www.sciencemag.org/cgi/content/full/science.aag2147/dc1 Supplementary Materials for How economic, humanitarian, and religious concerns shape European attitudes toward asylum seekers This PDF file includes Materials and Methods Supplementary Text Figs. S1 to S49 Tables S1 to S61 References Kirk Bansak, Jens Hainmueller,* Dominik Hangartner *Corresponding author. Email: jhain@stanford.edu Published 22 September 2016 on Science First Release DOI: 10.1126/science.aag2147 Other Supplementary Materials for this manuscript include the following: (available at www.sciencemag.org/cgi/content/full/science. aag2147/dc1) Data and replication archive (Harvard Dataverse doi:10.7910/dvn/kl0fdf) Preregistered analysis plan (http://dx.doi.org/10.7910/dvn/yunkul)
Contents Materials and Methods 3 Sample.......................................... 3 Sample weights..................................... 3 Experimental design.................................. 4 Outcomes........................................ 4 Survey text....................................... 5 Survey translations................................... 5 Statistical analysis................................... 5 Supplementary Text 6 General attitudes toward asylum seekers....................... 6 Notes on conjoint plots and regression tables..................... 6 Figures 8 Tables 57 2
Materials and Methods Sample We fielded our survey over the Internet in fifteen European countries that currently belong to the Dublin system, the institutional arrangement that coordinates asylum policy in Europe. We chose the sample of countries to represent a diversity of national characteristics, including coastal and non-coastal border countries, large and small economies, countries with major and minor political influence, and countries with varying degrees of popularity as asylum-seeker destinations. The sample includes Austria, the Czech Republic, Denmark, France, Germany, Greece, Hungary, Italy, the Netherlands, Norway, Poland, Spain, Sweden, Switzerland, and the United Kingdom. In each country, the survey firm Respondi sampled adult respondents from the population of eligible voters in late February and early March 2016. Respondi uses mostly online channels and, to a lesser extent, also computer-assisted telephone interviews (CATI) to recruit new panelists for its online panel (see the technical report for the Short-term Campaign Panel of the German Longitudinal Election Study 2009 (44) for details). After completing a profiling questionnaire covering basic sociodemographic information, panelists are then invited to participate in a few surveys. Respondi compensates its panelists for completing a survey. In our study, the modal incentive across all countries was EUR 2.00 for a median length of interview (LOI) of 20.53 minutes. In addition to Respondi s standard recruitment processes, the recruitment of panelists for our survey also employed age and gender quotas to roughly match the population margins for each of the fifteen countries in our study (post-stratification weights were constructed to account for remaining imbalances, as explained below). Among the panelists invited to take our survey, the response rate (calculated as the fraction of complete responses over invited, eligible participants) was 21.3%, averaged across all countries. The final sample included approximately 1,200 respondents per country (18,000 respondents overall). Descriptive statistics are reported in Tables S5-S7. The survey was conducted according to the University of Zurich s policy for human subjects research and approved by Stanford University s Institutional Review Board (protocol ID: 34881). Informed consent was obtained from each participant at the beginning of the survey. Sample weights Similar to most surveys based on internet panels, our samples from some of the countries included in our study were slightly skewed towards younger and more educated eligible voters compared to the overall population of eligible voters (in contrast, the samples were balanced on gender in all countries). To address these imbalances, we use entropy balancing (24) to reweight our sample data to match the demographic margins from the populations of each country. Unless otherwise noted, all statistical analyses presented in the main text and in this SM document employ these post-stratification weights. Note that data required for constructing the weights were missing for 147 respondents, and hence they were dropped from all weighted analyses. In Table S9 we also present unweighted analysis for comparison, 3
and the results are very similar to the weighted analysis. Experimental design We embedded in our survey a conjoint experiment designed to measure how a range of specific asylum-seeker characteristics affect voters willingness to welcome asylum seekers into their countries. Conjoint experiments, which ask subjects to evaluate profiles that combine multiple randomly assigned attributes, have long been used in marketing research and have become increasingly common in social scientific research to measure multidimensional preferences. We employed a standard fully randomized paired profiles conjoint design (27) in which each respondent was shown profiles of two different hypothetical asylum seekers displayed side-by-side (see Fig. S1 for an example). Previous research (36) has found that the paired profiles conjoint design performs best among several conjoint and vignette designs in terms of reducing social desirability bias and replicating real-world behavior. The profiles were comprised of nine attributes, including the asylum seeker s gender, age, country of origin, previous occupation, proficiency in the host country