defensiveness and debate should maintain group cohesion and avoid turning off social media users and enable information that debunks misinformation to reach users and affect them.

• Misinformation correction methods that create a safe, respectful communication environment may be effective if they create a space where social media users can share scientifically valid information and express support for messages that correct misinformation.