key: cord-0733679-0zfi0spw authors: Vijaykumar, Santosh; Rogerson, Daniel; Jin, Yan; de Oliveira Costa, Mariella Silva title: Dynamics of Feedback Behaviours to Social Peers Sharing COVID-19 Misinformation on WhatsApp in Brazil date: 2021-10-21 journal: J Am Med Inform Assoc DOI: 10.1093/jamia/ocab219 sha: b26d374508b66d1a09fd37c53360f4151c5c995c doc_id: 733679 cord_uid: 0zfi0spw OBJECTIVE: Online COVID-19 misinformation is a serious concern in Brazil, home to the second largest WhatsApp user base and the second highest number of COVID-19 deaths. We examined the extent to which WhatsApp users might be willing to correct their peers who might share COVID-19 misinformation. MATERIALS AND METHODS: We conducted a cross-sectional online survey using Qualtrics among N = 726 Brazilian adults to identify the types of social correction behaviours (SCBs) and health and technological factors that shape the performance of these behaviours. RESULTS: Brazil’s WhatsApp users expressed medium to high levels of willingness to engage in SCBs. We discovered three modes of SCBs: correction to the group, correction to the sender only, and passive or no correction. WhatsApp users with lower levels of educational attainment and from younger age groups were less inclined to provide corrections. Lastly, perceived severity of COVID-19 and the ability to critically evaluate a message were positively associated with providing corrections to either the group or the sender. DISCUSSION: The demographic analyses point to the need to strengthen information literacy among population groups that are younger with lower levels of educational attainment. These efforts could facilitate individual-level contributions to the global fight against misinformation by the World Health Organisation in collaboration with member states, social media companies and civil society. CONCLUSION: Our study suggests that Brazil’s WhatsApp users might be willing to actively respond with feedback when exposed to COVID-19 misinformation by their peers on small world networks like WhatsApp groups. Online misinformation,defined as "any health-related claim of fact that is false based on current scientific consensus", [1] has posed barriers to the promotion of preventive behaviours and caused social unrest during the COVID-19 pandemic. This global problem has invited a range of responses from several stakeholders such as social media companies and the World Health Organisation to developers of misinformation games and news, media, and information literacy initiatives. [2] [3] [4] [5] However, what is less understood is the role that the vast population of 3.8 billion social media users [6] could themselves play in tackling the issue of misinformation. Our paper focuses on the role and extent of WhatsApp users' willingness to correct their social peers who might knowingly or unwittingly share COVID-19 misinformation on this popular messaging platform. We first present the conceptual ideas that informed our study, describe the study context, detail the methods, present our findings, and discuss their implications for theory, policy and practice surrounding the management of online health misinformation that has now returned to thwart COVID-19 vaccination programs. Online misinformation is corrected through a range of interventions including myth busting campaigns by public health agencies, algorithm-based news dissemination and by using expert organizations, journalists and fact-checkers. [7] [8] [9] Besides being as effective as algorithmic corrections in correcting misperceptions, [7] the limited evidence surrounding corrections made by social media users reveals the psychological and technological dynamics at play in influencing this behavior. WhatsApp users can either witness someone else being corrected, experience being corrected, or perform a correction themselves. [10] This study focuses solely on the performance of correction which we refer to as social correction behaviors (or SCB). We conceptualize SCB as the voluntary act of feeding back with a deliberate intent to counter perceived misinformation sent by a social media user within one's WhatsApp network. Given the nascence of this research area, we draw upon contemporary evidence and classic theories of health behaviour to identify social psychological factors that inform our conceptual framework underpinning this study. Research has shown that social media users ignore fake news posts and offer corrections only to those with whom they have strong relationships. [11] This suggests that the closeness of social ties in small networks like WhatsApp groups that are oftentimes comprised of family members or friends might provide a trusted environment, leading members to engage in social correction behaviours. However, verifying and correcting every misleading claim can be a time-consuming process. [12] While researchers have examined the relationship between time spent on social media and psychological health and well-being, [13] the extent to which it can contribute to positive behaviors like verifying and correcting information with peers is little understood. Another possible consideration affecting users' decisions around whether or not to provide corrective feedback to peers could be influenced by how social norms are perceived in small networks like WhatsApp groups. Specifically, social peers have been found to be more accepting of the expression of positive rather