Europeans support a proportional allocation of asylum seekers

In the format provided by the authors and unedited. SUPPLEMENTARY INFORMATION VOLUME: 1 ARTICLE NUMBER: 0133 Europeans support a proportional allocation of asylum seekers Kirk Bansak, 1,2 Jens Hainmueller, 1,2 Dominik Hangartner 1,3,4, 1 Immigration Policy Lab, Stanford University, Stanford, CA 94305, and ETH Zurich, 8092 Zurich, Switzerland 2 Department of Political Science, Stanford University, Stanford, CA 94305 3 Center for Comparative and International Studies, ETH Zurich, 8092 Zurich, Switzerland 4 Department of Government, London School of Economics, London WC2A 2AE, UK e-mail: d.hangartner@lse.ac.uk Contents Supplementary Methods Supplementary Discussion Supplementary Tables Supplementary Figures Supplementary References NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 1

Supplementary Methods Sample We conducted our survey in fifteen European countries that belong to the Common European Asylum System (CEAS). The sample includes Austria, the Czech Republic, Denmark, France, Germany, Greece, Hungary, Italy, the Netherlands, Norway, Poland, Spain, Sweden, Switzerland, and the United Kingdom. This survey was also used by [1], and they provide a detailed description that we summarize and paraphrase below. The international survey firm Respondi recruited respondents from the population of eligible voters in each country. The field period was late February to early March 2016. Respondi recruits respondents for its survey panel using online channels and, to a lesser extent, also computer-assisted telephone interviews. Technical details are provided in the report for the Short-term Campaign Panel of the German Longitudinal Election Study 2009 [2]. New panelists fill out a profiling questionnaire that covers basic sociodemographic information and are then invited to participate in a few surveys. Panelists are compensated for completing the surveys. The modal incentive across all countries in our study was EUR 2.00 for a median length of interview (LOI) of 20.5 minutes. To approximate the population margins for each country, the recruitment of panelists for our survey also employed age and gender quotas. 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. Descriptive statistics are reported in Supplementary Tables 3-5 (reprinting the tables in [1]). Informed consent was obtained from each participant at the beginning of the survey. The survey was approved by Stanford University s Institutional Review Board (protocol ID: 34881) and conducted according to the University of Zurich s policy for human subjects research. Sample Weights As is typical for surveys based on internet panels, for some of the countries included in our study the samples were slightly skewed towards younger and more educated respondents compared to the population of eligible voters. To remove these imbalances, we followed [1] and used entropy balancing [3] to reweight our sample data to match the country-specific demographic margins from the populations. We dropped 147 respondents since data required for constructing the weights was missing. These entropy balancing weights are used for all analyses unless otherwise noted. Supplementary Figures 1, 2, and 3 replicate the main analyses presented in the paper without the entropy balancing weights. The results are substantively indistinguishable from the weighted analyses. Survey Translation We designed the survey instrument in English and subsequently professionally translated and back-translated the questionnaire in each of the country s languages to verify the quality and NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 2

accuracy of the translations. Randomized Design In each country, each respondent was randomly assigned to one of four conditions of the survey: The baseline condition, a condition where respondents are exposed to the information treatment, a condition where respondents are exposed to the consequences treatment, and a condition where respondents are exposed to the both the information treatment and consequences treatment. Supplementary Tables 6-20 provide covariate balance checks across the four conditions for each country and support the successful randomization of respondent assignment to the different conditions. Covariate descriptions can be found in Supplementary Tables 1 and 2. Baseline Condition For all four conditions, respondents were first shown a short introductory text that read as follows: Now, we would like to get your thoughts on policies toward asylum seekers in Europe (i.e. people who left their home countries and request legal safe-haven in Europe on the basis that they fear persecution in their home countries). European countries have adopted common regulations for granting asylum to refugees. We are interested to know your opinions regarding a number of asylum policy options that are currently being debated. For the baseline condition, we then asked respondents the following question to measure what type of allocation mechanism for asylum seekers they prefer: In your opinion, how should the number of asylum applications per country be determined? The number of asylum applications allocated to each European country should be based on the country of first entry (e.g. asylum seekers are required to submit their asylum application in the European country in which they initially arrive). the same for every European country (e.g. asylum seekers are allocated such that each European country receives the same number of asylum applications). proportional to the country s capacity (e.g. asylum seekers are allocated to each European country depending on its population, GDP, unemployment rate, and number of past applications). The baseline condition was designed to measure respondents preferences without exposing them to additional information. Note that each answer option includes explanations to ensure that respondents understand how the various allocations work. Also note that the wording for proportional allocation described in the answer option is directly based on the recent proposal by the European Commission that proposes to allocate asylum seekers based on the countries capacities as measured by the population, total GDP, number of past applications, and unemployment rate [4]. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 3

