Effect of anger, anxiety, and sadness on the propagation scale of social media posts after natural disasters

https://doi.org/10.1016/j.ipm.2020.102313Get rights and content

Abstract

Social media is widely used for sharing disaster-related information following natural disasters. Drawing on negativity bias theory, integrated crisis mapping model, and arousal theory, this study characterized the emotional responses of the public and tested the way emotional factors and influential users (with high numbers of followers and activeness) affect the number of reposts. Results indicated that after unpredictable earthquakes, the public showed negative responses, and negativity bias theory manifested especially when the posts came from influential users. During a typhoon or earthquake, the number of reposts grew as the number of anger-related words in posts increased. Anxiety- and typhoon-related posts from users with high numbers of followers negatively affected the number of reposts, whereas sadness-related posts had contrasting effects. These findings can help emergency managers formulate proper emotional response strategies after various natural calamities and help researchers test the abovementioned theories or models using real-word data from social media.

Introduction

Social media is widely used to share ideas and express emotions following disasters (Chung & Zeng, 2018; Ghafarian & Yazdi, 2020; Wei, Bu & Liang, 2012). Learning the source of information and studying how the public responds to extreme events is vital for emergency responders in reducing damage or improving disaster relief processes (Fathi, Thom, Koch, Ertl & Fiedrich, 2019; Li, Wang, Gao & Shi, 2017). Understanding key factors that influence the virality of disaster-related information is critical for emergency managers or decision-makers, as such data help them obtain knowledge on disaster response strategies and identify influential users (R. Chen, 2014; R. Chen & Sakamoto, 2013, 2014).

People show different emotional responses and information-sharing behavior during disasters (Lachlan, Spence & Eith, 2007; Jin, Pang & Cameron, 2012). For predictable disasters, information is likely to be transmitted to potential victims before the crisis happens (Lachlan et al., 2007). By contrast, information about an unpredictable crisis may lead to negative outcomes and threaten the safety of the public (H. Daramola, Oni, Ogundele & Adesanya, 2016; Park, 2017). Under unpredictable social conditions, the public shows affiliative behavior to cope with the challenges (Seyfarth & Cheney, 2013). Additionally, the public may feel anxious. People tend to share information frequently and extract meaning from the information to cope with the uncertainty caused by unpredictable changes (Oh, Kwon & Rao, 2010).

Emotional factors, such as sentiment (being positive or negative), significantly affect the reposted amount of disaster-related information on social media (Grabe & Kamhawi, 2006; Hornik, Shaanan Satchi, Cesareo & Pastore, 2015; Wu & Shen, 2015; Yoon, Kim, Kim & Song, 2016). For example, news with negative sentiments is more likely to be retweeted than news with positive sentiments (Wu & Shen, 2015). Other studies suggest that positive content may have more reposts than negative content (Jonah Berger & Milkman, 2012; Wojnicki & Godes, 2008).

In addition to sentiment, discrete negative emotions and their arousal level (activation level) can affect the virality of social media content on the basis of arousal and negativity bias theories (Jonah Berger & Milkman, 2012; J Berger & Milkman, 2010; Chen, 2014; Grabe & Kamhawi, 2006; Hornik et al., 2015). In the case of natural disasters, the public mainly expresses anger, anxiety, and sadness (Jin, 2009; Jin et al., 2012). Among these negative emotions, anger and anxiety are characterized by heightened arousal or activation, whereas sadness is characterized by low arousal or deactivation, as suggested by arousal theory (Feldman Barrett & Russell, 1998; Jonah Berger & Milkman, 2012). Anger and anxiety may have different effects. The cognitive functional model (CFM) by Nabi (2003) argues that anger can enhance deep information processing, whereas Cameron and Kim (2011) point out that people with anxiety may avoid the perceived dangers that may cause harm (Cameron & Kim, 2011; Nabi, 2003). Low arousal and certainty emotion of sadness lead to shallow information processing, but people may express increasing sadness through posting and reposting supportive information after natural disasters, especially after a predictable crisis (Cameron & Kim, 2011; Jin et al., 2012; Li et al., 2017; Jonah Berger & Milkman, 2012).

Social media users with high numbers of followers are key information sources (Wu & Shen, 2015; Riquelme & González-Cantergiani, 2016). They can influence and induce others to repost or comment on their posts and are thus known as “influential users” (Hou, Huang & Zhang, 2015; Kwon, Cha, Jung, Chen & Wang, 2013; Remy, Pervin, Toriumi & Takeda, 2013; Starbird & Palen, 2012; Wu & Shen, 2015). Moreover, the activeness of users is one perspective representing the influence of users (Riquelme & González-Cantergiani, 2016). If users actively participate in information propagation on social media, then their posts are likely seen and reposted by others (Hu, Zhang & Wei, 2019; Xu et al., 2018). Thus, we define users with high numbers of followers and posts as influential during natural disasters. Then, we test how these users can affect the propagation scale of disaster-related information.

