Hate in Context

Week 2: Demographic Change and Hate Crime · LENS preview

Violeta Haas

Department of Political Science

School of Social Sciences and Philosophy

Trinity College Dublin

This Week

  1. Theoretical backdrop — contact & group threat
  2. Part I — Hangartner et al. (2019)
  3. Part II — Cikara et al. (2022)
  4. Part III — Dancygier et al. (2022)
  5. Putting the readings together

Theoretical backdrop

Group discussion (≈8 min)

In groups of 3–4:

  1. What is contact theory? What does it predict when minority presence grows and under what conditions?
  2. What is group threat theory? What does it predict and what kinds of “threat” might it include?

Contact theory

[discussed in class]

Group threat theory

[discussed in class]

Part I — Hangartner et al. (2019)

Most or least likely case?

Hangartner et al. study Greek islands during the 2015–16 refugee crisis.

For each theory, is that setting a most likely or least likely case for the mechanism to be activated and why?

Hangartner et al. (2019): the setting

[discussed in class]

Identification strategy

X exposure Y hostility U island traits Z distance

Selection problem: Regions with more refugee arrivals often already differ:

Self-selection

  • Refugees might choose places with jobs, co-ethnic networks, or a reputation for being welcoming

Administrative selection

  • Authorities may send refugees to some municipalities and not others (e.g. available housing, reception capacity, or political considerations)

→ X and Y share many causes! Naive X–Y comparison mixes exposure with pre-existing differences. Confounding (U)!

Identification strategy: the instrument

Hangartner et al. Figure 3: Map of the Aegean Sea showing refugee arrivals per capita by island

Z distance X exposure Y hostility U island traits

Instrument Z: distance to Turkey.

  • Geography pushes boats to closer islands, not (self-)selection: welcomingness or admin readiness.
  • Two-stage least squares (2SLS): (Z → Y) / (Z → X)
    • First stage (Z → X): how distance changes exposure
    • Reduced form (Z → Y): how distance changes hostility
  • Must hold: distance only affects hostility through exposure.

→ Exclusion restriction: fails if being close to Turkey changes hostility for any reason other than more arrivals.

2 min. discussion with neighbour

Suggest an instrument Z (shifts X & affects Y only through X).

[discussed in class]

Disclaimer

IV solves confounding in theory; in practice the fights are about weak Z, who the effect is for (LATE), and whether exclusion is believable.

Additional resource: Scott Cunningham, Causal Inference: The Mixtape (IV chapters).

Findings

Hangartner et al. Figure 4: 2SLS estimates of refugee exposure on hostility, policy preferences, and behaviour

Findings: islands closer to Turkey (more exposure) show higher:

  • exclusionary attitudes/hostility
  • anti-asylum & anti-immigrant policy preferences
  • behaviour (petition/donate against refugees)

Puzzle: contact and group threat were both weak here. Yet, hostility still rose. What’s left is mere exposure: even a brief, chaotic presence of arrivals (with little lasting settlement) can harden attitudes → situational social threat.

Part II — Cikara et al. (2022)

Ebbinghaus illusion

Which hollow circle is bigger? Why might they look different even if they aren’t?

Ebbinghaus illusion: two equal hollow circles surrounded by differently sized dots

[discussed in class]

Group reference dependence

RQ: How does demographic change affect local intergroup dynamics?

Argument: Group reference dependence hypothesis

Violence and negative attitudes toward each minoritized group depend on the number and size of other minoritized groups in a community, i.e., their relative ranks in size.

2 min. discussion with neighbour

  1. What role do stereotypes or essentialized properties (e.g., unchangeable group characteristics) play in this framework?

[discussed in class]

  1. What psychological finding about how people judge group sizes do they build on to focus on ranks rather than absolute size?

[discussed in class]

Let’s try it out!

Part 1: Rank by size (≈3 min)

Rank Dublin City’s largest immigrant groups by size (largest → smallest).

QR code for POU44653 Week 2 Mapping Knowledge form

  1. Open Blackboard → Week 02 → POU44653 Week 2: Mapping Knowledge or Scan QR-code
  2. Drag the ten citizenships into order
  3. Do not consult others
  4. Anonymous

Part 2: Mapping vs. Metric Knowledge (≈5 min)

Dublin City had 580,237 residents in Census 2022. For each citizenship below, guess (1) how many people and (2) what share of the city that is.