Information Treatment The information treatment was designed to examine whether preferences change when we provide voters with policy relevant information about the different allocation mechanisms. Respondents assigned to the information treatment received additional information immediately before being asked the question about the preferred allocation. The information prompt read as follows: Under current regulations, asylum seekers are generally required to submit their applications in the country through which they first entered Europe (i.e. the country of first entry ). The goal behind this policy is to maximize efficiency. However, some people have pointed out that the current policy puts an unfair burden on border countries that are more likely to serve as entry points for asylum seekers. Accordingly, they recommend allocating asylum applications either equally across all countries or based on each country s capacity. The information prompt serves two purposes. First, it informs respondents that the country of first entry is, in fact, the current status quo regulation and provides the typical argument used in policy debates to justify this allocation. To the extent that respondents are uninformed about the issue, this might affect attitudes in two directions. On the one hand, if respondents exhibit a typical status quo bias, we would expect that they become more supportive of allocation by the country of first entry as they learn that this is the allocation rule that is currently in place [5, 6]. On the other hand, we might expect that respondents become less supportive of allocation by country of first entry if they enter the survey being concerned about the chaotic situation surrounding the crisis and now learn that country of first entry is the status quo regulation. This might lead them to conclude that the current rule has contributed to the crisis and needs to be changed. The second purpose of the information prompt is to inform respondents about a counterargument to the country-of-first-entry rule that is often advanced in policy debates in favor of a more balanced allocation. If public preferences on this issue are weak or superficial in the absence of substantive policy information, we might then expect that the information prompt will change the responses if voters are particularly swayed by one of the arguments presented. If instead voters already have relatively firm attitudes toward the specific allocation mechanisms or the principles behind them, we might then expect little change to be induced by the information prompt. Consequences Treatment The consequences treatment was designed to examine whether preferences change when we explicitly prime the consequentialist preferences of respondents. Respondents assigned to the consequences treatment received an alternative version of the question about the preferred allocation, in which a sentence was added at the end of each option specifying the associated number of asylum applications. The alternative question, using the example of the United Kingdom, read as follows: In your opinion, how should the number of asylum applications per country be determined? The number of asylum applications allocated to each European country should be NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 4

based on the country of first entry (e.g. asylum seekers are required to submit their asylum application in the European country in which they initially arrive). This would mean approximately 38,700 applications allotted to the United Kingdom. the same for every European country (e.g. asylum seekers are allocated such that each European country receives the same number of asylum applications). This would mean approximately 43,200 applications allotted to the United Kingdom proportional to the country s capacity (e.g. asylum seekers are allocated to each European country depending on its population, GDP, unemployment rate, and number of past applications). This would mean approximately 159,600 applications allotted to the United Kingdom. This randomized manipulation makes explicit what the consequences of the various allocation mechanisms would be in terms of the number of asylum seekers assigned to the respondent s country. This makes it easy for respondents who are driven by consequentialist logic to identify the specific allocation that would minimize the number of asylum seekers for their country. Importantly, to make the survey as realistic as possible, we piped in the actual number of asylum applications reported over the 2015 period by Eurostat for each country as the expected number of asylum seekers reported for the country-of-first-entry answer option. Using the actual empirical numbers for the status quo ensures that we capture the relevant benchmark for any policy reform since these numbers reflect both the current regulations as well as any departures from these rules that the countries might have engaged in. 1 The expected number of asylum seekers under equal and proportional allocation are also based on the empirically observed number of applications. In particular, we computed the total number of current applications for all countries and then multiplied them by 1/15 for the equal allocation option and by a country-specific weight for the proportional allocation option. The country-specific weights are based on the official allocation proposal by the European Commission and based on the following elements: 40% population, 40% total GDP, 10% number of past applications, and 10% unemployment rate [4]. Supplementary Table 27 provides the number of asylum seekers under each allocation for each country. Computing the expected numbers for each allocation in this way adds realism to the survey in that the numbers reflect our best guess of what we can expect to happen if these alternative allocation mechanisms were implemented. Priming the consequentialist preferences of respondents is important in order to examine the strength of the norms of distributive justice. To the extent that consequentialist considerations play an important role in determining public preferences, we expect to see a dramatic shift in attitudes towards the option that gives the lowest number for a specific 1 For example, the numbers listed for Germany reflect the fact that the government made use of the sovereignty clause and unilaterally accepted many asylum seekers in 2015. Note that these numbers only reflect submitted asylum applications and therefore do not capture asylum seekers who might have fled to a country but did not submit their claims yet. In addition, we did not attempt to compute the hypothetical numbers that would occur if the allocation based on country of first entry were perfectly enforced, since no systematic data on the country of first entry is available and also because perfect enforcement does not reflect real-world practice. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 5

