Snowflakes or Tough Cookies?

How citizens judge harassment against politicians
DIPAS Workshop 2025

Søren Damsbo-Svendsen
Karina Kosiara-Pedersen

September 23, 2025

Motivation

  • There is both loose conjecture (e.g., political debate) and careful theory about heterogeneous perceptions of harassment among citizens
  • But little solid empirical evidence
    • ‘solid’: large/broad data collection and broad ‘harassment’ concept
  • RQ: How severe is harassment against politicians to citizens, and what are theoretically and statistically significant gaps in perceptions of harassment?
    • i.e., are (harassed) politicians perceived as snowflakes or tough cookies? By whom?
  • Preregistered survey:
    • large sample (n = 8,899) from 4 European countries
    • 55 true stories of (Danish) politicians’ experiences with harassment
    • outcome: perceived severity
    • 43,460 judgments of a diverse set of harassment stories

Status

  • First draft
  • Data part of larger project with experiment
  • Analysis quite broad/deep and quite far
    • main results ‘final’
    • possibly need for cuts, possibility for extensions
    • gender and age weaved into every analysis for coherence
  • Framing and especially theory quite underdeveloped
    • we’d love to hear some of your takes on our findings

Design

  • New multinational preregistered survey
  • n = 8,899 respondents
    • Denmark: n = 5,008 (YouGov)
    • Germany: n = 1,519 (Prolific)
    • United Kingdom: n = 1,515
    • France: n = 857
  • Harassment stories
    • respondents read and evaluate 5 harassment stories
    • real stories about (Danish) politicians’ experiences with harassment
    • source: Danish 2021 municipal election candidate survey (n = 3,986)
    • stimulus sampling design with random draw from 55 curated stories

Examples of harassment stories

“I have experienced my election posters being ‘decapitated’ by unknown persons. My children, aged seven and nine, have seen this, which was very unpleasant.”

“I have experienced unpleasant advances in private messages on social media from people of all ages.”

“I have received death threats from unknown senders directed at both my little brothers and my own children.”

“There was a citizen who drove around in a car and measured how high my - and only my - posters were placed from the ground. He then reported me to the police for banners that were placed too low.”

How the stories are presented and judged


  • 1 of 4 politician gender-age categories (randomized)

  • Story


  • Judgment on three dimensions: seriousness, justifiableness, harmfulness


Repeats five times

Perceived severity

  • Judgments are combined into a severity index
  • I.e., respondents de facto rate each story’s severity from 0 to 100
  • => 43,460 judgments of 55 harassment stories
  • Analytic focus: theoretically and statistically significant gaps in perceived severity by key explanatory variables
    • a.k.a. predictors, covariates
    • e.g., gender, ideology, risk perception
    • four ‘themes’ →

Explanatory variables – four themes

  • A Gender and age
    1. gender
    2. age
    3. gender-age
    4. politicians’ gender-age
    5. gender-age affinity
  • B Ideology
    1. left-right placement (general)
    2. harassment attitudes (specific)
  • C Risk and likely candidates
    1. harassment risk perception
    2. general conflict aversion
    3. likely candidate (predicted)
    4. likely candidate (expressed ambition)
  • D Harassment type
    1. physical harassment, violence (threats), vandalism (non-person)
    2. sexual harassment
    3. social media (arena)
    4. system-insider (source)

Results: A Gender and age

Gender gap in perceived severity?


A Gender and age

  • A: Higher severity for women
  • B: Lower severity for young
  • C: Highest severity for older women, lowest for young men
  • D: Highest severity when politician (target) is young woman, lowest when older man

A Gender-age affinity

  • A: Subset to stories with gender-age affinity (~25%)
  • A: Gaps similar to full sample
  • B: Gender-age affinity causes higher severity for women
  • B: Gender-age affinity causes lower severity for older men
  • B: Largely driven by perceptions related to politician’s identity

Results: B Ideology

B Ideology

  • A: Higher severity for left-leaning respondents
  • B: Higher severity for respondents with clear anti-harassment attitudes
  • → Specific harassment-related ideology more important than general ideological leaning
  • Ideological gaps ~2-4 %-points smaller for women

Results: C Risk and likely candidates

C Risk and likely candidates

  • A: Higher severity for respondents with high risk perceptions
  • B: Higher severity for respondents with high conflict avoidance
  • C: Higher severity for likely future candidates (index),
    but …
  • D: Lower severity for citizens with expressed political ambition (??)

C Risk and likely candidates


  • We interact the two measures of potential candidates
  • Within each x-axis category (‘resources’1, or predicted likelihood of running for office)
    • lower severity for those with expressed political ambition
  • Within each color category (expressed political ambition)
    • higher severity for those with higher index values

C Risk and likely candidates


  • Plausible interpretation:
    • those who actively consider politics (expressed ambition) have factored harassment in, devalued the severity, are less surprised
    • or, due to self-selection, they have ‘thicker skin’ to begin with
  • Raises questions:
    • about who’s willing to run/stay
    • about whether/when/how candidates’ perceptions of harassment change

Results: D Harassment type

D Harassment type

  • A-B: higher severity for stories of physical harassment and (threats of) violence
  • C: Lower severity for stories of vandalism (i.e., higher when target is a person)
  • D: No gap for sexual vs. non-sexual harassment
  • D: But large positive gap for young and small negative gap for older people

D Harassment type

  • E: Lower severity for stories in social media arena
  • E: Especially among young people (gap smaller among women)
  • F: Lower severity when party colleague is ‘source’
    (more excusable?)
  • F: Gap smaller among women

Summary of gaps

  • A Gender and age
    1. gender: large
    2. age: moderate (-)
    3. gender-age: large
    4. politicians’ gender-age: small
    5. gender-age affinity: small
  • B Ideology
    1. left-right placement: large
      (gap smaller for women)
    2. harassment attitudes: very large (gap smaller for women)
  • C Risk and likely candidates
    1. harassment risk perception: very large
    2. general conflict aversion: very large
      (gap larger for women/young)
    3. likely candidates (predicted): small (-)
    4. likely candidates (expressed ambition): small
  • D Harassment type
    1. physical harassment, violence, vandalism: large
    2. sexual harassment: none
      (large gap for young)
    3. social media (arena): large
    4. system-insider (source): large

To do: theory/interpretation/discussion


  • Why do we find these gaps? How do we explain them? What are their causes?
  • What are theimplications?
  • How (much) do we discuss?
  • Best framing?
  • Paper structure (four ‘themes’ etc.)? Need to throw out parts or parcel into another paper?

Conclusions

  • We asked how citizens perceive harassment against politicians in terms of severity
  • Focus on important gaps within four distinct themes:
    • A Gender and age
    • B Ideology
    • C Risk and likely candidates
    • D Harassment type
  • Find numerous important and interesting gaps
    • some expected (e.g., gender, ideology, risk and conflict aversion, physical)
    • others unexpected (e.g., age, gender-age affinity, sexual, party colleague)
    • also interesting gender/age gaps in other gaps (e.g., woman x ideology, age x sexual)
  • Causal effects of the stories in separate paper

Thank you