How individuals cope with anger- and sadness-induced narrative misinformation on social media: roles of transportation and correction

2.1 Misinformation on social media

Misinformation is defined as false or misleading information based on the best available scientific evidence and expert knowledge at the time (Vraga and Bode 2020). The threat of misinformation can be heightened by health issues, such as unverified COVID-19 vaccine mandates, Zika Virus conspiracies, and fabricated opioid anecdotes. Recent research shows that misinformation spread online poses a greater threat than its offline counterpart, due to digital technologies and social media amplifying its reach and engagement (Peng et al. 2022; Chen et al. 2021a). A meta-analysis shows that health misperceptions, due to exposure to misinformation on social media, can generally be corrected (Walter et al. 2021). Corrective information can increase awareness surrounding the severity of a crisis, aiding in changing attitudes and emotions (van der Meer and Jin 2020).

However, in the social media context, emerging evidence has found the correction of narrative misinformation, such as personal stories, anecdotes, or testimonials, is less successful, compared to the correction of blatantly wrong claims (Zhao and Tsang 2024). Narrative misinformation is defined as false or misleading information that deviates from scientifically validated knowledge, conveyed through a narrative format consisting of a sequence of events and characters (Kreuter et al. 2007; Vraga and Bode 2020). One reason could be that social media users perceive fake testimonials and stories shared by like-minded others as more relatable and authentic, thereby lowering verification motivation. Another reason could be the emotional impact of narrative misinformation, as reliance on heuristic cues such as strong emotions can strengthen the impact of misinformation.

To enhance our understanding of individual coping mechanisms with narrative misinformation in the social media environment, this study introduces a theoretical model detailing responses to narrative misinformation that elicits anger and sadness, each with unique appraisal tendencies. We discuss narrative communication theories and Appraisal Tendency Framework (ATF) in the misinformation context as a basis for the model as follows.

2.2 Narrative misinformation and emotions

A narrative is ‘the telling of someone’s experience about something’ (Dahlstrom 2021, p. 2). A fundamental part of human communication, narratives are a popular approach to conveying health information and promoting healthy behavior change (Hinyard and Kreuter 2007). Extensive research has shown narratives to be powerful influences in health-related beliefs, attitudes, and behaviors (see Shen et al. 2015). However, just as narratives are utilized for communicating scientific health information, they can also be used to spread misinformation, disinformation, and conspiracy theories. Narratives are a common strategy for creating online health misinformation (Peng et al. 2022; Chen et al. 2021b). For example, in a content analysis of HPV vaccine-related posts on Instagram, Massey et al. (2020) found that anti-vaccine posts were more likely to include personal narratives than pro-vaccine posts.

The primary mechanism for narratives to persuade is through transporting readers to the narrative world, namely, drawing their attention to the story and evoking strong emotions (Green and Brock 2000; Moyer-Gusé 2008). Similarly, online narrative misinformation can induce transportation. This is because narrative misinformation tends to feature vivid and detailed descriptions that can capture attention and arouse strong emotions. The consequence of being transported into a story world is to adopt story-relevant beliefs, attitudes, and behaviors. For example, Ma and Yang (2022) found transportation was positively related to intentions to reduce alcohol use and seek information about alcohol use. Zhao and Tsang (2024) found that narrative (vs. statistical) misinformation led to greater transportation, and consequently, increased intentions to share the misinformation on social media.

Although emotions are a critical component of narrative transportation, they are often measured by asking participants to indicate their overall feelings (e.g., This story affected me emotionally). Based on the appraisal approach to emotions (Lazarus 1991), researchers have explored the role that discrete emotions play in narrative processing and effects (e.g., Banerjee and Greene 2012; Lillie et al. 2021). However, the impact of different emotions on misinformation processing and its behavioral outcomes remains unexplored.

2.3 Appraisal tendency framework: anger and sadness

The Appraisal-Tendency Framework (ATF) explains the effect of emotions on individual judgments and decision-making (Lerner et al. 2007). Namely, ATF suggests each emotion links to specific appraisal tendencies – or set of motivations and cognitive patterns. Appraisal tendencies then give rise to associated sets of judgments, actions, and behaviors (Lazarus 1991). For example, sadness is linked to situations where circumstances are seen as controlling the event and appraisals of helplessness and loss are present (Tiedens and Linton 2001). Anger, on the other hand, is tied to experiences where other individuals are seen as controlling the event and appraisals of certainty, control, and perceptions of unfairness are found (Lerner et al. 2007). While ATF addresses both positive and negative emotions, a large portion of misinformation is found to contain negative emotions, such as anger and sadness (Lee et al. 2022). Additionally, Han et al. (2020) suggest anger makes individuals more susceptible to misinformation as the emotion adds to a propensity to resolve a perceived unjust situation. Sadness-eliciting misinformation on social media was also found to increase attention and sharing efforts (Lee et al. 2022). Thus, the study examines anger and sadness as two discrete emotions that merit additional attention within narrative misinformation.

