Measuring the Effect of Media Framing on Behavior Towards Refugees

ECPR General Conference

Innsbruck

25 August 2022

Dawid Walentek

University of Warsaw

About us

• Natalia Letki Principal Investigator (University of Warsaw)
• Dawid Walentek (University of Warsaw)
• Artem Graban (University of Warsaw)
• Ulf Liebe (University of Warwick)
• Peter Thisted Dinesen (University College London)
   
• Funding: NCN (National Science Centre Poland) grant no 2019/33/B/HS6/0084

Background

• 2015 Refugee Crisis
• Strong presence of the topic in the media
• Great variation in the kind of behaviour we observe
• Ongoing relevance: crisis on the Polish-Belarussian border, war in Ukraine

Literature

• Mobilisation of sentiments towards refugees and asylum seekers linked to media reporting (Aschauer & Mayerl, 2019; Krzyżanowski, 2020; Abou-Chadi, Cohen, & Wagner, 2022)
  • ​right-wing political preferences and ethnocentric attitudes increasing susceptibility to negative news
• Relation between stated preferences and actual behaviour
  • sometimes it works in the same direction (Adida, Lo & Platas, 2018)
  • ... and sometimes it does not (Liebe et al., 2018)
• We lack systematic evidence on the presence and degree of the media effect on actual behavior in respect to refugees

Data & Methods

Data

• Online survey experiment in early 2022
• 7 EU Member States
  • Germany, Austria, Poland, Hungary, Slovenia, Bulgaria and Croatia
• 8,832 respondents
• 3 treatment groups
• Well-balanced across treatment arms and quota representative
  • age, gender, NUTS, education following Eurostat 
• Embedded in a larger survey
• Preregistration with EGAP

Treatments

• Three visual frames
  • Welfare
  • Security
  • Humanitarian
• Images accompanied by a short caption
• Widely circulated in media in relation to the relevant primes

Methods

• Final part of the survey
• Respondents receive first general information about the International Rescue Committee
• Then respondents are presented with the real-effort task, where they must decide whether to donate half of the survey remuneration to the IRC

Results & Discussion

Results

• Control group:
  • PL, DE & BG ~ 25%
  • HU & SL ~ 30%
  • AU & HR ~ 35%

Results

• Treatments: Humanitarian only
  • Region FE (NUTS 2)
  • Bonferroni
    correction
  • ATE = .05
  • 5% higher donation

Results: heterogeneity

• Heterogeneity in the treatment effect
  for different levels
  of ethnocentrism
• Ethnocentrism
  based on a set of survey questions

Results: attitudes
and behaviour

• We see a correspondence between attitudes
  and behaviours
• Average rating assigned to profiles
  with a relocation
  attribute level

What do we learn?

• Responsive to humanitarian media frame, but not across the (ethnocentric) board
• Overlap between stated preference and behavioural benchmark
• Corroborates prior research pointing to the role of humanitarian concern (Bansak et al., 2016; Adida et al., 2018)

Thank you