Influential users tend to express intense emotions of fear, anger, disgust, and sadness (Kanavos, Perikos, Hatzilygeroudis & Tsakalidis, 2018). The emotion of trust distinguishes influential users from others, whereas anger and fear significantly contribute to user influence (Chung & Zeng, 2018). Therefore, on the basis of the significant effect of influential users on the popularity of information, this study verifies how the influence of users moderates the effect of emotional factors on the popularity of information on social media after natural disasters.

A review of recent literature reveals that no study has characterized public emotional response according to different types of natural disasters. Few studies have described how influential users moderate the effect of emotional factors on the propagation scale of disaster-related information based on large-scale, real-time social media data. However, understanding how emotional factors and influential users affect the transmission of disaster-related information on social media helps authorities learn the main emotions of the public, including attributions, attitudes, and behavioral intentions, and how to publish disaster-related information or provide emotional support through influential users (E. J. Lee & Kim, 2014; C. Zhang, Fan, Yao, Hu & Mostafavi, 2019). Therefore, this study fills these research gaps by answering the following research questions:

  • RQ1: What are the emotional responses of social media users after predictable and unpredictable natural disasters?

  • RQ2: Does the sentiment of the post and discrete emotions affect the reposted amount of the disaster-related information?

  • RQ3: How do influential users affect the outcome of emotional factors on the reposted amount of disaster-related information?

  • RQ4: Do differences exist among the effects of emotional factors and influential users on the propagation scale of disaster-related information after predictable and unpredictable disasters?

This study characterizes the effects of emotional factors and influential users on the number of reposts of social media content after two natural disasters. Then, this study advances the integrated crisis management (ICM) model by empirically examining the emotional responses of the public and how the emotional factors affect the number of reposts of social media content during natural disasters (Jin et al., 2012). On the basis of negativity bias and arousal theory, this study assumes that social media posts with negative emotions are more popular than posts with positive emotions. Similarly, social media posts with high-arousal emotions are more popular than posts with low-arousal emotions. By defining influential users as having a high following and activeness, we further test how these users affect the number of reposts and its moderating effect on emotional factors. Regression analysis shows that the effect of anger on the number of reposts about earthquakes and predictable typhoons are similar. By contrast, the anxiety-related posts of users with a high following have more reposts during an earthquake than a predictable typhoon. Moreover, sadness-related posts about predictable and unpredictable disasters present different effects on the number of reposts. In addition, this study provides practical implications by identifying the specific effect of negative emotions and influential users on the number of reposts. Thus, results of the study can help authorities understand the emotional response of the public to different natural disasters. Subsequently, authorities can construct proper emotional support and strategies to alleviate negative emotions.

The remainder of this paper is organized as follows. Section 2 shows the theoretical background and hypotheses. Section 3 discusses the datasets, preprocessing of data, extraction of hypothesis-related variables, and summary statistics of all the variables. Section 4 expounds the regression analysis that is used to test the proposed hypotheses and describe the results. Section 5 concludes the study. The theoretical and managerial implications as well as the limitations of this study are summarized in Section 6.

Section snippets

Emotional appeal and information-sharing behavior

Emotional appeal is a method of persuasion designed to create an emotional response to a message by using emotional content (e.g., horror movie, sad story, triumphant music, etc.). Emotional appeal effectively persuades an individual with little motivation or low ability to process a message cognitively (Petty and Cacioppo, 1986). It deals with the ways in which emotional responses are triggered as a function of the motivational relevance of a message to individuals (J. Lee & Hong, 2016;

Two natural disasters

To test how emotional factors and influential users affect the number of reposts of disaster-related information, this study collected data from two disasters, namely, the Mangosteen typhoon and the Yiliang earthquake. These disasters were chosen because both of them are natural disasters. One is predictable, and the other one is unpredictable. As such, the different emotional reactions of the public after two different types of natural disasters can be identified.

Typhoon Mangosteen hit

The emotional responses of the public after different natural disasters

To reveal the different emotional responses of the public on two different natural disasters, we plotted the positive and negative emotional words proportions in Fig. 1, which we extracted from LIWC. For three commonly expressed negative emotions, such as anger, anxiety, and sadness, we plotted their amount by date in Fig. 1.

As shown in Fig. 1(A), Typhoon Mangosteen was generated on September 7, 2018 and made landfall in Guangdong on the September 16. The emotional response on the day before

Theoretical implications

This study contributes to the extant literature on ICM in characterizing the emotional responses of the public after disasters. By characterizing such response, this study reveals the rationality and limitation of the ICM. The ICM reasonably reveals the emotional responses of the public after the occurrence of predictable and unpredictable natural disasters. However, the ICM model fails to explain the emotional responses of the public before the occurrence of the predictable typhoon. Therefore,

Conclusions

Drawing from negativity bias theory, CFM, ICM, and arousal theory, this study characterizes the emotional responses of social media users and verifies how emotional factors affect the number of reposts of social media content after two natural disasters (predictable and unpredictable disasters). In addition, results from defining the influential users as those with many followers and high activity users and then characterizing how they affect the number of reposts after natural disasters

Acknowledgements

The authors thank the constructive comments and suggestions from the editors and anonymous reviewers. This work was supported by the key project of National Social Science Foundation of China Grant No 15AZZ002, the National Natural Science Foundation of China (NSFC) Grant Nos. 71972164 and 71672163, the Guangdong Provincial Natural Science Foundation No. 2020A1515011217. This work was also supported by the National Social Science Major Special Foundation of China under Grant 17VZL017.

