October 29, 2018

Violence and Social Media






When do frames work/fail?

Frame success and violence

Social Media and Violence

Does social media cause violence?

Keep in mind:

1. Coordination vs. Persuasion

2. What makes frames work:

  • credibility (believable; credible messenger)

  • salience


Difficult to say…

1) Do social media frames change enough?

2) Delayed effects of social media?

3) Selection into "Treatment"

Evidence from German Facebook

Mueller and Schwarz

Does social-media hate speech translate into real-world behavior?

Anti-Refugee content on Facebook and Anti-Refugee violence

  • Alternative für Deutschland (AfD) Facebook Page

Mueller and Schwarz

Does AfD Facebook content drive anti-immigrant violence?

If anti-refugee sentiment spreads through Facebook, we would expect hate crimes to be more likely to occur in municipalities with higher exposure to social media

Mueller and Schwarz: Data

Anti-Refugee Violence

  • 3335 events from January 2015 to early 2017
  • Reported in media outlets
  • property damage, assault, arson, demonstrations

Mueller and Schwarz: Data

Facebook anti-Refugee rhetoric

  • ~200K posts on AfD page
  • use word "Refugee"

Mueller and Schwarz: Data

Facebook Exposure

  • ~90K users on AfD Facebook page; locations for 40K
  • ~63K users on Nutella Germany Facebook page; locations for 20K

Nutella captures Facebook exposure, without selection on far-right politics

Mueller and Schwarz: Design

(1) Compare municipalities to themselves over time

  • "Control" for national trends; unchanging attributes of cities

(2) "Treatment" of Facebook Rhetoric same for all

(3) Exposure to "Treatment" varies with Facebook users

Mueller and Schwarz: Design

Do cities with more Facebook users have greater increases in violence in weeks with anti-immigrant Facebook activity compared to weeks without anti-immigrant Facebook activity than cities with fewer facebook users?

Mueller and Schwarz: Results

Compared to Frankfurt (6.5 AfD Facebook Users/10,000), Dresden (9.6 AfD Facebook Users/10,000) estimated to have 50% more attacks on refugees in a week with average anti-refugee Facebook posts.

Mueller and Schwarz: Results

Results the same

  • Looking at Nutella Facebook users
  • Comparing cities in same county in the same week

Mueller and Schwarz: Results

  • Effects go away when places have internet disruptions/Facebook issues
  • Effects only significant for anti-Refugee rhetoric (as opposed to other topics) on AfD page
  • Effects reduced when competing news (Brexit, Trump, UEFA)
  • Effects of social media persist even when accounting for "traditional" media coverage

Mueller and Schwarz: Interpretation

Likely that social media exposure drives anto-refugee violence

Is it persuasion or coordination?