Hate in Context

Week 5: Threatening Events and Hate Crime

Violeta Haas

Department of Political Science

School of Social Sciences and Philosophy

Trinity College Dublin

This Week

  1. Theoretical backdrop
  2. Álvarez-Benjumea & Winter (2020)
  3. Riaz, Bischof & Wagner (2024)
  4. Dipoppa, Grossman & Zonszein (2023)
  5. Outlook

Theoretical backdrop

Discussion: expert groups

Split into Expert Group A and Expert Group B (≈5 min).

Group A — Social norms

  1. Define social norms.
  2. How do attitudes, social norms, and behavior differ? (provide an example for each)
  3. Can behavior change without attitudes changing?

Group B — Vicarious retribution

  1. Sketch a schematic figure of the logic of vicarious retribution.
  2. How do triggering events translate into hate crime?

Then: exchange between experts — Group A presents their definitions and examples; Group B presents their schematic figure.

1. What is a social norm?

Bicchieri (2006, 2017): A social norm is a rule of behavior such that individuals prefer to conform to it on condition that they believe that:

(a) Empirical expectation

(= descriptive norm)

Most people in their reference network conform to it.

(b) Normative expectation

(= injunctive norm)

Most people in their reference network believe they ought to conform to it.

2. Attitudes, behavior, and norms

Attitudes

Individual evaluations or personal beliefs about whether something is good, bad, pleasant, or important

Social norms

Informal, shared expectations about what is typical (descriptive) or socially approved (injunctive)

Behavior

What an individual actually does in a given situation

→ individually motivated

→ socially motivated

→ individually or socially driven

e.g. “Cutting in line is unfair.”

e.g. “Everyone waits their turn, and others would disapprove if I jumped ahead.”

e.g. joining the back of the queue

Can behavior change without attitudes changing?

If the player fails, use: youtube.com/watch?v=o8BkzvP19v4

Can behavior change without attitudes changing?

Attitudes need not equal behavior. People often have a conditional preference: they act on a social rule only if they expect others to follow it (descriptive) and to think one ought to follow it (injunctive). So norms can suppress or enable behavior even when private attitudes haven’t changed.

People often look for environmental cues — what others are doing — to figure out how to act (descriptive norms).

Meme: when you have no idea what you are doing but pretend like you do

Book recommendation

Book cover: Bicchieri, Norms in the Wild

Bicchieri, C. (2017). Norms in the wild: How to diagnose, measure, and change social norms. Oxford University Press.

Triggering event

Triggering event: A1 assaults B1

An antecedent provocation (triggering event): a member of Group A harms a member of Group B.

Shared grievance and collective blame / liability

Collective blame / collective liability from A1 to A2; shared grievance from B1 to B2

A shared grievance spreads across Group B and Group A is collectively blamed / liable.

Vicarious retribution

Vicarious retribution: uninvolved B2 retaliates against uninvolved A2

A third party in-group member (B2) retaliates against an uninvolved out-group member (A2).

Vicarious retribution (Lickel et al., 2006):

A group member aggresses against out-group members for an assault or provocation that had no personal consequences for them, but did harm a fellow in-group member.

Prolonged intergroup conflict

Prolonged conflict cycle after vicarious retribution

Third-party retaliation invites counter-retribution and a vicious cycle of prolonged group conflict.

From trigger to hate crime: cognition and affect

How do triggering events translate into hate crime? Two key pathways (Anderson & Bushman 2002; Riaz et al. 2024):

Cognition

Galvanizing events can harden negative attitudes toward the out-group and raise threat perceptions.

→ Guide how people interpret situations and what (behavioral) responses come to mind.

Affect

Attitudes alone often do not change behavior — emotional reactions may mediate. Galvanizing events trigger anger.

→ Reduces inhibitions & primes aggressive behavior.

Cognition shifts how the out-group is seen; affect supplies the short-run emotional push.

Today’s papers ask how discrete shocks (triggering events) shift hostility — offline and online — including by changing what seems acceptable (social norms) and by redirecting blame onto uninvolved out-group members (vicarious retribution).

Álvarez-Benjumea & Winter (2020)

Overview

  • Research question: Do terrorist attacks increase online hate speech — and do descriptive norms moderate that effect?
  • Y: hate score of forum comments (1–9 hostility ratings)
  • X: terrorist attacks during fieldwork (Würzburg & Ansbach, July 2016)
  • Moderator: descriptive norms (no / weak / strong)
  • Theory: attacks can weaken antiracist norms → latent xenophobic attitudes become behavior (hate speech)
  • Design: unexpected event during survey/experiment (UEDS) + lab-in-the-field norm manipulation

Unexpected Event During Survey (UEDS) design

Idea: attacks fall between two waves of the same experiment. Compare comments from the before wave vs the after wave.