language, religion, reason for migrating, consistency of the asylum testimony, and special vulnerabilities. Table S1 describes the full list of attributes and the possible values each attribute could take. We chose these attributes in consultation with asylum policy experts from the Migration Policy Group, UNHCR, and the Swiss Refugee Council, and based on the detailed handbook (45, 46) that the Swiss State Secretariat of Migration provides for its asylum officers, which specify the topics on which the officers must elicit information during asylum interviews. This ensures that we captured the most relevant characteristics that officials typically consider when making decisions to grant or reject asylum. We also included attributes that have been previously identified in the academic literature as important for generating support for the admission of immigrants (36, 39). Each respondent was asked to evaluate five separate pairs of asylum-seeker profiles, and for each profile we randomly assigned the values of all attributes. We also randomly assigned for each respondent the order in which the attributes were listed in the conjoint table to avoid order effects. Fig. S1 displays an example of a pair of profiles presented to the respondents. Outcomes After being shown a pair of asylum-seeker profiles, respondents were asked to perform two tasks. First, for each profile, respondents were asked to rate on a scale from 1 to 7 how supportive they would be of allowing the hypothetical asylum seeker to stay in the host country. These data were used to construct a rating outcome variable. Second, out of the pair of profiles, respondents were then asked to choose the one asylum seeker they would prefer to be allowed to stay in the country. This resulted in a choice outcome variable, which was coded as 1 for the preferred profiles and 0 for the rejected profiles. The Survey text subsection below describes the exact question wording, and Table S2 describes the dependent variables used for analysis in greater detail. 4
Survey text The following is the text used to introduce the respondents to the conjoint section of the survey, as well as the questions used to measure the respondents evaluations of the asylum-seeker profiles and hence construct the outcome variables. Prelude: Now we would like to show you the profiles of potential applicants for asylum in Europe. You will be shown pairs of asylum seekers, along with several of their attributes. We would like to know your opinion regarding whether you would be in favor of sending each applicant back to their country of origin or allowing them to stay in [Respondent s Country]. In total, we will show you five comparison pairs. Please take your time when reading the descriptions of each applicant. People have different opinions about this issue, and there are no right or wrong answers. Questions: 1. (Rating) On a scale from 1 to 7, where 1 indicates that [Respondent s Country] should absolutely send the applicant back to their country of origin and 7 indicates that [Respondent s Country] should definitely allow the applicant to stay, how would you rate each of the asylum seekers described above? 2. (Choice) Now imagine that you had to choose one applicant who would be allowed to stay in [Respondent s Country], and the other applicant would be sent back to their own country of origin. Which of the two applicants would you personally prefer to be allowed to stay in [Respondent s Country]? In addition to the conjoint component, the survey instrument also contained a number of questions that measured various characteristics and attitudes of the respondents. Tables S3 and S4 describe these additional variables. Survey translations The survey instrument was designed in English and then professionally translated in each of the country s languages. To verify the quality and accuracy of the translations, the translated questionnaires were also professionally back-translated. Statistical analysis Each of the approximately 18,000 respondents evaluated five pairs of profiles, resulting in a total of approximately 180,000 profiles being evaluated. Given that the attribute values were randomly assigned across all profiles and respondents, we can identify the so-called average marginal component-specific effects (AMCEs) which measure the average causal effect of each attribute on respondents acceptance of an asylum seeker (27). To estimate the AMCEs, we use linear (weighted) least squares regression to regress the rating and choice 5
outcomes (see Table S2 for precise details on the dependent variable construction) on sets of indicator variables that capture the values of each attribute while omitting one level of each attribute as the reference category and clustering the standard errors by respondent (27). Unless otherwise noted, all regressions employ the sample weights. Since the results are very similar regardless of whether we use the rating or choice outcome, we focus the bulk of our presentation of the results on the latter. The robustness of the results when using the rating outcome is shown in Figs. S16-S21 and Tables S9, S26-S31. For replication code and data, see Harvard Dataverse (doi:10.7910/dvn/kl0fdf). A preregistered analysis plan is available at the Political Science Registered Studies Dataverse (http://dx.doi.org/10.7910/dvn/yunkul). Supplementary Text General attitudes toward asylum seekers The number of profiles that each respondent accepted in their conjoint tasks, based on the Binary Rating version of the dependent variable (see Table S2 for description), is highly correlated (ρ = 0.47) with respondents general attitudes towards asylum seekers as measured by an additional question in our survey that asked respondents whether they want to decrease or increase the number of people granted asylum in their country (see Table S4 for description). This suggests that respondents judgments of individual profiles are closely linked to their support for broader asylum policies. Notes on conjoint plots and regression tables 1. All plots of the conjoint AMCEs contained in the figures in the main text and this SM document are based on linear (weighted) least squares regression estimates. The dots with horizontal lines indicate point estimates with cluster-robust 95% confidence intervals. The unfilled dots on the zero line denote the reference category for each asylum-seeker attribute. 