than negative emotions across Facebook, Instagram, WhatsApp and Twitter. [14] However, when these patterns were disambiguated by platform, expressing negative emotions was found to be most acceptable on WhatsApp (ibid). Placing this evidence in context, it is likely that social correction behaviours in a WhatsApp group might generate negative or positive emotions in the original sender and other members of the group could affect the decision of the user to perform such corrective behaviours. In addition to the abovementioned technological factors related to exposure to misinformation -critical evaluation of the message, time spent on social media, and the social norms of WhatsApp groups-we suggest that beliefs related to health risks might also affect SCB against misinformation. This argument is supported by extant evidence which shows that the perceived severity of, and perceived susceptibility to COVID-19 contribute to cyberchondria (compulsively seeking online information related to illness or symptoms) with males more likely to share news without verifying its accuracy. [15] Conversely, we are interested to find out if perceived severity of and susceptibility to COVID-19 may persuade WhatsApp users to issue social corrections to those spreading misinformation in their online groups. Lastly, while it is known that demographic factors like age, sex, income and education play a role in shaping misinformation beliefs [8 16 17] , the extent to which they affect SCB is unknown. The decision to perform SCB is thought to be based on a variety of factors such as cognitive and time costs of evaluating the veracity of information, normative perceptions about the appropriateness of such behaviour, and variations in our understanding of what constitutes as misinformation. These factors were reflected in decisions made by users including whether or not to make a social correction, when to make a social correction (now or later), and to whom to make a social correction (only to the sender or the group). Based on this framework, our study seeks to investigate the following questions. RQ1: How do WhatsApp users in Brazil engage in social correction behaviors? RQ2: How do these social correction behaviors differ by demographic factors? RQ3: How do factors related to misinformation consumption (exposure and beliefs), technology, and health beliefs jointly predict these social correction behaviors? We examine these questions in Brazil, a country with the second highest number of COVID-19 deaths globally. [18] Brazil also constitutes the second largest market with 146.6 million users. [19] The application allows members to create and subscribe to groups, which function as hives of information sharing among users. [20] However, its popularity has meant that WhatsApp has been prolifically misused not only for political propaganda during the 2018 national general elections but also as a channel for COVID-19 misinformation. [21 22] More than 70% of Brazilians believed in COVID-19 misinformation with WhatsApp being the main vector of misinformation in the country. [23] Consistent with incidents seen in other countries, misinformation circulating on social media sought legitimacy by false attribution to the Oswaldo Cruz Foundation (Fiocruz), one of the country's main public health institutions. [24] Subsequently, Fiocruz disseminated corrective information via online social networks to debunk false information attributed to itself and shared illustrated step-by-step guidelines to help people verify health information on online social networks. Other initiatives to combat COVID-19 misinformation by the Brazilian Health ministry included like the "Saúde sem Fake News' website -not updated since July 2020 -where a dedicated team of journalists would factcheck WhatsApp messages sent by users. [25] They found that messages categorized as false were usually about health authorities, prevention, prognosis of the disease, therapy and vaccination and that COVID- 19 lower human development index (HDI) found it more difficult to detect false information. [26] In correlated efforts to curb the effects of misinformation, a multi-center study was conducted to examine the impact of a digitally delivered intervention where medical students from 12 colleges demystified fake news and responded to COVID-19 questions from 375 elderly residents. The initiative highlighted how interpersonal communication facilitated by digital channels like WhatsApp could enable a greater understanding of the risks of misinformation among the elderly population. [26] While these programs suggest disparate institutional efforts to curb online COVID-19 misinformation, we are interested in the dynamics of individual-level actions undertaken to tackle this problem. In this cross-sectional study, we analysed a subset of questions enshrined within a larger dataset of a Portuguese-language online survey of WhatsApp users in Brazil, assessing their responses to COVID-19 misinformation and corrective information. As part of this online survey, a convenience sample of WhatsApp users from Brazil were recruited online by Qualtrics (a global professional survey firm headquartered in the United States) between 26 th May and 10 th June, 2020. All participants were required to be adult WhatsApp users (18 years of age or over), who had heard of COVID-19. In the survey questionnaire, we also asked participants about their sex, age, highest level of education attained, household income, and time spent using WhatsApp to discuss COVID-19 (see Table 1 in Findings section for detailed participant profiles). 