Fighting the Infodemic on Two Fronts: Reducing False Beliefs Without Increasing Polarization

Abstract

Actors aiming to remedy the effects of health misinformation often issue corrections focused on individual outcomes (i.e., promoting individual health behaviors) rather than societal outcomes (i.e., reducing issue polarization). Yet, for highly politicized health crises like the COVID-19 pandemic, such interventions run the risk of exacerbating societal cleavages, driving those holding opposing views further apart from one another. Interventions yielding individual benefits but causing societal harm are certainly not ideal. But is the design of such dual-focus corrections even possible? We believe this to be the case. Here, we delineate an agenda for future research that should help social scientists in identifying the characteristics of corrections that might reduce false beliefs without increasing polarization.

The novel coronavirus (COVID-19) pandemic has become one of the deadliest global disease outbreaks in a century, placing unprecedented strain on global health and economic systems. The pandemic’s sudden emergence has been met with an equally rapid proliferation of false and misleading claims (i.e., misinformation)—especially on social media platforms, like Facebook, YouTube, and Twitter (see Southwell et al., 2018; Suarez-Lledo & Alvarez-Galvez, 2021). Such claims have spread doubt on the virus’s origins, severity, and even existence while also discouraging preventive behaviors necessary to curb its spread (Barua et al., 2020). Recent research even linked COVID-19 misinformation to preventable death and hospitalization (Islam et al., 2020). Concerns about the impact of this misinformation at the individual and societal levels have been raised in response.

First, at the individual level, misinformation can increase the prevalence of false beliefs and promote attitudes and behaviors inconsistent with public health recommendations (Bridgman et al., 2020). It can reduce people’s intention to get a COVID-19 vaccine, for example (Kantar, 2020). Second, at the societal level, ¹ misinformation can exacerbate the gap between attitudes in society (i.e., issue polarization) and hostility among different-minded groups (i.e., affective polarization; Iyengar et al., 2012). COVID-19 misinformation, for instance, can cause people to arrive at different assessments of risk magnitude, thus fostering animosities between societal groups—be they liberals and conservatives, (information) haves and have-nots, or the young and the old (de Bruin et al., 2020; Hart et al., 2020; Motta et al., 2020; see #covidiot on Twitter). To the extent that they are considered to threaten identity, corrections could increase polarization (Hart & Nisbet, 2012). While these so-called backfire effects are less common than previously assumed, misperceptions often persist due to corrections’ influence decaying overtime or by being drowned out by competing views that are more consistent with people’s partisan or group identity (Nyhan, 2021).

The implication of all this is clear: Misinformation cannot be left unchallenged (see Southwell et al., 2018). In the COVID-19 pandemic, public health officials, politicians, fact checkers, and numerous other actors entered the fray in an attempt to fight the infodemic, defined as the viral spread of misinformation (World Health Organization, 2020). Currently, the main goal of such interventions, and the primary method in which their effectiveness is assessed, is by determining their capacity to reduce people’s false beliefs and adopt preventive behavior (see Swire-Thompson et al., 2020). Given the urgency of the ongoing health threat, the decision to concentrate on individual outcomes made sense. Yet, as we are entering the second year of the pandemic, an exclusive focus on this level is insufficient. Of particular concern is how corrections to misinformation may embolden individuals to engage in downward social comparisons against those holding differing views—thus, how they may motivate people to embark on a high horse, what we call a high-horse effect of corrections to misinformation. This might lead to greater division between those passionately holding COVID-19-related misperceptions and those with diametrically opposing viewpoints. This could occur when people feel extremely confident that they are in possession of the ultimate truth—a characteristic found in many who favor restrictive measures to curb the COVID-19 pandemic and in those who reject them. Corrections could then prompt defensive processing for some, and affirmation for others. But the result could call individuals’ attention to outgroup comparisons, which could weaken the opportunity for reducing COVID-19 polarization. Thus, strategies to correct COVID-19 misinformation should consider a dual-focus approach that improves accuracy of individual-level beliefs, behaviors, and attitudes while also taking into account societal-level polarization.

The High-Horse Effect and Polarization

To understand why corrections to misinformation might have unintended effects, it is helpful to put oneself in the situation of an individual holding false beliefs at the time the correction is encountered, and that of one holding accurate beliefs, respectively.

First, defensive processing is a likely outcome to a person hearing that what they think to be true is in fact false. After all, being informed that one’s belief is untrue often creates unpleasant psychological dissonance. In a highly politicized context like the COVID-19 pandemic, it seems unlikely that people would simply abandon their version of reality, especially when their reality is built to support a deep-rooted identity. Sticking to one’s truth is, cognitively speaking, easier and more pleasant (Major & Jankowski, 2020). Indeed, a person convinced to be in possession of the ultimate truth, but who is frequently informed that they are wrong, might engage in downward social comparisons with those who reject and challenge their beliefs. This may involve derogation of the source of the correction and, with it, all those who subscribe to the version of reality presented in the correction.

In contrast, how would those who never fell for misinformation but still received a correction ² feel about their fellow citizens who were less resilient? By less resilient, we mean both those who bought into misinformation in the first place and those who have done so and rejected the correction provided to them. In the current pandemic, for instance, it may not even have occurred to a person that drinking bleach does not cure COVID-19 until having encountered a corrective message stating so. Certainly, anxiety about the well-being of those falling for such claims would be a natural response. But, as social beings, people also have a natural tendency to compare themselves with one another (Suls & Wills, 1991). In this case, people might feel superior to those who fall victim to misinformation, perhaps even gloat over the naiveté of less resilient others—a process that may further divide an already polarized populace.