2. The estimated AMCEs displayed in all plots are numerically presented in corresponding regression tables in this SM document. Captions to all relevant figures include a reference to the corresponding regression table. 3. All regression models are estimated by linear (weighted) least squares regression. Point estimates are displayed, and standard errors clustered by respondent are shown in parentheses in the regression tables. 4. Unless otherwise noted, all regression models employ the choice outcome as the dependent variable (Forced Choice). Where specified, certain models also employ two other dependent variables. The first is a dichotomized version of the rating outcome (Binary Rating), which was coded as a 1 for profiles rated higher than 4 (out of 7) and 0 otherwise, thus serving as an indicator for acceptance. The second is a scaled version 6
of the rating outcome (Scaled Rating), where the original 1-7 coding is rescaled to vary between 0 and 1. See Table S2 for more detail on the alternative dependent variables. 5. Unless otherwise noted, all regression models regress the dependent variable on the full set of asylum-seeker profile attributes (a dummy variable for all levels except for one reference level per attribute). 6. For all regression models, the reference categories for the nine attributes are as follows. Asylum Testimony: No inconsistencies, Female, Country of Origin: Syria, 21 Years, Previous Occupation: Unemployed,, Reason for Migrating: Political persecution, Christian, Language Skills: Fluent. 7. Unless otherwise noted, all regression models are estimated using the post-stratification weights. 8. In each model, the number of observations reflects the number of survey respondents pertaining to that model multiplied by 10, since each respondent evaluated 10 profiles (5 pairs) total. 9. There were 18,030 respondents, but data were missing for 156 respondents (data necessary for constructing the post-stratification weights were missing for 147 respondents, while other data were missing for an additional 9 respondents). This resulted in 178,740 total observations in the pooled weighted regression models. 7
Figures Fig. S1. Example profile pair. This figure displays an example of a pair of asylum-seeker profiles seen by the survey respondents. This example comes from the English version of the survey administered to respondents in the United Kingdom. Each respondent saw and evaluated five separate pairs. The order of the attributes (rows) was fully randomized between respondents, but for each respondent, the order was kept constant across the five pairs they were shown. The specific attribute levels (values in the cells in the last two columns) were fully randomized between and within respondents. Two different asylum seekers ATTRIBUTE APPLICANT 1 APPLICANT 2 Age 21 Years 62 Years Language Skills Previous Occupation Speaks broken English Unemployed Speaks fluent English Teacher Different attributes of the asylum seekers Religion Christian Muslim Consistency of Asylum Testimony Vulnerability Minor inconsistencies Post-traumatic stress disorder (PTSD) Major inconsistencies No surviving family members Origin Iraq Pakistan Reason for Migrating Seeking better economic opportunities Persecution for ethnicity Gender Male Male 8
Fig. S2. 9 AMCEs across country. Tables S11 and S12 present the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Austria Czech Republic Denmark France Germany Greece Hungary Italy Netherlands Norway 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance Poland Spain Sweden Switzerland United Kingdom
Fig. S3. AMCEs across political ideology. Table S13 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S3 describes how political ideology was coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Far Left Left Center Right Far Right Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 10
Fig. S4. AMCEs across ideological placement of party of identification. Table S14 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S4 describes how party ideological placement was coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Far Left Left Center Right Far Right Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 11
Fig. S5. AMCEs across age groups. Table S15 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies 29 and under 30 39 40 49 50 59 60 and over Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 12
13 Fig. S6. AMCEs across education level. Table S16 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S3 describes how education level was coded. < Lower Secondary Lower Secondary Lower Tier Upper Secondary Upper Tier Upper Secondary Advanced Vocational BA Level >= MA Level Broken Fluent Language skills: Muslim Agnostic Christian Economic opportunities Ethnic persecution Religious persecution Political persecution Reason for migrating: Handicapped No surviving family Victim of torture PTSD Doctor Teacher Accountant Farmer Cleaner Unemployed Previous occupation: 62 years 38 years 21 years Iraq Ukraine Pakistan Eritrea Kosovo Afghanistan Syria Country of origin: Male Female Major inconsistencies Minor inconsistencies No inconsistencies Asylum testimony: 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S7. AMCEs across income level. Table S17 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S3 describes how income level was coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Income Quintile 1 Income Quintile 2 Income Quintile 3 Income Quintile 4 Income Quintile 5 Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 14