Citizenship Absolute number % of total population
Italy
China
United States
Romania
Poland
France
Brazil
Spain
India
United Kingdom

Solution

[discussed in class]

Finding 1

Cikara et al. Fig. 3: marginal effect of size rank on hate crimes

Baseline (red): being 1st / 2nd / 3rd in local size rank → more victimization than being 4th (strongest for 1st).

Rank changes (blue): same pattern when a group’s rank changes over time in the same county, i.e., comparing the same groups in the same places before vs after a rank switch.

→ Relative size rank predicts hate-crime victimization, even after accounting for group size and county/group fixed-effects (time-invariant characteristics).

Why only violent hate crimes?

They also re-run the analysis excluding non-violent hate crimes.

Why do that?

[discussed in class]

Finding 2

Cikara et al. regional rank encoding

Geographic scales (small → large): LSOA → MSOA → LAD

What does the plot show? At which scale is the effect strongest, and what does that imply about how people learn ranks?

[discussed in class]

2 min. discussion with neighbour

If people learn about ranks from everyday life, what would they actually notice?

→ Can you come up with an everyday example where group A now feels larger relative to group B than it used to?

Example: a rank switch in the news

CBS News, 2003: Hispanics now largest U.S. minority

→ Census coverage of Hispanics surpassing Black Americans as the largest US minority.

Example: public holidays as cues

NBC News: NYC makes Diwali an official school holiday

→ NYC adds Diwali to the school calendar in recognition of the city’s growing South Asian and Indo-Caribbean communities.

Example: 59 Brick Lane

Huguenot chapel (1743) → Methodist (1819) → synagogue (~1897) → mosque (1976)

1743 sundial Umbra Sumus from Huguenot chapel

Huguenot: La Neuve Église
Sundial dated 1743: Umbra Sumus (“we are but shadows”). Built by French Protestant silk weavers in Spitalfields.

Wesleyan Methodists chapel record

Wesleyan Methodist chapel
Same building from 1819 until 1897 (Methodist chapel record).

Herbert Cutner drawing of Spitalfields Great Synagogue

Spitalfields Great Synagogue
Cutner (1930s): Machzike Hadath; synagogue from 1897; Eastern European Jewish East End.

Brick Lane Jamme Masjid today

Brick Lane Jamme Masjid
Mosque since 1976, mainly the local Bangladeshi Muslim community.

→ One address, successive congregations: local institutions sometimes track relative group size over time.

What kinds of cues?

  • Face-to-face interactions
  • Media attention to demographic change
  • (Dis)appearance of (cultural) institutions (e.g., religious buildings, public holidays, LGBTQ+ venues)

Finding 3

Cikara et al. Supplementary Fig. F.6: symmetric effects of moving up vs down in rank

Supplementary Fig. F.6 from Cikara et al. (2022)

Up vs down: when a group moves up in local size rank (e.g. 2nd → 1st), hate-crime victimization rises. When it moves down (1st → 2nd), victimization falls.

→ Rank effects are roughly symmetric. Authors read this as substitution of prejudice across targets (fixed amount of discrimination redistributed among minority targets).

But: the same pattern can arise without substitution. Example: Group that dropped in rank might see fewer attacks because members move away (compositional change, not prejudice transferred).

Part III: Dancygier et al. (2022)

Who commits hate crime?

Perpetrators are overwhelmingly male.

  • Germany (right-wing extremist hate crime): roughly 85–97% men (cited in Dancygier 2023)
  • London (hate-crime, 2014–16): 78% men (Walters & Krasodomski-Jones 2018)

→ Hate crime perpetrators ≈ mostly men (often young / mid-life).

Reminder: two families of explanation

Individual-level

Psychological traits & offender behaviour

Cognitive and affective processes that drive perpetrators to commit hate crime

Contextual

Would-be perpetrators’ social environment

Political, cultural, sociological, and economic conditions that make bias-motivated violence more or less likely

→ Dancygier et al. (2022) integrate both, by looking at profile-specific contextual drivers.