country. If instead the norm of equal or proportional allocation is robust, we might expect less of a change in attitudes even if respondents are prompted with the consequences. Note that by making the consequences so readily salient to respondents, the study likely provides an upper bound on the importance of consequentialist considerations in determining public preferences. In reality, we might expect that not all voters will learn about the consequences, and therefore we might not see consequentialist logic being activated as much as when respondents are prompted directly with the consequences as in our survey. In addition, in a real political framing battle voters would likely also be confronted with countervailing primes about fairness to justify the proportional allocation. We do not test such fairness primes here, but they might increase support for proportional allocation. Information and Consequences Treatment In the fourth condition, respondents received both the information treatment and the consequences treatment. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 6

Supplementary Discussion SUPPLEMENTARY INFORMATION This section describes additional analyses referenced in the main study. Unweighted Results Supplementary Figures 1, 2, and 3 replicate the main analyses presented in the paper without the entropy balancing post-stratification weights. The results are substantively indistinguishable from the weighted analyses. Effect of Information Treatment by Knowledge To test whether the information treatment effect was moderated by knowledge, we performed country-specific interaction regressions. For each country, using the sub-samples of respondents assigned to no treatment and only the information treatment, we ran a linear weighted least squares regression of support for proportional allocation on an indicator for receiving the information treatment, our knowledge index measure, and an interaction between the two. Variable descriptions can be found in Supplementary Tables 1 and 2, and Supplementary Table 28 shows the distribution of the knowledge measure. Moderation of the information treatment by knowledge would be captured by the interaction coefficient. Supplementary Table 29 reports the estimates, along with 95% confidence intervals, of the interaction coefficient for each country-specific regression. None of the interaction estimates are statistically or substantively significant. Support for Proportional over Status Quo Allocation of Asylum Seekers, by Mechanism Ranking Supplementary Figure 4 presents the results of an alternative analysis of the respondents preferences on the asylum-seeker allocation mechanism. Immediately after the respondents were asked to specify their most preferred allocation option, they were then asked to rank all three options from most preferred to least preferred. A binary indicator was then constructed, where 1 denotes having selected proportional allocation as more preferred (higher ranked) than the status quo of country of first entry, and 0 otherwise. Supplementary Figure 4 shows the estimated percentage who prefer the proportional allocation mechanism over the status quo of country of first entry, broken out by whether respondents were told the number of asylum seekers allocated to their country under the three allocation mechanisms or not. The estimates employ the entropy balancing weights. The results are similar to the results based on the percentage difference between preference for proportional allocation and preference for the status quo, as presented in Figure 3 in the main text. Support for Proportional Allocation Across Subsets of Respondents Here we examine whether the support for proportional allocation versus the status quo allocation varies across sub-samples of respondents. Supplementary Figure 5 shows the support among respondents at the left, center, and right of the political spectrum when exposed to NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 7

the consequences treatmment or not. The results corroborate the findings for the overall sample. When not shown the numbers of asylum seekers, a large majority of respondents prefers proportional allocation over the status quo. This strong support is weakened but persists even when respondents see the implied numbers. This pattern holds true for all three subsets of respondents at the left, center, and right of the political spectrum. Supplementary Figure 6 replicates the same test for respondents with varying levels of political knowledge, and the results are again similar. Even among voters who are politically uninformed, there remains more support for proportional allocation than the status quo, even when respondents are shown the consequences. Supplementary Figure 7 replicates the same test for respondents who vary in their general support for asylum seekers, and the results are again similar. Even among respondents who support decreasing the number of asylum seekers in general, there is more support for proportional allocation than the status quo, and this persists even when respondents are told about the consequences, which implies a higher number of asylum seekers under proportional allocation for most respondents. Taken together, these results suggest that the support for proportional allocation over the status quo is widely shared among different subgroups of respondents. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 8