19) does not spread in places with warm/hot weather", "You can protect yourself from coronavirus if you eat hot food or drink hot water", and "Hot pineapple water can cure coronavirus". Participants indicated 'Yes' or 'No' as to whether they had seen these messages on WhatsApp prior to taking part in the study, with more 'Yes' answers indicating greater exposure to the misinformation message. Participants were then asked to rate the accuracy of the five statements individually using a 5-point Likert-type scale ranging from 1= "Completely inaccurate" to 5= "Completely accurate". Given that the COVID-19 situation at the time of data collection was still rapidly evolving, we leveraged the local expertise of one of our co-authors, a health communication Health Beliefs: Perceived severity of and perceived susceptibility to COVID-19 were both measured using a three-item 5-point Likert scale adapted from Witte et al. (1996) , [27] ranging from 1= "Strongly disagree" to 5 = "Strongly agree". Specifically, perceived severity was measured with statements like "I believe Coronavirus (COVID-19) has serious negative consequences" (α = .69). Perceived susceptibility was measured with statements like "It is likely that I will get Coronavirus (COVID-19)" (α = .86). A self-structured three-item measure of information seeking on WhatsApp was created for this study (α = .82) and presented to participants to respond using a 5-point Likert scale, ranging from 1= "Strongly disagree" to 5 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 information on WhatsApp each day?", on a 5-point Likert-type scale: where 1 = "no time spent at all", 2 = "less than one hour", 3 = "between one and three hours", 4 = "between three and five hours" and 5 = "more than five hours". Points four and five were combined for analysis to maintain similar proportions of the sample in each of the categories (see table 3 for frequencies). Based on the conceptual framework, a self-structured 10-item scale 5-point Likert scale (1= "Strongly disagree" to 5= "Strongly agree") comprising different combinations of whether or not participants would send a correction, when they would send the correction and to whom they would send it were developed. For instance, one of the question stems said: "If you find that there is incorrect or fake COVID-19 misinformation in a WhatsApp forward you have just received you will…", with the following responses: inform the sender immediately / inform the sender after waiting a while / not inform the sender at all. A complete list of statements is available in Table 2 . Ethical approval for the study was granted by a research university in the United Kingdom. Participants were recruited online by Qualtrics through multiple platforms: social 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 media, e-mail invitations to propriety panels and online advertising (including online survey platforms). An anonymous link to the study survey was used in all recruitment methods. Participants received remuneration for their participation based on how they were recruited, some received points which could be redeemed for items whilst others were directly reimbursed the monetary value for participation. Participants were shown an information sheet outlining the nature of the study and informing them of what they would be required to do, if they consented. Those who did not consent to take part in the research were skipped to the end of the survey. Those who consented first provided demographic information. Participants then completed the survey in the order described in the Questionnaire section before being debriefed. Contact details for the principal researcher were included in the information and debrief sheet should participants need further information, prior, during or after their participation in the study. Participants took approximately 10-15 minutes to complete the survey. All data were analysed using SPSS version 27 [29] . Descriptive statistics were first calculated for the predictor and control variables in the study (see Table 1 ). Three participants who identified their sex as 'other' were excluded from the analysis. The consideration of gender minority communities was outside the scope of this study but merits future investigation. The final analysis thus accounted for only male and female respondents, with gender being treated as a continuous variable. [30] We performed a principal component analysis (PCA) using orthogonal rotation on the 10 SCB items (see Table 3 for results and new factors). Independent samples t-tests were then 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 used to determine any statistical differences between any of the control predictors with two levels (see Table 4 ) and a one-way independent groups ANOVA for control predictors with three levels (see Table 5 ) and the dependent variables. Predictor variables were then input into a correlation matrix (see supplementary information) to assess any multicollinearity issues. Finally, three hierarchical regressions (see Table 6 ) were run (one for each of our dependent variables) to ascertain which of the variables significantly contributed to the final models. We initially recruited 1100 participants of whom 197 did not provide consent, 102 were not