The Emotional Developments in Social Media: Conversations Following a Terror Attack

Emotions are an essential part of our communication. With the exceptions of guilt or shame, experiencing an emotion awakens a need to share it with others. The emergence of online social platforms during the recent decade has created new opportunities for such emotion sharing, and although research on expressions of emotion on social media has already yielded interesting insights, there are still gaps in our knowledge. Terror attacks are a particularly emotion-evoking type of event with heavy consequences on the emotional climate and well-being within a community. It is therefore a particularly interesting context in which to understand emotional mechanisms, and understanding them better can be used to inform actions that help members of the community overcome
their trauma better. In addition, a better understanding of online communication
dynamics will be helpful to emergency aid organizations in their attempt to filter relevant situational information in social media feeds during and after disasters. The research question posed and answered in this dissertation is How are different emotions expressed on social media in the wake of a terror attack?
This project started out by examining the existing wisdom on emotions on social media through a structured literature review. We found that theories are not extensively used or developed in the domain, that the most explored area of research is opinion mining, that the terms opinion and emotion are often used interchangeably, and that most studies study emotions through looking at polarity (positive vs. negative) rather than distinct emotion states.
One particularly interesting observation from the literature was that the findings
regarding the role of emotions in information sharing are not always in agreement with each other. It is not clear whether these differences arise from contextual or cultural particularities, or differences in how emotions are analyzed. Because of these differences, and because information sharing has been found to be the primary use of social media in disaster situations, we decided to take a closer look at the relationship between emotions and information sharing. Our next step was therefore to examine the relationship
between distinct state emotions and information sharing in an online conversation following a terror attack. We found that there were differences between negative emotions: fear and contempt were associated with lower retweeting rates, but anger and sadness did not have a notable correlation with retweeting. Positive emotions had a small positive correlation with elevated retweeting. It caught our attention that tweets from the affected area of the terror attack had a higher average of positive emotion and lower average of negative emotions than tweets from farther away: one would expect the opposite to be true for people who have recently been exposed to a traumatic event. Tweets from close by were also significantly more retweeted than other tweets. This prompted us to take a closer look into what people in different
proximity areas are talking about, and how emotions are associated to different
conversation streams. We analyzed the topics and emotions of three proximity regions over time following a terror event, and found that while some reactions are global, there are location specific collective emotions. As a result of this study, we propose a process model of the phases of post-terror online conversations, and outline how topics and emotions evolve during those phases.
The emotion expression on social media deviates from what literature says about emotions in two ways. Firstly, the theory on the social sharing of emotions states that the target of the sharing is a close person. However, emotion expression on social media is not targeted towards close people in specific, but to anyone who has access to the medium.
Secondly, the overall duration of the post-terror discussion online is shorter than the phase of frequent discussions outlined by the social stage model of coping. Establishing causality requires future research, but it seems that social practices related to emotion expression online differ from those that occur offline. This project contributes to our knowledge on how emotions are expressed on social media following a terror attack by finding that different emotions have different roles, that some of the collective emotional developments are specific to the proximity to the terror event location, and that emotional and topical trends are phase-specific, as well as proposes a model for the phases of post-terror online discussions.