Supplementary Tables Supplementary Table 1. Additional variables. This table describes other respondent attitudes and characteristics measured in the survey. Variable Description Age Gender Education Income Employment status Self-reported age. Self-reported gender. Self-reported highest level of education achieved. The customized country-specific educational level questionnaire options used by the European Social Survey were employed for each country, and the respondents educational levels were mapped onto the harmonized European version of the International Standard Classification of Education (EISCED) scale. The scale contains seven major categories: less than lower secondary (1), lower secondary (2), lower tier upper secondary (3), upper tier upper secondary (4), advanced vocational/sub-degree (5), lower tertiary education/ba level (6), and higher tertiary education/ MA level (7). Self-reported household income. Respondents household income levels were measured using the customized country-specific household income decile bins constructed in the European Social Survey (the most recent available version of the European Social Survey was used for each country in the sample). Self-reported employment status. Respondents were asked to answer the following question: Which of these options best describes your situation (in the last seven days)? Answer options included: Paid employee (including temporary leave of absence due to maternity/paternity, accident, illness or vacation), Self-employed (e.g. freelancer, independent contractor, or family-owned business), Student (excluding employer-sponsored education), Unemployed, actively searching for a job, Unemployed, not actively searching, Chronic illness or permanent disability, Retired, and Working at home, caring for children or others. Employed Indicator for whether respondent reported being employed. Variable is coded as 1 if respondents chose Paid employee or Self-employed in the employment status question, and 0 otherwise. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 9

Supplementary Table 2. Additional variables, continued. This table describes other respondent attitudes and characteristics measured in the survey. Variable Description Political ideology Knowledge Index Support for Asylum seekers Self-reported placement on the left-right political ideological spectrum. Ideological placement was measured using a standard self-identification scale ranging from 0 on the left to 10 on the right. The question wording used to measure the respondents ideology was as follows: In politics people often talk of left and right. On this scale from 0 (left) to 10 (right), where would you classify your own political views? For Supplementary Figure 5, political ideology was coarsened into three bins: left (0-4), center (5), and right (6-10). Measures the extent to which respondents are familiar with international affairs, related and unrelated to the asylum crisis. Respondents were presented with three knowledge questions. For each question, a correct answer was scored as a 1, and 0 otherwise. The values for the three questions were summed to create the index. The questions were as follows. (1) Which of the following countries was the most common country of origin for asylum seekers in Europe in 2014? Answer choices were: Syria (correct), Ukraine, Iran, Georgia, Honduras, China, and Sri Lanka. (2) From what you have read or heard, what is a tariff? Answer choices were: A tax exemption for low earners, A tax exemption for buyers of locally produced goods, A tax imposed on imported goods and services (correct), A tax imposed on luxury goods, and Don t know. (3) What is the name of the European Union law that determines the EU Member State responsible for examining an application for asylum seekers? Answer choices were: Dublin Regulation (correct), Rome Regulation, Paris Regulation, Vienna Regulation, Prague Regulation, Berlin Regulation, Madrid Regulation, Lisbon Regulation, and Maastricht Regulation. Self-reported preference for increasing or decreasing the number of asylum seekers granted asylum in respondent s home country. Respondents were asked: Do you think [RESPONDENT s COUNTRY] should increase or decrease the number of people it grants asylum to? Answer options included: Greatly increase, Increase, Neither increase nor decrease, Decrease, and Greatly decrease. The analysis presented in Figure 1 of the main text is based on a collapsed binary version of this variable where 1 indicates selection of Greatly increase or Increase, and 0 otherwise. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 10