Having proponents of both sides of an argument embark on a “high horse” is certainly not a welcome development, as it can exacerbate already existing deep societal divides. It would be helpful if there would be more dialogue between these polarized groups. How then can corrections increase understanding for the other side and the way they perceive the world while also correcting the misguided belief? Careful attention must be paid to how to correct the individual-level effects of misinformation such that polarization is not increased. The aim of this article then is to make specific recommendations for future research seeking to identify how corrections should be designed to accomplish the dual focus of reducing false beliefs while without increasing polarization.

Two Propositions for Future Studies

Proposition 1: Increase the Appeal of Corrective Messages

Currently, the most common way in which health communicators attempt to correct misinformation is through factual elaboration on why something was wrong and by stating the facts—often in a fact-checking format (see van der Meer & Jin, 2020). We argue that other formats may be more courteous (and perceived as less offensive) toward those taken in by misinformation, thus potentially decreasing both these individuals’ reactance and others’ contempt toward them (see Hart et al., 2009).

We provide two specific suggestions with regard to alternative formats. First, instead of just-the-facts accounts, corrections should consider telling stories about likeable individuals who realize they fell for a false or misleading claim (see also Dan, 2021). A recent study by Sangalang et al. (2019) demonstrated that corrections employing this storytelling technique were very successful in correcting false beliefs. To our knowledge, no study to date has contrasted people’s views of the outgroup when the correction was provided as facts only versus using a human-interest story. However, related research on narratives suggests that a more lenient judgment will be passed on understanding the circumstances that led to a person’s gullibility (see Green & Donahue, 2018). Also, those in need of this correction may experience less reactance to it and resent the fact-checking entity less.

Our second suggestion concerns the use of visuals in corrections. Given that the human brain is attracted to images and hardwired to trust that which they record as reality (Coleman, 2010), it seems advisable to support corrective messages with visual cues and evidence (Dan, 2018, 2021). Existing research found verbal and visual corrections to be more enjoyable and more convincing than purely textual corrections (Amazeen et al., 2016; Garrett et al., 2013; Hameleers et al., 2020), especially those employing a video format (Young et al., 2018).

From the above, it follows that researchers should work toward the identification of such alternative formats and testing their effects on individual and societal outcomes. We would expect that video corrections in a human-interest format would yield best results. An indispensable component of these efforts are partnerships with fact-checking entities, as research must ensure that the alternatives tested meet these actors’ normative expectations and thus have a real-life relevance.

With regard to methodology, we recommend embracing selective-exposure designs (Knobloch-Westerwick, 2015). Indeed, while available misinformation research has a predilection for forced-exposure designs (Walter & Tukachinsky, 2020), we must factor in our designs the high probability that—considering the high number of stimuli competing for attention—people are unlikely to delve into corrections on all topics in their daily lives. This would allow scholars to test how to increase the appeal of corrections, for instance, through the use of different formats (see also Swire-Thompson et al., 2020).

Proposition 2: Join Forces With Scholars of Polarization

Research into how the noxious effects of health-related misinformation can be contained has been typically addressed by cognitive psychology and communication research (Swire-Thompson et al., 2020; Walter & Tukachinsky, 2020). In order to expand the focus to include societal effects, we argue that work in polarization, conducted mostly in political science, should be incorporated into the existing body of knowledge (e.g., Iyengar et al., 2012).

Such a collaboration with polarization scholars would have methodological ramifications, such that larger samples than usual would have to be used. This is because, when assessing the effects of corrections at the individual level, we focus on their ability to bring about mean change typically through a pretest-posttest design. For example, a study might record whether people’s level of agreement with a statement such as “Drinking bleach will cure those infected with COVID-19” can be reduced following exposure to a corrective message designed in a specific way. If pre-post mean change achieves statistical significance in the hypothesized direction, evidence would suggest that corrections might bring individual benefits (i.e., improved health attitudes and beliefs). However, in a study examining polarization, this type of insight would be insufficient. Rather, we would also be interested in uncovering evidence of potential societal-level effects (i.e., the distribution of values on a continuum). We would want to know whether the correction increased or decreased the number of people located at the extremes. Analyses of this type will require careful consideration of statistical power such that the potential effects of corrections on polarization can be reliably assessed.

Conclusion

In addition to contributing to social science by conducting basic research on effective ways to fight the infodemic on two fronts, the perspective suggested here would also deliver social value by substantially improving the state of knowledge in such a way as to enable a society that is better informed and less divided on issues related to public health. The insights of effective message design strategies will prove relevant to fact-checking entities. As the most efficient corrections are those that are delivered rapidly (Walter & Tukachinsky, 2020), and given that opinion on emerging issues tends to be highly malleable, building an arsenal of ready-to-use techniques will improve actors’ ability to act fast in fighting the infodemic in the current and future pandemics on both fronts.

Yet no one form of messaging acts as a panacea to misinformation surrounding this pandemic or future health crises. The problems that have surfaced during the pandemic may be symptoms of larger trends, some of which could have also been noticed before. For instance, some actors may choose to spread false and misleading claims while knowing very well that what they are stating is not accurate—perhaps to advance their own political agendas, consequences be damned. In any case, regardless of whether actors spread such claims knowingly (i.e., disinformation) or unknowingly (i.e., misinformation), as news consumers and social media users, we are obliged to work even harder to confront attempts of undue influence.