2.4 Appraisal tendency framework and health misinformation

Within existing ATF literature, negative emotions and associated appraisal tendencies have been found to play a critical role in the evaluation and understanding of health misinformation. Liu et al. (2021) examined the role of emotions in e-cigarette-related misinformation and found that anger mediated the relationship between exposure to misinformation and people’s perception of e-cigarette risks, making them more likely to believe that e-cigarettes were not harmful and reducing their perception of associated risks. Similarly, Šrol et al. (2021) found that anxiety increased people’s belief in COVID-19 misinformation, suggesting that different negative emotions can amplify the acceptance of misinformation.

Emotion heuristics – or mental shortcuts that encourage judgments based on emotions rather than logic – are often used to spread misinformation (Kata 2010). Thus, messages with misinformation often elicit more intense negative emotions than messages with accurate information (Paschen 2020). Ali et al. (2022) also found misinformation designed to discourage vaccination frequently utilized anger to motivate people to share and spread the misinformation. Since anger triggers appraisals that make individuals feel more certain about information accuracy (Lerner and Tiedens 2006), they are more likely to share misinformation on vaccination (Ali et al. 2022).

ATF has begun to be used to examine health behaviors and better understand misinformation processing, but limited research has applied the framework to narrative misinformation on social media. Further explorations into different discrete emotions, particularly the use of strong negative emotions such as sadness and anger within narratives offer opportunities to expand the literature.

3.1 Misinformation coping: roles of anger and sadness

Misinformation coping literature suggests individuals faced with misinformation turn to information-seeking as a method of verifying the information (Taddicken and Wolff 2020). Message truthfulness involves one’s judgment of content veracity (Zhao and Tsang 2024). Tandoc et al. (2018) suggested people often utilize multiple strategies to authenticate and determine if the message is truthful. Corrective intentions, such as debunking or fact-checking misinformation, are also useful in helping individuals develop new narratives around misperceptions (van der Meer and Jin 2020). For this study, the message content contains misinformation. The study thus considers the perceived truthfulness of misinformation and correction intention as cognitive and behavioral forms of misinformation coping.

In examining the role of discrete emotions within misinformation coping, ATF research finds anger to be an activating emotion linked to a tendency to take actions without systematic cognitive responses (Lazarus 1991). The misinformation literature also supports that anger can increase perceived message credibility (Han et al. 2020). This can reduce their likelihood of corrective actions, a key outcome in misinformation research defined as ‘outward behavior that seeks to rectify what is seen as flawed’ (Koo et al. 2021, p. 12). Following O’Keefe’s (2003) recommendation to analyze effect-based message variables (e.g., fear appeal) as mediating states rather than direct outcomes, we propose the following mediation hypothesis:

By contrast, sadness triggers more deliberative and detail-oriented information processing (Bago et al. 2022; Lerner and Tiedens 2006). Following this rationale, individuals viewing sadness-inducing misinformation might put additional thought into the perceived truthfulness of misinformation and become more apt in identifying and implementing corrections to counteract misinformation.

3.2 Risk coping: roles of anger and sadness

As health misinformation can result in the spread of disease or other potential harms, individuals have also been found to cope with misinformation by formulating perceptions of risk (Austin et al. 2023; Chang 2021). In assessing risk, individuals consider the prevalence, severity, and susceptibility of the suggested risk (Weinstein 1993). In other words, people first determine whether a threat is present before taking action through either treatment or prevention behaviors (Tandoc et al. 2018). This study considers risk perception and coping intention as cognitive and behavioral forms of risk coping.

Within the context of risk coping, the research on ATF suggests that angry people tend to rely on heuristics in judgment-making at the expense of systematic risk evaluation (Carnahan et al. 2023; Tiedens and Linton 2001), thereby leading individuals to underestimate potential risks (Slovic 1987). Anger also triggers appraisals that make people feel more certain regarding information accuracy, and thus, less likely to take precautions (Kim 2021). Therefore, anger-eliciting misinformation should mitigate risk coping by reducing risk perception and coping intention. We propose the following hypothesis.