Supplementary Table 3. Number of respondents per country. Sample Size Austria 1206 Czech Republic 1202 Denmark 1201 France 1203 Germany 1200 Greece 1200 Hungary 1200 Italy 1200 Netherlands 1200 Norway 1202 Poland 1201 Spain 1203 Sweden 1203 Switzerland 1208 United Kingdom 1201 Total 18030 Note: Main analyses presented in the paper drop 147 respondents for whom weights could not be constructed due to missing data. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 11

Supplementary Table 4. Summary statistics. This table displays the distribution of respondents along key demographic and ideological dimensions, by country. Gender Age Income Quintile Political Ideology Female 29/under 30-39 40-49 50-59 60/over First Second Third Fourth Fifth Left Center Right Austria 50% 22% 19% 24% 18% 17% 29% 24% 22% 17% 8% 34% 37% 29% Czech Republic 51% 42% 28% 16% 9% 5% 15% 19% 23% 28% 15% 22% 43% 35% Denmark 50% 20% 21% 22% 20% 17% 26% 20% 25% 16% 13% 37% 25% 38% France 51% 23% 20% 21% 20% 15% 29% 23% 17% 23% 9% 30% 35% 36% Germany 50% 20% 18% 25% 21% 16% 25% 23% 21% 20% 11% 41% 36% 24% Greece 47% 33% 29% 25% 10% 3% 37% 28% 19% 11% 5% 36% 36% 28% Hungary 51% 43% 28% 14% 8% 6% 18% 20% 15% 21% 26% 19% 48% 33% Italy 50% 19% 22% 23% 19% 17% 27% 26% 22% 18% 7% 35% 26% 39% Netherlands 50% 22% 17% 23% 21% 17% 30% 22% 19% 19% 9% 28% 33% 39% Norway 49% 22% 22% 21% 19% 16% 28% 26% 19% 13% 13% 34% 24% 42% Poland 49% 19% 30% 25% 15% 11% 15% 22% 20% 24% 20% 23% 39% 38% Spain 50% 22% 25% 22% 17% 14% 23% 25% 18% 16% 17% 51% 23% 27% Sweden 49% 22% 20% 21% 19% 18% 22% 15% 20% 18% 25% 37% 26% 37% Switzerland 50% 22% 20% 23% 19% 16% 33% 29% 20% 14% 4% 29% 30% 41% United Kingdom 52% 26% 19% 21% 18% 16% 29% 21% 19% 19% 12% 27% 43% 30% NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 12

Supplementary Table 5. Summary statistics, continued. See Section?? for variable descriptions. Lower Secondary and Below corresponds to EISCED 1-2, Upper Secondary EISCED 3-4, Advanced Vocational EISCED 5, BA Level EISCED 6, and MA Level and Above EISCED 7. Education Lower Upper Advanced BA Level MA Level Secondary Secondary Vocational and Above and Below Austria 8% 53% 20% 5% 13% Czech Republic 5% 56% 15% 10% 14% Denmark 22% 37% 10% 22% 8% France 10% 45% 21% 9% 16% Germany 5% 38% 29% 15% 13% Greece 4% 32% 19% 29% 16% Hungary 4% 49% 18% 19% 9% Italy 13% 48% 4% 11% 23% Netherlands 32% 34% 7% 12% 15% Norway 13% 35% 17% 23% 12% Poland 9% 35% 13% 14% 29% Spain 23% 19% 16% 17% 25% Sweden 12% 38% 28% 11% 11% Switzerland 10% 50% 24% 7% 9% United Kingdom 17% 37% 14% 21% 11% NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 13

Supplementary Table 6. Balance Check, Austria. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 292 296 316 302 Female 0.56 0.49 0.48 0.47 0.11 Age 42.08 43.77 42.97 42.94 0.57 (0.85) (0.87) (0.8) (0.83) Employed 0.61 0.63 0.66 0.63 0.75 Education 4.05 4.07 4.16 4.16 0.72 (0.09) (0.09) (0.09) (0.09) Income Decile 4.17 4.5 4.64 4.47 0.16 (0.15) (0.15) (0.15) (0.15) Political Ideology 4.75 4.79 4.67 5.04 0.12 (0.11) (0.12) (0.11) (0.12) Knowledge Index 2.04 1.99 2.05 2.13 0.24 (0.05) (0.05) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 14