Comparison around the attacks

Different participants in the same online experiment

UEDS design: Wave 1 on 10 July; Würzburg and Ansbach attacks 18–24 July; Wave 2 from 3 August

Experimental forum

Álvarez-Benjumea and Winter 2020 Figure 1: Screenshot of the no-norm experimental forum condition

  1. Participants join an online forum, see a prompt photo and seed comments that manipulate the descriptive anti-hate norm:
    • No norm: mix of hostile, neutral, and positive comments (norm ambiguous)
    • Weak norm: only neutral and positive comments (anti-hate norm signaled)
    • Strong norm: only very positive comments (anti-hate norm strongly signaled)
  2. Participants then leave their own comment.
    • Comments are rated for hostility on a hate score from 1–9.

Findings: mean change after attacks

Álvarez-Benjumea and Winter 2020: Post-attack mean change in hate score by norm condition

After the attacks, hate toward refugees rises when no norm is shown — but not under weak or strong norms. No parallel shift for unrelated group (i.e., gender rights).

Findings: extreme comments

Álvarez-Benjumea and Winter 2020: Quantile treatment effects of terrorist attacks by norm condition

Descriptive norms especially dampen the most extreme comments: effects concentrate in the upper tail of the hate-score distribution.

Group discussion (≈5 min)

Two main responses to online hate (Siegel 2020): content moderation vs counter-speech.

In groups of 3–4:

  1. For each approach, collect 2 points in favor and 2 drawbacks.
  2. What do Álvarez-Benjumea & Winter’s results speak for?
  3. What do you think is the best way to deal with hate speech?

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).

Pros and cons

Approach A: Content moderation

In favor

  • Can cut exposure fast by removing or demoting high-reach hate
  • Clear rules (and legal bans where they exist) may deter some producers

Drawbacks

  • Over-removal and uneven enforcement (who decides what counts as hate?)
  • Speakers migrate to fringe platforms; hate can reappear under new accounts

Approach B: Counter-speech

In favor

  • Preserves open debate while challenging norms that make hate acceptable
  • Can mobilize bystanders and support targets without relying only on platforms

Drawbacks

  • Risks amplifying hate (replies boost visibility) or burdening targets to respond
  • Hard to scale; effects often weak against committed haters or coordinated campaigns

What do Álvarez-Benjumea & Winter speak for?

What is the best way to deal with hate speech?

Riaz, Bischof & Wagner (2024)

Overview

  • Research question: Do local immigrant-attributed crimes raise hate crimes against refugees?
  • Y: daily probability / count of hate crimes against refugees
  • X: immigrant-attributed crime event
  • Theory: vicarious retribution, i.e., natives retaliate against uninvolved out-group members after a triggering crime
  • Design: regression discontinuity in time (RDiT) around local crime events (Germany, 2015–2019)

Wait… RDiT? That sounds familiar!

Regression Discontinuity in Time (RDiT) design

Idea: a sudden local shock is a cutoff in time. Compare hate crimes from the before period vs the after period.

Comparison around local crime events

Same county, days close to the event

RDiT design: compare hate crimes before vs after immigrant-attributed crime, excluding day 0

Scope conditions

Scope condition: the subset of cases / conditions under which a theory applies.

Vicarious retribution applies when…

  • An antecedent event leaves one group with a grievance against another
  • A definable out-group is blamed for the triggering event
  • The event receives considerable publicity (King & Sutton, 2013)
  • An antecedent event leaves one group with a grievance against another
  • A definable out-group is blamed for the triggering event
  • The event receives considerable publicity (King & Sutton, 2013)

→ Riaz et al. (2024): considerable publicity is not necessary — smaller local triggering events can also have an effect.

How do natives—including potential hate crime perpetrators—hear about immigrant-attributed crime in their locality?

(1) Direct exposure to media reporting

Pew survey: among local news topics, Americans get news about crime more than any topic except weather

Crime is the second-most followed local news topic (after weather) — so even without national publicity, local crimes can be highly visible.

(2) Indirect exposure through social networks

Screenshot of refcrime.info: Refugee and Migrant Crime Map aggregating immigrant-attributed crimes

Findings: daily hate-crime rate around events

Riaz et al. 2024 Figure 2: Daily rate of hate crimes in the two weeks before and after an immigrant crime event

Hate crimes surge locally right after immigrant-attributed crimes — but the spike fades within ~4 days, consistent with an emotional trigger (affect) rather than lasting attitude change.

Findings: where the backlash is stronger

Riaz et al. 2024 Figure 3: Heterogeneous LATE estimates by AfD strength and foreign-population context

Backlash is stronger where far-right mobilization potential is high and where the foreign population has recently increased — not simply where many foreigners already live.

Anything else that didn’t convince you?

Outlook

Next week: political elites

Week 6 (Mon 19/Wed 21 Oct): political elites and hate crime.

Required readings

  • Romarri, A. (2020). Do far-right mayors increase the probability of hate crime? Evidence from Italy. Working paper.
  • Dancygier, R. (2023). Hate crime supporters are found across age, gender, and income groups and are susceptible to violent political appeals. Proceedings of the National Academy of Sciences, 120(7), Article e2212757120.
  • Zonszein, S., & Grossman, G. (2025). Voted in, standing out: Public response to immigrants’ political accession. American Journal of Political Science, 69(2), 718–733.

Weekly comments due Sunday 18 Oct, 11:59 pm.

Questions?

Goodbye!

See you next week.