Intervention (Hidden)
We can divide our survey into four sequential sections: 1. demographic questions for screening (these are used to reach certain fixed quotas for country representativeness), 2. (conjoint) choice experiment, 3. perspective-taking experiment, and 4. collection of pre-treatment variables.
We randomly assign respondents to different conditions or “treatments” in sections 2 and 3, as we describe below. (All treatment assignments are stratified by country.)
1. Choice experiment
We use a conjoint choice experiment (henceforth, simply “choice experiment”) to evaluate formally any differences in attitudes associated with the refugees’ demographics, reasons for leaving their country, and costs (see Mariel et al. 2021 for an overview).
A choice experiment is characterized by five main features: the number of choice tasks, the alternatives presented in a choice task, the attributes used to describe each alternative, and the levels used to describe the corresponding attribute.
In our study, the choice task consists of presenting the respondents with two alternative scenarios describing a group of refugees asking which of the two alternatives, if any, they would prefer to offer temporary protection in the EU (“Please, compare groups A and B described in the table and indicate which of the two, if any, you would prefer to offer temporary protection in the EU.”). Each respondent will perform six of such choice tasks.
For each choice task, any alternative scenario is characterized by the following attributes (the attribute’s levels are in parenthesis):
• Country of origin of the refugees (An Eastern European country, A Middle Eastern country, A sub-saharan African country)
• Displaced by: Main reason forcing them to leave their country (poverty in their country, climate change making their country uninhabitable, a war in their country)
• Fraction of women (20 out 100, 50 out 100, 80 out of 100)
• Fraction of children (10 out 100, 30 out 100, 50 out 100)
• Religion (90 out 100 are Christian, 45 out of 100 are Christian, 45 are Muslim, 90 out of 100 are Muslim)
• Cost of protection (5,10,20,40,70,100)
Testing all possible scenarios (3 x 3 x 3 x 3 x 4 x 6 = 1,944 possible combinations) is unfeasible. Therefore, we restrict the choice sets to six blocks of combinations and we will sort the respondents into one of these blocks at random.
The level combinations in these blocks will be selected using a Bayesian d-efficient design optimized for Multinomial Logit models (Ferrini and Scarpa 2007), where the priors will come from a pilot study on a sample of about 450 respondents from four countries.
On top of the randomization into blocks, we will also randomize the choice task into three conditions where respondents are asked to given their preference for different types of temporary protection. The treatment conditions will be as follows:
1. Access to the labor market: In this condition, we test whether an emphasis on giving refugees temporary access to the labor market has any effect on the individuals’ support to temporary protection. This emphasis may stimulate people’s fear of a negative impact of the migrants on the labor market, which may lead to less inclusionary attitudes. We expect the effect to be larger for low-skilled individuals who may have a greater fear of direct migrant competition (see, e.g., discussion in Ford and Mellon 2020)
2. Housing: In this condition, we test whether putting more emphasis on giving refugees access to housing has any effect on the individuals’ support to temporary protection.
3. Generic protection (“Control”): This is the control condition with no emphasis on the implications of temporary protection for housing or the labor market.
Finally, we will also randomize the order of the alternatives displayed in the table shown in each block to prevent any ordering effect.
2. Taking vs. getting perspective
When asking to what extent individuals support specific aspects of temporary protection offered to Ukrainians, we will randomize the types of narratives/messages preceding the question. The treatment conditions will be as follows:
1. Perspective-getting: We will present respondents with an open questions about their feelings on the account of a refugee’s experience, thus to induce a reflection on the refugee’s perspective. We hypothesize that this perspective-giving reflection will persuade people to adopt more inclusionary attitudes, as shown in another experimental study described by Kalla and Broockman (2021).
2. Perspective-taking: this condition poses individuals an open question about what they would do if they were in the same conditions of a refugee from Ukraine. Adida, Lo, and Platas (2018) shows that a similar perspective-taking treatment condition successfully induced a significantly positive effect on Americans’ attitudes towards refugees in the context of the Syrian refugee crisis.
3. Unrelated (“Control”): This condition asks respondents an unrelated question that is not expected to trigger any feelings or perspectives about refugees (i.e., the question is about how they stay informed about the war in Ukraine).