Fig. S8. AMCEs across gender. Table S18 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Male Female 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 15
16 Fig. S9. AMCEs across employment status. Table S19 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S3 describes how employment status was coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Paid employee Self employed Student Unemployed, searching Unemployed, not searching Chronic illness or permanent disability Retired Working at home 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S10. Comparing AMCEs of native-born respondents. Table S20 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken All Respondents Only Native Born Respondents 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 17
Fig. S11. AMCEs across levels of empathy (empathic concern and perspective taking). Table S21 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S4 describes how empathy levels were coded. This figure employs the combined empathy index. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Empathy Quintile 1 Empathy Quintile 2 Empathy Quintile 3 Empathy Quintile 4 Empathy Quintile 5 Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S12. AMCEs across levels of empathy (empathic concern). Table S22 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S4 describes how empathy levels were coded. This figure employs the empathic concern index. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Empathy Quintile 1 Empathy Quintile 2 Empathy Quintile 3 Empathy Quintile 4 Empathy Quintile 5 Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S13. AMCEs across levels of empathy (perspective taking). Table S23 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Table S4 describes how empathy levels were coded. This figure employs the perspective taking index. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Empathy Quintile 1 Empathy Quintile 2 Empathy Quintile 3 Empathy Quintile 4 Empathy Quintile 5 Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S14. AMCEs across tasks. As described in the Materials and Methods section, each respondent evaluated five pairs of asylum-seeker profiles. In this figure, the task number denotes the particular pair (e.g. the first pair evaluated vs. the fifth pair evaluated) to which each observation belongs. Table S24 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. The median and mean absolute difference between all pairwise comparisons of effects across all five tasks is only 1 percentage point. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Task 1 Task 2 Task 3 Task 4 Task 5 Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S15. AMCEs across length of time to complete survey. In this figure, Long refers to those respondents who took greater than the median amount of time to complete the survey, and Short otherwise. Table S25 presents the regression estimates plotted in this figure. The estimates displayed here employ the Forced Choice dependent variable. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Short Long 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 22
Fig. S16. 23 AMCEs using alternative dependent variable (Binary Rating). Similar to Fig. 2 of the main text, this figure plots the AMCEs but uses an alternative dependent variable. Table S9 presents the regression estimates plotted in this figure. The estimates displayed here employ the Binary Rating dependent variable. Table S2 describes how this variable is coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S17. 24 AMCEs using alternative dependent variable (Scaled Rating). Similar to Fig. 2 of the main text, this figure plots the AMCEs but uses an alternative dependent variable. Table S9 presents the regression estimates plotted in this figure. The estimates displayed here employ the Scaled Rating dependent variable. Table S2 describes how this variable is coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 Change in rating level
Fig. S18. 25 Subsetted AMCEs using alternative dependent variable (Binary Rating). Similar to Fig. 3 of the main text, this figure plots the subsetted AMCEs but uses an alternative dependent variable. Table S26 presents the regression estimates plotted in this figure. The estimates displayed here employ the Binary Rating dependent variable. Table S2 describes how this dependent variable is coded, and Table S3 describes how the subset variables are coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Left Right Young Old Low Education High Education Below Median Income Above Median Income 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S19. 26 Subsetted AMCEs using alternative dependent variable (Scaled Rating). Similar to Fig. 3 of the main text, this figure plots the subsetted AMCEs but uses an alternative dependent variable. Table S27 presents the regression estimates plotted in this figure. The estimates displayed here employ the Scaled Rating dependent variable. Table S2 describes how this dependent variable is coded, Table S3 describes how the subset variables are coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Left Right Young Old Low Education High Education Below Median Income Above Median Income 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Change in rating level