Sadness, on the other hand, encourages problem-solving through inward contemplation (Nabi 1999). Sad individuals often focus on issue-relevant thinking and perceive higher levels of risk (Kim 2021), and risk perception can motivate coping intention (Weinstein 1993). We thus hypothesize:

3.3 Role of transportation

Transportation is defined as “a distinct mental process, an integrative melding of attention, imagery, and feelings” (Green and Brock 2000, p. 701). The consequences of transportation include losing access to the real-world facts, experiencing strong emotions, and adopting story-consistent beliefs and attitudes (Green and Brock 2000; Moyer-Gusé 2008). Therefore, we predict transportation will deepen the experiences of discrete emotions and the degrees of risk coping and misinformation coping in processing narrative misinformation. We hypothesize transportation following narrative misinformation exposure increases felt emotions, risk coping, and misinformation coping (see Figure 1).

3.4 Role of correction

Last, meta-analyses have found corrections, such as fact-checking or debunking, to be effective strategies for mitigating misinformation (Walter et al. 2021). After exposure to health misinformation, corrections can counter misperceptions by reducing misinformation credibility (Walter et al. 2021). As such, correction following narrative misinformation is expected to aid misinformation and risk coping, thus, we hypothesize:

4.1 Pilot studies

The first pilot was conducted in September 2022 to select two misinformation topics with emotional impact. Participants (N = 75) reported their emotions on 7-point scales for false Facebook posts on several topics. The results revealed two issues with the strongest emotional impact: the first focused on the impact of a climate change bill on the increase in grocery prices (abbreviated as climate change below), arousing anger (M = 3.95, SD = 0.21), and the second on alleged incidents of dollar bills covered with fentanyl causing unintentional overdoses (abbreviated as fentanyl overdose below), arousing sadness (M = 4.27, SD = 0.29). Specific measures of anger and sadness are included in the measurement section of the main study.

A second pilot (N = 64) was launched in September 2022 to test narrative versus non-narrative versions of the two topics and ensure narrative misinformation aroused significantly higher anger for climate change and higher sadness for fentanyl overdose when compared to non-narrative conditions. For climate change policy, narrative versions (M = 5.08, SD = 0.88) elicited more anger than the non-narrative versions (M = 3.36, SD = 1.78), t (39) = −3.89, p < 0.001. For opioid overdose, narrative versions (M = 5.27, SD = 1.62) elicited more sadness than the non-narrative versions (M = 4.05, SD = 1.76), t (40) = −2.34, p = 0.041. Results also confirmed that participants distinguished between a narrative post and a non-narrative post: t (36) = −4.52, p < 0.001 for climate change and t (36) = −2.20, p = 0.032 for opioid overdose.

A third pilot (N = 60), launched in October 2022, tested whether various stories of the same issue could arouse anger and sadness, respectively, to remove the potential confounding effect of the issue. For example, for climate change policy, a fake story about a single mother suffering from the policy was designed to elicit sadness, whereas a fake story about an angry consumer toward the policy was designed to elicit anger. Results showed various stories on the same issue generated both high anger and sadness, suggesting the two discrete emotions cannot be manipulated separately using the same issue. To ensure internal validity, we kept the climate change story of the angry consumer to primarily arouse anger and the fentanyl overdose story of a sad mom to primarily arouse sadness. The implications of this approach were discussed in the section on Limitations and Future Directions.

4.2 Main study

We conducted a 2 (Narrative: present vs. absent) × 2 (Topic: climate change policy vs. accidental fentanyl overdose) × 2 (Correction: present vs. absent) between-subject experiment in November 2022. A total of 444 participants were recruited. However, 43 participants were removed due to incomplete responses or failure to pass attention checks, yielding a final sample of 401 participants. Most respondents were male (49.4 %, n = 198) with roughly 73 % (n = 291) of participants between 18 and 40 years old. Approximately 13 % had at least a high school degree, 72 % had some college education or more, and the remainder had a graduate degree. The average household income fell between $50,000 and $59,9999.

Participants were randomly assigned to view Facebook post stimuli regarding a fabricated climate change policy or a fabricated fentanyl overdose (see Appendix A for stimuli). For the 200 participants viewing climate change stimuli, each one was randomly assigned to see an anger-arousing narrative post (experimental condition) or a non-narrative post (control condition). For the remaining 201 participants viewing the fentanyl stimuli, each one was randomly assigned to see one a sadness-arousing narrative post (experimental condition) or a non-narrative post (control condition). The stimuli creation was informed by all pilots to elicit the discrete emotions (for detail, see the section of Pilot Studies and Appendices). After message exposure, participants answered questions on their perceptions and behaviors related to the corresponding topic. Respondents were then randomly assigned to view a correction post or non-correction (i.e., click to proceed; see Appendix B) and reported message perceptions using similar measures worded in a slightly different way. Last, demographic information was collected.