Mating competition

Central concept: Mating competition

  • Function of the sex ratio (e.g., 1:1 → little competition)

Assumptions

[discussed in class]

Finding 1

Dancygier et al. finding 1: sex ratio and anti-refugee hate crime

→ Areas where men significantly outnumber women (more excess males) have higher probabilities and levels of anti-refugee hate crime.

From Macro to Micro

Finding 1 is ecological (established on the macro-level only): male-heavy places → more anti-refugee hate crime.

If the player fails, use: youtube.com/watch?v=5Vxbv138MuQ

Why go to the micro level?

1. Danger of ecological fallacy: Areas with skewed sex-ratios experience more hate-crimes, therefore men in those places commit hate crime because they fear mate competition.

→ place pattern ≠ individual mate-competition motive: place pattern could omit important context.

2. Theory implications for the individual-level

  1. Who should feel threatened?

[discussed in class]

  1. Which refugees should be the target?

[discussed in class]

Macro-level analysis alone can’t settle either. A survey can!

Finding 2

Dancygier et al. finding 2: perceived mate competition

→ Perceptions of native–refugee mate competition rise in areas with excess males, especially among men in the most mating-active age range.

What about (support for) hate crimes?

Dancygier et al. Figure 3: list experiment vs direct questions on hate-crime support

Figure 3 from Dancygier et al. (2022)

Problem: direct questions on hate-crime support invite social desirability bias (people hide unpopular views).

→ List-experiment: ask only how many list items you agree with.

  • Control: 3 harmless items (e.g., “Taking a few minutes a day to relax matters in a hectic world”)
  • Treatment: same 3 + sensitive item

→ Difference in means ≈ share endorsing the sensitive item.

Finding: people do not hide support. List estimate ≈ 15%; asked directly ≈ 18% (insignificant difference). If anything, direct answers are slightly higher.

Finding 3

Dancygier et al. finding 3: mate competition predicts hate crime support

→ Concerns over mate competition are a robust predictor of hate-crime support, even with controls for other attitudes.

→ Effects are stronger for goal-oriented items (violence as a means to send a message / keep refugees out) than for a value item (whether politicians should condemn attacks more forcefully).

Two unsettling implications

1. Integration can raise hostility

“[M]ajority opposition against minorities rises when the latter are in fact integrating into majority settings and intergroup barriers begin to fall.”

(Dancygier et al. 2022)

→ Refugees entering dating/marriage markets could be interpreted as a form of social integration but, Dancygier et al. (2022) show, this intensifies threat.

2. While perpetrators are few… supportive attitudes are not

→ A sizable share of survey respondents support violent acts against Refugees and do not feel the need to hide these attitudes.

→ Perpetrators may be the exception, but they can be empowered by broader attitudes favoring hate crime.

Group discussion (≈5 min)

In groups of 3: pick one of the two cases below. What follows for mate competition and for hate crime?

(A) Ukrainian refugee inflows

Gender composition of Ukrainian refugee inflows

(B) Rising LGBTQ ID among Gen Z women

NBC News: Nearly 30% of Gen Z women identify as LGBTQ, Gallup survey finds

Putting it together

From size to composition

  1. Hangartner et al. (2019). Sudden local exposure to refugee arrivals hardens hostility, exclusionary preferences, and far-right support. Mere passage can suffice; lasting economic or cultural threat is not required.

  2. Cikara et al. (2022). Which minoritized group is targeted depends on its relative size rank among other minorities in the locality, not only absolute size or group stereotypes (group reference dependence hypothesis).

  3. Dancygier et al. (2022). Who arrives matters: a male-skewed inflow can activate mate competition among native men, raising anti-refugee hate crime where local sex ratios are already unbalanced.

Outlook

Next week: economic causes

Week 3 (Mon 28 / Wed 30 Sep): economic causes and hate crime.

Required readings

  • Riaz, S. (2023). Does inequality foster xenophobia? Evidence from the German refugee crisis. Journal of Ethnic and Migration Studies, 50(2), 359–378.
  • Bestenbostel, A., & Peralta, A. (2026). WARNings: The impact of negative economic news on racial animus. Journal of Economic Behavior & Organization, 243, Article 107399.

Weekly comments due Sunday 27 Sep, 11:59 pm.

Questions?

Goodbye!

See you next week.