Supplementary Table 7. Balance Check, Czech Republic. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 290 302 284 326 Female 0.51 0.54 0.52 0.47 0.31 Age 34.45 35.63 34.48 33.67 0.25 (0.73) (0.74) (0.68) (0.67) Employed 0.61 0.66 0.6 0.58 0.16 Education 4.39 4.66 4.44 4.45 0.10 (0.08) (0.08) (0.09) (0.07) Income Decile 5.73 5.66 5.67 5.63 0.97 (0.15) (0.15) (0.16) (0.14) Political Ideology 5.3 5.56 5.31 5.25 0.21 (0.11) (0.12) (0.12) (0.11) Knowledge Index 1.69 1.72 1.73 1.63 0.31 (0.04) (0.04) (0.04) (0.04) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 15

Supplementary Table 8. Balance Check, Denmark. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 313 296 300 292 Female 0.53 0.53 0.44 0.49 0.08 Age 43.64 43.09 44.74 43.76 0.56 (0.81) (0.83) (0.81) (0.86) Employed 0.53 0.48 0.5 0.53 0.59 Education 4.11 4.13 4.04 4.12 0.90 (0.1) (0.1) (0.1) (0.1) Income Decile 4.84 4.82 5.12 4.76 0.39 (0.16) (0.16) (0.16) (0.16) Political Ideology 5.2 4.97 4.93 5.08 0.48 (0.14) (0.14) (0.14) (0.12) Knowledge Index 1.44 1.43 1.49 1.39 0.55 (0.05) (0.05) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 16

Supplementary Table 9. Balance Check, France. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 300 320 287 296 Female 0.5 0.51 0.5 0.54 0.74 Age 42.38 44.26 42.87 41.74 0.14 (0.82) (0.79) (0.81) (0.81) Employed 0.65 0.62 0.59 0.62 0.55 Education 4.49 4.32 4.45 4.54 0.36 (0.09) (0.09) (0.09) (0.09) Income Decile 4.53 4.54 4.87 4.66 0.39 (0.16) (0.15) (0.16) (0.16) Political Ideology 5.25 5.17 5.1 5.2 0.90 (0.14) (0.13) (0.14) (0.14) Knowledge Index 1.47 1.48 1.49 1.51 0.87 (0.04) (0.04) (0.04) (0.04) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 17

Supplementary Table 10. Balance Check, Germany. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 314 284 273 329 Female 0.5 0.51 0.52 0.47 0.67 Age 44.23 43.68 44.28 44.13 0.95 (0.75) (0.82) (0.85) (0.78) Employed 0.63 0.69 0.64 0.62 0.36 Education 4.46 4.53 4.35 4.58 0.32 (0.09) (0.09) (0.09) (0.09) Income Decile 4.89 4.91 4.88 4.75 0.88 (0.16) (0.16) (0.17) (0.15) Political Ideology 4.33 4.52 4.52 4.64 0.21 (0.11) (0.11) (0.11) (0.1) Knowledge Index 1.75 1.82 1.78 1.88 0.26 (0.05) (0.05) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 18

Supplementary Table 11. Balance Check, Greece. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 290 298 314 298 Female 0.42 0.43 0.53 0.5 0.02 Age 35.61 37.01 35.77 35.45 0.31 (0.64) (0.64) (0.66) (0.66) Employed 0.59 0.55 0.51 0.54 0.30 Education 5.18 5.12 5.13 5.03 0.62 (0.08) (0.08) (0.07) (0.08) Income Decile 3.79 3.9 3.67 3.85 0.65 (0.14) (0.14) (0.13) (0.15) Political Ideology 4.73 4.83 4.68 4.77 0.84 (0.13) (0.11) (0.12) (0.12) Knowledge Index 1.89 1.96 1.83 1.85 0.29 (0.05) (0.05) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 19

Supplementary Table 12. Balance Check, Hungary. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 290 291 302 317 Female 0.53 0.5 0.5 0.5 0.82 Age 34.63 34.35 33.3 34.92 0.40 (0.72) (0.76) (0.69) (0.71) Employed 0.6 0.57 0.56 0.63 0.24 Education 4.54 4.56 4.65 4.68 0.48 (0.07) (0.08) (0.07) (0.08) Income Decile 5.67 5.77 5.99 6.13 0.23 (0.18) (0.18) (0.17) (0.17) Political Ideology 5.41 5.31 5.39 5.49 0.76 (0.12) (0.12) (0.11) (0.12) Knowledge Index 1.98 1.9 2.06 2.03 0.07 (0.04) (0.05) (0.05) (0.04) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 20