6.1 Theoretical and practical implications

This study makes several theoretical contributions. First, the proposed theoretical model enriches the ATF by broadening its application to narrative misinformation on social media. Findings advance the ATF by demonstrating that narrative misinformation elicits anger and sadness with unique appraisal tendencies, influencing coping outcomes through distinct mechanisms and pathways. Appraisals associated with anger-laden narrative misinformation are more direct and favor behavioral coping measures including message correction and risk reduction. The lack of cognitive coping in this process suggests that anger can motivate individuals to adopt coping behaviors without systematic message evaluation. Additionally, appraisals associated with sadness-laden misinformation enhance misinformation coping by sharpening the ability to discern falsehoods. However, the role of sadness in risk coping might be affected by transportation, as people more immersed in fake stories become more vulnerable to misinformation by developing a higher risk perception of the fake hazard. These findings contribute to ATF by elucidating how narrative transportation complicates the interplay between distinct emotional appraisals and the varied responses to misinformation.

Second, it contributes to narrative persuasion research by proposing and testing a theoretical model of narrative misinformation coping. Our results show that narrative-based health misinformation can influence both risk coping and misinformation coping through narrative transportation and discrete emotions. In general, the more transported readers are, the greater emotions they feel, leading to a stronger need for risk coping. However, the impact of transportation on misinformation coping varies by topic. Transportation enhances perceived truthfulness and subsequently reduces the intention to correct misinformation about climate change; yet, for narratives about fentanyl overdose, transportation increases misinformation corrective intention. This finding indicates that misinformation embedded in a narrative format exerts influence similarly to true information presented in a narrative style. It further suggests that a promising way to curb narrative misinformation may be to disrupt narrative transportation experience, especially when it induces anger. This nuanced approach recognizes the distinct effects of specific emotions on narrative transportation within the realm of misinformation. Future research is needed to address this question.

Our findings carry important practical implications. Specifically, our results highlight that sadness decreases susceptibility to narrative misinformation’s negative effects, and anger prompts intuitive actions. These insights can inform responsible social media platforms and policymakers in enhancing posting guidelines. Additionally, our findings can contribute to the development of targeted policies aimed at mitigating online misinformation dissemination and provide a roadmap to executing effective correction measures. For example, practitioners developing correctives should emphasize the necessity of fact-checking, but also communicate proper levels of risk perception and appropriate coping behaviors.

This study also underscores the powerful impact of transportation and narratives. Our findings reveal that the narrative format of misinformation can magnify the negative effects of transportation and emotions on misinformation or risk coping. Health practitioners should consider interventions that help individuals better identify and mitigate these emotional and cognitive experiences. For example, since narrative misinformation is often more enjoyable and entertaining than non-narrative misinformation, practitioners could teach people to be cautious when encountering entertaining information as it might contain false claims. Moreover, practitioners could employ interventions that guide individuals to reflect on their emotional experience and regulate them on social media (Bago et al. 2022; Ma et al. 2024), which may help attenuate these emotional responses. Lastly, emerging research suggests that narratives can be effective in countering inaccurate information (Krishna and Amazeen 2022; Lillie et al. 2024). Health practitioners may consider using narrative refutations to address narrative misinformation (Lepoutre 2024; Shelby and Ernst 2013).

6.2 Limitations and future directions

Several limitations of this study should be taken into consideration. First, fabricated stories eliciting anger and sadness were used for the stimuli. People’s emotional states, however, rarely consist of a singular emotion. Future research should examine how individuals respond to narrative misinformation eliciting additional discrete emotions or even multiple emotions to map a comprehensive range of discrete emotions in the context of narrative misinformation processing. Second, even though our messaging highlighted the personalized impacts of both issues, the topics of climate change policy and accidental fentanyl overdose may confound the effects of narrative misinformation. Participants might engage with these issues to different extents, leading to varying levels of emotional reactions. Future research should include covariates such as issue involvement or employ additional issues to tease out any potential confounding effects. Another limitation of our study is the minimal variation in the self-other perceptual gap, which restricted our ability to explore its potential impact on risk coping behaviors. Future research should investigate this perceptual gap more deeply, particularly in the context of misinformation and its influence on public support for media censorship. Last, our cross-sectional online experiment does not reveal the longitudinal effect of narrative misinformation. The emotional impact of narrative misinformation might persist over an extended period, warranting future research on this aspect.