Fig. S20. 27 AMCEs across country using alternative dependent variable (Binary Rating). Similar to Fig. S2, this figure plots the ACMEs across country but uses an alternative dependent variable. Tables S28 and S29 present the regression estimates plotted in this figure. The estimates displayed here employ the Binary Rating dependent variable. Table S2 describes how this variable is coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Austria Czech Republic Denmark France Germany Greece Hungary Italy Netherlands Norway Poland Spain Sweden Switzerland United Kingdom 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance
Fig. S21. 28 AMCEs across country using alternative dependent variable (Scaled Rating). Similar to Fig. S2, this figure plots the ACMEs across country but uses an alternative dependent variable. Tables S30 and S31 present the regression estimates plotted in this figure. The estimates displayed here employ the Scaled Rating dependent variable. Table S2 describes how this variable is coded. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Austria Czech Republic Denmark France Germany Greece Hungary Italy Netherlands Norway Poland Spain Sweden Switzerland United Kingdom 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Change in rating level
Fig. S22. AMCEs by consistency of asylum testimony, using Forced Choice dependent variable. Table S32 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies No Inconsistencies Minor Inconsistencies Major Inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 29
Fig. S23. AMCEs by consistency of asylum testimony, using Binary Rating dependent variable. Table S33 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies No Inconsistencies Minor Inconsistencies Major Inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 30
Fig. S24. AMCEs by consistency of asylum testimony, using Scaled Rating dependent variable. Table S34 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies No Inconsistencies Minor Inconsistencies Major Inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Change in rating level 31
Fig. S25. AMCEs by asylum seeker s gender, using Forced Choice dependent variable. presents the regression estimates plotted in this figure. Table S35 Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Female Male 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 32
Fig. S26. AMCEs by asylum seeker s gender, using Binary Rating dependent variable. presents the regression estimates plotted in this figure. Table S36 Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Female Male 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 33
Fig. S27. AMCEs by asylum seeker s gender, using Scaled Rating dependent variable. presents the regression estimates plotted in this figure. Table S37 Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken Female Male 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Change in rating level 34
Fig. S28. AMCEs by asylum seeker s country of origin, using Forced Choice dependent variable. Table S38 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Syria Kosovo Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 35
Fig. S29. AMCEs by asylum seeker s country of origin, using Binary Rating dependent variable. Table S39 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Syria Kosovo Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 36
Fig. S30. AMCEs by asylum seeker s country of origin, using Scaled Rating dependent variable. Table S40 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Syria Kosovo Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Change in rating level 37
Fig. S31. AMCEs by asylum seeker s age, using Forced Choice dependent variable. Table S41 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies 21 years 38 years 62 years Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 38
Fig. S32. AMCEs by asylum seeker s age, using Binary Rating dependent variable. Table S42 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies 21 years 38 years 62 years Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 39
Fig. S33. AMCEs by asylum seeker s age, using Scaled Rating dependent variable. Table S43 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies 21 years 38 years 62 years Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Change in rating level 40
Fig. S34. AMCEs by asylum seeker s previous occupation, using Forced Choice dependent variable. Table S44 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Unemployed Doctor Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 41
Fig. S35. AMCEs by asylum seeker s previous occupation, using Binary Rating dependent variable. Table S45 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Unemployed Doctor Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Effect on probability of acceptance 42
Fig. S36. AMCEs by asylum seeker s previous occupation, using Scaled Rating dependent variable. Table S46 presents the regression estimates plotted in this figure. Asylum testimony: No inconsistencies Minor inconsistencies Major inconsistencies Female Male Country of origin: Syria Afghanistan Kosovo Eritrea Pakistan Ukraine Iraq 21 years 38 years 62 years Previous occupation: Unemployed Cleaner Farmer Accountant Teacher Doctor Unemployed Doctor PTSD Victim of torture No surviving family Handicapped Reason for migrating: Political persecution Religious persecution Ethnic persecution Economic opportunities Christian Agnostic Muslim Language skills: Fluent Broken 0.2 0.1 0 0.1 0.2 0.2 0.1 0 0.1 0.2 Change in rating level 43