Supplementary Table 13. Balance Check, Italy. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 310 282 307 301 Female 0.51 0.54 0.47 0.49 0.44 Age 43.65 42.72 45.39 43.9 0.11 (0.75) (0.81) (0.77) (0.78) Employed 0.62 0.53 0.62 0.57 0.08 Education 4.66 4.79 4.67 4.51 0.23 (0.1) (0.1) (0.1) (0.09) Income Decile 4.67 4.48 4.42 4.35 0.44 (0.14) (0.15) (0.15) (0.14) Political Ideology 5.11 4.89 5.15 5.03 0.63 (0.15) (0.15) (0.14) (0.15) Knowledge Index 1.82 1.8 1.82 1.76 0.77 (0.05) (0.05) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 21

Supplementary Table 14. Balance Check, Netherlands. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 311 280 293 316 Female 0.47 0.49 0.5 0.54 0.25 Age 44.42 44.2 44.71 42.87 0.40 (0.8) (0.89) (0.83) (0.83) Employed 0.53 0.49 0.52 0.47 0.50 Education 4.02 3.9 3.76 3.92 0.40 (0.11) (0.11) (0.11) (0.11) Income Decile 4.91 4.49 4.53 4.48 0.16 (0.15) (0.16) (0.16) (0.16) Political Ideology 5.38 5.05 5.08 5.39 0.07 (0.11) (0.13) (0.12) (0.12) Knowledge Index 1.22 1.26 1.11 1.22 0.07 (0.04) (0.05) (0.04) (0.04) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 22

Supplementary Table 15. Balance Check, Norway. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 301 296 280 325 Female 0.49 0.5 0.51 0.46 0.62 Age 42.4 42.77 42.65 43.37 0.86 (0.81) (0.8) (0.89) (0.83) Employed 0.61 0.6 0.53 0.56 0.19 Education 4.6 4.51 4.42 4.59 0.50 (0.09) (0.09) (0.1) (0.09) Income Decile 4.83 4.49 4.44 4.6 0.33 (0.16) (0.16) (0.17) (0.16) Political Ideology 5.21 5.29 5.17 5.31 0.85 (0.14) (0.12) (0.14) (0.13) Knowledge Index 1.35 1.31 1.22 1.25 0.18 (0.05) (0.05) (0.05) (0.04) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 23

Supplementary Table 16. Balance Check, Poland. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 331 299 296 275 Female 0.52 0.56 0.43 0.45 <0.01 Age 41.14 41.61 40.97 41.36 0.93 (0.7) (0.73) (0.73) (0.77) Employed 0.67 0.61 0.68 0.69 0.14 Education 5.07 4.78 4.83 4.99 0.14 (0.09) (0.1) (0.1) (0.1) Income Decile 5.79 5.53 5.63 6.14 0.04 (0.15) (0.16) (0.16) (0.16) Political Ideology 5.42 5.39 5.55 5.67 0.42 (0.12) (0.13) (0.13) (0.14) Knowledge Index 1.37 1.43 1.35 1.38 0.66 (0.05) (0.05) (0.04) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 24

Supplementary Table 17. Balance Check, Spain. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 305 322 304 272 Female 0.51 0.48 0.52 0.5 0.82 Age 42.54 42.04 41.72 41.79 0.87 (0.76) (0.73) (0.77) (0.8) Employed 0.59 0.66 0.59 0.61 0.23 Education 4.64 4.69 4.62 4.54 0.84 (0.12) (0.11) (0.11) (0.13) Income Decile 5.17 5.07 5.19 4.93 0.67 (0.16) (0.15) (0.16) (0.18) Political Ideology 4.32 4.07 4.18 4.23 0.64 (0.14) (0.14) (0.15) (0.15) Knowledge Index 1.6 1.6 1.61 1.57 0.93 (0.04) (0.04) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 25

