Hate in Context

Week 4: (Social) Media and Hate Crime · LENS preview

Violeta Haas

Department of Political Science

School of Social Sciences and Philosophy

Trinity College Dublin

This Week

Catch-up — Week 3 policy pitch
Week 04 – Siegel (2020)
Offline consequences — Müller & Schwarz (2023)
Combating online hate speech
Outlook

Catch-up — Week 3 policy pitch

Policy advice (10-15min)

Scenario. The government has funding for precautionary measures to avoid spikes in hate crime. They have become aware of research linking economic conditions to bias-motivated crime and they are worried about AI-driven job replacement.

You are one of three advocacy organisations competing for the contract:

Institute for Intergroup Safety (IIS)
Alliance Against Bias Crime (AABC)
Coalition for Hate Crime Prevention (CHCP)

Your task (3 Groups)

Design a short policy brief: what to worry about, what to fund, and why it should work. Draw on our readings.
Each group decide on one representative to pitch your brief (2 minutes).
Vote on who gets the job!

A policy brief usually has…

Title
Executive summary / recommendation
Problem / context
Evidence
Options
Preferred option
Implementation
Risks & caveats
Next steps

Collect notes (5 min)

If you were here last week

Groups: IIS, AABC, CHCP

Collect your notes from the policy brief.
Decide who will pitch (2 minutes).

If you missed last week

While others prep, work on Siegel (2020) — Online hate speech:

What do we know about the producers of hate speech?
What do we know about the targets of hate speech?
How prevalent is hate speech?

Institute for Intergroup Safety
(IIS)

Alliance Against Bias Crime
(AABC)

Coalition for Hate Crime Prevention
(CHCP)

Who gets the job?

Institute for Intergroup Safety (IIS)
Alliance Against Bias Crime (AABC)
Coalition for Hate Crime Prevention (CHCP)

You can vote using the QR code or here:

on Blackboard → Week 03 → POU44653 Week 3: Who gets the job?

Week 04 – Siegel (2020)

Defining Hate Speech

Siegel (2020): there is no single agreed definition, definitions range from maximal to minimal.

Maximal (broad)

[discussed in class]

Minimal (narrow)

[discussed in class]

Typical producers & targets

Typical producers

[discussed in class]

Typical targets

[discussed in class]

Prevalence of online hate speech

On platforms (content)

[discussed in class]

In surveys (exposure)

[discussed in class]

Offline consequences

Müller & Schwarz (2023)

Müller and Schwarz 2023 Figure 1: Hate crimes and Twitter usage by US county

Question: Can social media amplify antiminority sentiment?

Setting:

Where: US counties
When: around Trump’s political rise (especially after the 2016 primaries)
Outcome: anti-Muslim hate crimes (FBI UCR)
Exposure: county-level Twitter usage

Additional data sources

Hate groups (SPLC)

Annual census of active US hate and antigovernment groups (Hate Map)
Tracks organizations, not individuals
Based on publications, reports, field sources, and investigations

Victimization (NCVS)

Bureau of Justice Statistics’ main national survey of criminal victimization
~240,000 people in ~150,000 households each year
Covers crimes reported and not reported to police

Global Terrorism Database

START (University of Maryland) open-source event database
200,000+ domestic and international terrorist attacks since 1970
Date, location, weapons, targets, casualties, perpetrators

ADL hate/incident data

Tracks bias incidents of harassment, vandalism, and assault
Includes criminal and non-criminal incidents
Drawn from victims, law enforcement, media, and partner reports

Motivation: hate crime trends

Figure 3

Anti-Muslim hate crimes rise sharply with the Trump campaign period. Total hate crimes less so until later.

Findings: Twitter and hate crimes

Did higher county Twitter use raise anti-Muslim hate crimes after Trump’s rise?

Figure 4

Figure 6

Same pattern for raw Twitter use (Fig. 4) and the SXSW instrument (Fig. 6): rise after 2015, flat before.

SXSW as instrument

South by Southwest is an annual festival/conference in Austin. The March 2007 event was a tipping point for Twitter’s early US growth.

How they use it: Counties of people who started following SXSW on Twitter around the 2007 festival predict later county Twitter use. Controlling for pre-festival SXSW followers addresses selection into interest in Austin/SXSW.

Heterogeneous effects

Does Twitter amplify preexisting local hatred?

Figure 7

The Twitter effect concentrates where hate groups already exist (Panel A) and where hate crime was already high (Panel B).

Did Trump actively contribute to the spread?

Do Trump’s Muslim tweets trigger anti-Muslim hate crimes?

Figure 8

Trump’s Islam-related tweets and anti-Muslim hate crimes move together over the campaign period.

Identification: golf days

Figure 9

On golf days: more tweets about Muslims (Panel A), but not more tweets overall (Panel B). Instrument for content, not volume.

Discussion (2 min): golf days

President Trump golfing (White House photo)

Turn to a neighbor (≈2 min). Discuss:

Why might tweet content change on golf days?
What alternative explanations would threaten the exclusion restriction (golf → hate crime only through Muslim tweets)?

Discussion: answers

[discussed in class]

Identification: golf days (estimates)

Figure 10

Müller and Schwarz 2023 Figure 10 Panel C: Reduced form

OLS: hate crimes rise after Muslim tweets (Panel A). First stage: golf days spike Muslim tweets (Panel B). Reduced form: golf → hate crimes (Panel C).

Spillovers to followers

Do Trump’s tweets spill over among his followers?

Figure 11

Müller and Schwarz 2023 Figure 11 Panel C: Anti-Muslim tweets by Trump followers

Trump’s Muslim tweets spike retweets (Panel A) and new Muslim tweets (Panel B) the same day; anti-Muslim hashtags concentrate among his followers (Panel C).

Mechanism: the news cycle

Do Trump’s tweets also shape the cable news cycle?

Table 6

Trump’s Muslim tweets raise cable-news mentions of Muslims the same day, especially on Fox News.

→ Xenophobic rhetoric that travels from Twitter into (often uncritical) broadcast coverage is one suggested trigger for hate crime.

Critique

Anything else that didn’t convince you?

Combating online hate speech

Group Discussion: How to combat online hate?

Groups of 3–4 (≈5 min). For each approach, collect 2 points in favor and 2 drawbacks. Then: what do you think would be a useful way to deal with hate speech online?

Approach A: Content moderation

Ban accounts/communities; enforce platform rules (and, where relevant, legal standards).

Approach B: Counter-speech

Challenge or dilute hate with opposing messages (users, campaigns, or designed interventions).

How to combat online hate?

[discussed in class]

Outlook

Next week: threatening events

Week 5 (Mon 12/Wed 14 Oct): threatening events and hate crime.

Required readings

Dipoppa, G., Grossman, G., & Zonszein, S. (2023). Locked down, lashing out: COVID-19 effects on Asian hate crimes in Italy. The Journal of Politics, 85(2), 389–404.
Riaz, S., Bischof, D., & Wagner, M. (2024). Out-group threat and xenophobic hate crimes: Evidence of local intergroup conflict dynamics between immigrants and natives. The Journal of Politics, 86(4), 1146–1161.
Álvarez-Benjumea, A., & Winter, F. (2020). The breakdown of antiracist norms: A natural experiment on hate speech after terrorist attacks. Proceedings of the National Academy of Sciences, 117(37), 22800–22804.

Weekly comments due Sunday 11 Oct, 11:59 pm.

Questions?

Goodbye!

See you next week.