Analyzing the Digital Traces of Collective Emotions after a Terrorist Attack

class: center, middle, inverse, title-slide

.title[
# Analyzing the Digital Traces of Collective Emotions after a Terrorist Attack
]
.author[
### David Garcia <br> <em>ETH Zurich</em>
]
.date[
### Social Data Science
]

---

layout: true
<div class="my-footer"><span>David Garcia - Computational Social Science Lab - TU Graz + CSH Vienna</span></div>

---

## The Social Function of Collective Emotions

- **Theory of collective effervescence (Durkheim, 1912):**
Shared emotions generate social identity, reinforce shared beliefs, and lead to higher solidarity

- **Collective emotions (von Scheve and Salmela, 2014):**   
Emotional states shared by a large amount of people at the same time

---

background-size: 40%
background-image: url(figures/Trauma.png)
background-position: 90% 60%

## Emotional Responses to Collective Traumas

**How do societies respond to traumatic events?**  
**Is there a social function of collective emotions?**
.pull-left[
- Emotional synchronization: Experience of simultaneous negative emotions

- Collective emotion lasts longer than individual emotional reactions

- Individuals that participate in the collective emotions show higher levels of long-term solidarity ]

---

## Twitter Digital Traces after a Terrorist Attack

Focus on Paris Attacks of of Nov 13, 2015  
Removed bots, news media, and organizations. Final sample of 62,114 users  
Retrieved historical timeline of users. Total of more than 27 Million tweets (no RT)
<center>
<img src="figures/ParisCase.png" width="900" />

---

## Linguistic Signals

**Linguistic Inquiry and Word Count, LIWC (pronounced “Luke”)**  
- Simple word matching method  
- Validated by psychologists, years of implementation (2001-2015)  
- Multiple classes, calibrated for netspeak and neologisms

.pull-left[
<img src="figures/LIWC.png" width="1000" />
]
.pull-right[
**LIWC classes in this study:**  
Positive Affect, Negative Affect  
Anxiety, Sadness, Anger  
Social processes  
Prosocial terms (Frimer, et. al, 2014)  
French values (libert*, egalit*, fraternit*)
]

---

## Agent-Based Modelling of Collective Emotions

.pull-left[
<img src="figures/Framework.png" width="1000" />

ABM calibrated with experiment data
- φ: post-shock AR(1) term
- φ~0: subcritical response
- φ>0: critical, collective behavior
]

.pull-right[
<img src="figures/Simulation.png" width="450" />
<center>
Mean valence in simulations
</center>
]

<div style="font-size:14pt">
[The Dynamics of Emotions in Online Interaction. Garcia et al. Royal Society Open Science 3 (2016)](https://royalsocietypublishing.org/doi/full/10.1098/rsos.160059)

---

## Evidence of Collective Emotions
<center>
<img src="figures/TS.png" width="750" />
</center>

---

### Collective Dynamics of Social Resilience Indicators
<center>
<img src="figures/TS2.png" width="720" />
</center>

---

# Emotional Synchronization Effect

.pull-left[
<center>
<img src="figures/TS3.png" width="470" />
</center>
]

.pull-right[
Division of users into two groups based on their emotional expression two weeks after the attacks

Frequency of social process terms:
- Very similar before the attacks
- Strong difference after the attacks
- Difference lasts for months

Similar effect for prosocial terms and shared values terms
]

---

class:center

## The Vienna terrorist attack of Nov 2nd, 2020

---

class:center

## Austrian Emotions on Twitter

<div style="font-size:15pt"> 
Paper: [Dashboard of Sentiment in Austrian Social Media During COVID-19. Frontiers in Big Data. Pellert et al. (2020)](https://www.frontiersin.org/articles/10.3389/fdata.2020.00032/full)  
Dashboard: https://mpellert.at/covid19_monitor_austria
---

class:center

## Austrian Emotions in Der Standard Tickers

<img src="figures/Dashboard2.png" width="900" />
<div style="font-size:15pt"> 
Paper: [Dashboard of Sentiment in Austrian Social Media During COVID-19. Frontiers in Big Data. Pellert et al. (2020)](https://www.frontiersin.org/articles/10.3389/fdata.2020.00032/full)  
Dashboard: https://mpellert.at/covid19_monitor_austria

---

background-size: 50%
background-image: url(figures/OpenDoor.gif)
background-position: 90% 60%

## Shelter offerings on Twitter

.pull-left[
<img width=1000 src="https://pbs.twimg.com/media/EmAbRICXYAA0WqB?format=png&name=large">
- Number of tweets with hashtags similar to [#Schwedenplatztür](https://twitter.com/hashtag/schwedenplatzt%C3%BCr)
- More in Twitte thread:
https://twitter.com/dgarcia_eu/status/1324097342551842817
]

---

# Conclusions

- Terrorist attacks trigger collective emotions that we can observe online

- Terms related to social resilience increase after collective emotions

- Individuals expressing stronger emotions used on average more terms related to social processes, prosocial behavior, and shared values

- Collective emotions are not just venting, they can keep us together

- Negative effects: inter-group conflict, intolerance, short-term orientation...

- Online interactive visualization at: http://dgarcia.eu/ParisAttacks.html

[**Collective Emotions and Social Resilience in the Digital Traces After a Terrorist Attack. David Garcia Bernard Rimé. Psychological Science (2019)**](https://journals.sagepub.com/doi/full/10.1177/0956797619831964)