WhatsApp users, and 162 were excluded as duplicates, incomplete datasets or for not completing the study within time parameters. Three participants who identified their sex as 'other' were excluded from the analysis due to marginal representation. The final analysis accounted for N=726 participants or 66% of the initially recruited sample. (Table 1) Our sample was predominantly male with a majority of the participants having completed at least an undergraduate degree (55%) and most (~40%) belonging to the middleincome bracket (R$3,000-6,999). Exposure to misinformation about COVID-19 not spreading in hot weather was the highest (65.3%) and the curative powers of pineapple was the lowest (17.6%). More than 7 in 10 participants spent fewer than three hours discussing COVID-19 on WhatsApp every day. Perceived severity of COVID-19 among the participants was higher than perceived susceptibility to being infected by it. (Table 2) Means, standard deviations and reliability scores for all scales are presented in Table 2 . Overall, we found low levels of misinformation belief among participants but high levels of perceived severity. All the scales used in the survey had high levels of reliability measured by Cronbach's  ranging from 0.82 to 0.90. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 and the relevant reliability and descriptive statistics for these factors. Values in bold indicate the best fit for each item on to the 3 factors. Our analyses revealed three significant differences. A significant effect of age on passive or no correction was found (p = .01): younger participants ( Notes: In the T-test statistic columns, t refers to the t statistic, p refers to significance of the tested difference <.05 denotes a significant difference (<.05* and <.01**), d refers to the Cohens D a way of measuring the size of the effect found. conducting more critical message evaluation and being more likely to send corrective information privately to the sender, were more likely to engage in correction to group as a ways of social correction. Sex was also found to be a significant predictor of correction to group: female participants reported a lesser preference to engage in group feedback than males (b = -.08, p = .03). Among the significant predictors, the standardised beta-weights suggest that correction to sender had the strongest relationship with group feedback, thus functioning as the strongest predictor for participants' behaviour surrounding correction to group. Correction to Sender: The final model was able to significantly account for 24% of the variance in correction to sender (R 2 = .24, p < .01). Despite education having a significant effect on correction to sender prior to the regression (see results for RQ2), education was not found to be a significant predictor on this SCB (b = .05, p=.15). Five of the remaining variables were found to be positive predictors of correction to sender: perceived severity (b = .14, p < .01), information seeking on WhatsApp (b = .21, p < .01), critical message evaluation (b = .10, p < .01), time spent discussing COVID-19 (b = .11, p = .01), and correction to group (b = .20, p < .01): participants perceiving higher COVID-19 severity, seeking more information on WhatsApp, conducting more critical message evaluation, spending more time discussing COVID-19, and more likely to send corrective information to a group, were more likely to engage in correction to sender. Passive or no correction was found to be a negative 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 predictor of correction to sender (b = .15, p < .01). Information seeking on WhatsApp was found to be the strongest predictor of correction to sender. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 misinformation), and Passive/No Correction. Our survey first revealed the pattern of how different demographics influenced the three types of SCB: first, younger participants exhibited greater passivity in engaging with social correction; second, higher educational attainment was associated with providing correction to the original sender; third, male participants were more likely in to sending the correction to the entire group. Information seeking and critical message evaluation significantly affected all three SCB types. Information seeking was positively associated with all three SCB types (i.e., the more individuals seek information about COVID-19 on WhatsApp, the more likely they are to engage in both active and passive SCB). Given the nascence of this research area, our study does not offer immediate explanations for this finding. We speculate that this association there might be related to the content of the COVID-19 information sought on WhatsApp or subsequent information processing, neither of which our study has not captured. Contrastingly, critical message evaluation is an important determinant of whether one engages in active (i.e. group and sender corrections) or passive SCB (i.e. no correction). Individuals who critically evaluate messages are thus more likely to correct misinformation via group-and/or sender-corrections on WhatsApp. We also found that each SCB has a distinct strongest predictor: namely, correction to sender for correction to group (positive association), information seeking for correction to sender (positive association), and perceived severity for passive/no correction (negative association). These findings suggest that encouraging users to