References (84)

  • Xiaolin Lin et al.

    Examining gender differences in people's information-sharing decisions on social networking sites

    International Journal of Information Management

    (2020)
  • H. Park

    Exploring effective crisis response strategies

    Public Relations Review

    (2017)
  • F. Riquelme et al.

    Measuring user influence on Twitter : A survey

    Information Processing and Management

    (2016)
  • Y. Wang et al.

    Attitude of the Chinese public toward off-site construction: A text mining study

    Journal of Cleaner Production

    (2019)
  • J. Wei et al.

    Estimating the diffusion models of crisis information in micro blog

    Journal of Informetrics

    (2012)
  • F.A. White et al.

    Emotion expression and intergroup bias reduction between Muslims and Christians: Long-term Internet contact

    Computers in Human Behavior

    (2015)
  • B. Wu et al.

    Analyzing and predicting news popularity on Twitter

    International Journal of Information Management

    (2015)
  • J. Xu et al.

    Empirical and modeling studies of WeChat information dissemination

    Physica A: Statistical Mechanics and Its Applications

    (2018)
  • H.G. Yoon et al.

    Opinion polarity detection in Twitter data combining shrinkage regression and topic modeling

    Journal of Informetrics

    (2016)
  • C. Zhang et al.

    Social media for intelligent public information and warning in disasters: An interdisciplinary review

    International Journal of Information Management

    (2019)
  • Y. Zhang et al.

    Research note: Examining gender emotional differences in Web forum communication

    Decision Support Systems

    (2013)
  • E. Bakshy et al.

    Everyone's an influencer: Quantifying influence on twitter

  • E. Bakshy et al.

    The role of social networks in information diffusion, 10

    Physics and Society

    (2012)
  • E.O.C. Bantum et al.

    Evaluating the validity of computerized content analysis programs for identification of emotional expression in cancer narratives

    Psychological Assessment

    (2009)
  • A. Beasley et al.

    Emotional states vs. emotional words in social media

  • J. Berger et al.

    Social transmission, emotion, and the virality of online content

    Wharton Research Paper

    (2010)
  • Jonah Berger et al.

    What makes online content go viral?

    Journal of Marketing Research

    (2012)
  • A. Bilgihan et al.

    Generation Y's dining information seeking and sharing behavior on social networking sites

    International Journal of Contemporary Hospitality Management

    (2014)
  • P. Bogdanov et al.

    Modeling individual topic-specific behavior and influence backbone networks in social media

    Social Network Analysis and Mining

    (2014)
  • P. Burnap et al.

    Tweeting the terror: Modelling the social media reaction to the Woolwich terrorist attack

    Social Network Analysis and Mining

    (2014)
  • G.T. Cameron et al.

    Emotions matter in crisis: The role of anger and sadness in the publics’ response to crisis news framing and corporate crisis response

    Communication Research

    (2011)
  • S. Chai et al.

    Factors affecting bloggers’ knowledge sharing: an investigation across gender

    Journal of Management Information Systems

    (2011)
  • R. Chen

    Sharing of crisis information in social media: The roles of distance, perspective-taking, and feelings[d]

    (2014)
  • R. Chen et al.

    Perspective matters: Sharing of crisis information in social media

  • R. Chen et al.

    Feelings and perspective matter: Sharing of crisis information in social media

    Proceedings of the Annual Hawaii International Conference on System Sciences

    (2014)
  • Q. Cheng et al.

    Assessing suicide risk and emotional distress in Chinese social media: A text mining and machine learning study

    Journal of Medical Internet Research

    (2017)
  • W. Chung et al.

    Dissecting emotion and user influence in social media communities: An interaction modeling approach

    Information and Management

    (2018)
  • T.M. De los Santos et al.

    Emotionally charged: Exploring the role of emotion in online news information seeking and processing

    Journal of Broadcasting and Electronic Media

    (2019)
  • S. Deng et al.

    The interaction between microblog sentiment and stock return: An empirical examination

    MIS quarterly

    (2018)
  • E. Diener et al.

    The personality structure of affect

    Journal of Personality and Social Psychology

    (1995)
  • R. Fan et al.

    Anger is more influential than joy: Sentiment correlation in Weibo

    PloS one

    (2014)
  • R. Fathi et al.

    VOST: A case study in voluntary digital participation for collaborative emergency management

    Information Processing and Management

    (2019)
  • Cited by (52)

    View all citing articles on Scopus
    View full text