actively seek (accurate) health information could make them more likely to send corrections to the sender which emerged as the "anchor" SCB -meaning, it connected and helped predict other SCB types in positive (Correction to Group) and negative (Passive/No Correction) directions, respectively. It seems that misinformation correction among Brazilian WhatsApp users is a cultivated process: those who are keen on correcting the sender alone might be encouraged or facilitated to share their corrections more publicly, in a group setting. In terms of passive SCB, we find that an elevated perception of the severity of the health risk might discourage users from engaging in SCBs. From a health risk communication standpoint, this finding highlights the importance of calibrating messages around the severity of health risks in a manner that is commensurate with the actual level of risk. Interestingly, while the predictive power of age and education seemed to be to correct people in their networks. [13] In terms of health beliefs, we found that perceived severity was a powerful predictor of all three feedback routes of social correction, based on our study among WhatsApp users. Campaigns to help enhance the perception of threat severity could help with social correction in small networks like WhatsApp. We found that critical message evaluation was important amongst the technological factors. Message evaluation, which is important in the primary information vetting process, [34] seemed to lead the user towards providing feedback. Because of this, our findings highlight the need to build critical thinking and informational literacy skills among social media users. Spending a greater amount of time on WhatsApp discussing COVID-19 was significantly associated with providing correction to the sender, and negatively associated with providing no correction. Therefore, it appears that the time spent deliberating over and discussing information received on WhatsApp facilitates peer-to-peer dialogue-based social correction. The private route, although not amplifying the correction in a group setting, might encourage the sender to consider vetting the information further (even revising their own original judgement on the information and themselves), thus initiating secondary information vetting[ibid]. Misinformation belief was significantly associated with "passive/no correction" but not with "active correction". This finding implies that those with high belief in ability to function as community-based misinformation watchers and alleviate the burden on fact-checking agencies. Equally, these fact-checking agencies could help develop community capacity in identifying misinformation, using online information verification tools, and generating source-based corrections that could be sent by users. [35] Our study has several limitations that constrain the generalisability of our findings and could be addressed by future research. First, the current study only studied looked at two elements of the health belief model (HBM), perceived severity and perceived susceptibility. Other HBM elements, such as perceived benefits and barriers of preventive action and perceived self-efficacy, needs to be further examined in future studies applying the full HBM 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 model to studies of managing the "infodemic" (spread of excessive and false information). [2] Second, this study was conducted at a single point in time. limited to a potentially arbitrary selection of misinformation messages found on social media. As the quality and nature of the content of each message differ, subsequent SCB may vary depending on the quality of content to which users are exposed. The frequency of exposure to such messages was not captured in the current study. Additionally, temporal factors (e.g., the time from exposure to content on social media and later peer interaction on social networking platforms like WhatsApp) that may influence peer interaction behaviors need to be investigated by future studies. Lastly, the definition of misinformation we employed -as any claim that is false based on current scientific consensus -must be consumed with caution for two reasons. One, it may be challenging to measure consensus among scientists at any one point in time. And two, if deployed loosely and prematurely, the consensus can be exclusionary of other less well-held perspectives that may be more evidentiarily robust and, in doing so, create new ground for misinformation. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 More than a year after the COVID-19 pandemic, online misinformation continues to impede the public health response to of vaccination programs by exacerbating vaccine hesitancy and denial through conspiracy theories and misleading claims about safety and side-effects. [37] [38] [39] In this scenario, when public health systems are stretched, sometimes to the limit, individual-level behaviors to combat misinformation could complement systemslevel infodemic management efforts. Our study adds new elements to these efforts by identifying factors related to health belief and social media use, along with key demographic factors and misinformation belief, as essential drivers that explain and predict infodemic management outcomes at the levels of the individual user and their social media community. Our findings also highlight the importance of activating all viable routes for misinformation correction, unlocking the power of social correction as amplifying source joining a conduit linking the influence of self-correction via information vetting and corrective communication and fact-checking efforts initiated by public health authorities and news media. We also note the responsibility of public and private health institutions to invest in information literacy initiatives by distilling complex scientific jargon into more accessible content and infuse a greater sense of personal responsibility in being vigilant to COVID-19 misinformation circulating on WhatsApp and other social networks. This work was supported by Facebook Inc. through the WhatsApp Social Science Misinformation Research Award. Authors have no competing interests to declare. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 SV led the conceptualization and writing. DTR led the methodology, statistical analysis and reporting of findings. YJ contributed to conceptualization and writing. MDC contributed to writing the manuscript This project was supported by the WhatsApp Social Science Misinformation Award. We thank all the participants for their time in responding to our survey, Ms. ES for assistance in formatting the manuscript and Mr. AM for feedback on the social correction behaviour survey items. The data underlying this article will be shared on reasonable request to the corresponding author. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Where we go from here: health misinformation on social media Framework for managing the COVID-19 infodemic: methods and results of an online, crowdsourced WHO technical consultation The fake news game: actively inoculating against the risk of misinformation Does media literacy help identification of fake news? Information literacy helps, but other literacies don't Major tech platforms say they're 'jointly combating fraud and misinformation' about COVID-19. Secondary Major tech platforms say they're 'jointly combating fraud and misinformation' about COVID-19 2020 Digital 2020: 3.8 Billion people use social media See something, say something: correction of global health misinformation on social media How shades of truth and age affect responses to COVID-19 (Mis) information: randomized survey experiment among WhatsApp users in UK and Brazil Creating news literacy messages to enhance expert corrections of misinformation on Twitter Dysfunctional information sharing on WhatsApp and Facebook: The role of political talk, crosscutting exposure and social corrections Diffusion of disinformation: How social media users respond to fake news and why Misinformation as a misunderstood challenge to public health When a story contradicts: correcting health misinformation on social media through different message formats and mechanisms. Information Norms of online expressions of emotion: Comparing Facebook What drives unverified information sharing and cyberchondria during the COVID-19 pandemic? An Examination of Factors Contributing to the Acceptance of Online Health Misinformation Vulnerable populations and misinformation: A mixed-methods approach to underserved older adults' online information assessment COVID-19 Coronavirus Pandemic. Secondary COVID-19 Coronavirus Pandemic WhatsApp in Brazil -statistics & facts. Secondary WhatsApp in Brazilstatistics & facts 2021 Mis) information dissemination in WhatsApp: Gathering, analyzing and countermeasures. The World Wide Web Conference Detection of Misinformation About COVID-19 in Brazilian Portuguese WhatsApp Messages. International Conference on Applications of Natural Language to Information Systems Using misinformation as a political weapon: COVID-19 and Bolsonaro in Brazil Pesquisa alerta que sete a cada dez brasileiros acreditam em fake news sobre Covid-19 Secondary Pesquisa alerta que sete a cada dez brasileiros acreditam em fake news sobre Covid-19 Fact or fake? An analysis of disinformation regarding the Covid-19 pandemic in Brazil Fake news no cenário da pandemia de Covid-19 Knowledge about COVID-19 in Brazil: cross-sectional web-based study Predicting risk behaviors: Development and validation of a diagnostic scale Adolescents' mediarelated cognitions and substance use in the context of parental and peer influences IBM SPSS Statistics for Windows Gender can be a continuous variable A Cocreational Approach to Social-Mediated Crisis Communication: Communicating Health Crises Strategically on Social Media. The Handbook of Strategic Communication: The Handbook of Strategic Communication Crisis information seeking and sharing (CISS): Scale development for measuring publics' communicative behavior in social-mediated public health crises Seeking formula for misinformation treatment in public health crises: The effects of corrective information type and source Information vetting as a key component in social-mediated crisis communication: An exploratory study to examine the initial conceptualization Secondary The fact-checking preferences of Africacheck.org's audiences in Nigeria, Kenya and South Africa A Promising but Difficult Domain: Complex Health-related Crises and Academic-Professional Collaboration What happens in Brazil? A pandemic of misinformation that culminates in an endless disease burden Measuring the impact of COVID-19 vaccine misinformation on vaccination intent in the UK and USA Social media and vaccine hesitancy: new updates for the era of COVID-19 and globalized infectious diseases