Supplementary Table 18. Balance Check, Sweden. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 295 301 295 312 Female 0.45 0.5 0.55 0.47 0.10 Age 43.91 42.74 43.81 44.08 0.66 (0.85) (0.83) (0.85) (0.81) Employed 0.59 0.61 0.58 0.6 0.91 Education 4.4 4.58 4.36 4.41 0.33 (0.09) (0.09) (0.09) (0.09) Income Decile 5.77 5.73 5.68 5.57 0.86 (0.18) (0.18) (0.17) (0.17) Political Ideology 5.08 4.87 4.79 5.02 0.43 (0.15) (0.14) (0.14) (0.13) Knowledge Index 2.08 2.05 2.01 2.07 0.67 (0.04) (0.04) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 26

Supplementary Table 19. Balance Check, Switzerland. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 302 301 315 290 Female 0.54 0.51 0.48 0.47 0.31 Age 42.71 43.63 43.85 42.13 0.43 (0.86) (0.8) (0.81) (0.86) Employed 0.66 0.68 0.61 0.61 0.18 Education 4.36 4.51 4.38 4.4 0.44 (0.07) (0.08) (0.08) (0.07) Income Decile 3.98 4.21 4.07 3.82 0.25 (0.13) (0.14) (0.14) (0.14) Political Ideology 5.4 5.38 5.27 5.28 0.83 (0.12) (0.13) (0.13) (0.13) Knowledge Index 2.01 1.96 1.95 2.05 0.45 (0.05) (0.05) (0.05) (0.05) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 27

Supplementary Table 20. Balance Check, United Kingdom. Table shows covariate means (with standard errors) for respondents randomly assigned to each of the four survey treatment conditions, with p values from ANOVA tests of equal means. Condition A denotes the baseline condition, Condition B denotes assignment to only the information treatment, Condition C denotes assignment to only the consequences treatment, and Condition D denotes assignment to both treatments. Variable Condition A Condition B Condition C Condition D ANOVA pval N 323 308 291 279 Female 0.55 0.52 0.49 0.52 0.62 Age 43.46 43.09 43.09 41.58 0.42 (0.79) (0.81) (0.91) (0.86) Employed 0.58 0.6 0.56 0.59 0.73 Education 4.29 4.44 4.39 4.32 0.67 (0.09) (0.1) (0.1) (0.1) Income Decile 4.76 4.84 4.66 4.96 0.63 (0.16) (0.16) (0.17) (0.17) Political Ideology 5.19 5.04 5.17 4.95 0.44 (0.11) (0.12) (0.12) (0.13) Knowledge Index 1.3 1.33 1.35 1.39 0.58 (0.04) (0.05) (0.05) (0.04) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 28

Supplementary Table 21. Public Support for Increasing the Number of Asylum seekers. Table displays the weighted estimates (with standard errors) of the percentage support for increasing the number of asylum seekers in each of the fifteen surveyed countries, as diplayed in Figure 1 in the main text. Country Estimate (SE) Austria 16.95 (1.35) Czech Republic 10.78 (1.41) Denmark 18.65 (1.23) France 18.96 (1.43) Germany 14.10 (1.14) Greece 16.99 (2.04) Hungary 19.58 (1.95) Italy 17.15 (1.50) Netherlands 20.29 (1.34) Norway 23.78 (1.35) Poland 17.45 (1.39) Spain 39.46 (1.73) Sweden 19.26 (1.35) Switzerland 21.31 (1.70) United Kingdom 20.89 (1.34) Pooled 19.78 (0.39) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 29

Supplementary Table 22. Public Support for Various Allocations of Asylum Seekers, Baseline Condition. Table shows the weighted estimates of percentage support for each of the three allocation options in the baseline condition, as diplayed in the top left panel of Figure 2 in the main text. Country Proportional Same for All Status Quo Germany 57.92 20.88 21.20 Hungary 76.22 4.57 19.22 Sweden 64.21 17.13 18.66 Austria 69.75 6.64 23.61 Norway 71.99 4.53 23.48 Switzerland 79.34 8.20 12.46 Denmark 72.12 8.03 19.85 Netherlands 70.32 10.89 18.78 Greece 87.38 7.77 4.85 Czech Republic 68.22 2.46 29.32 Italy 78.74 11.85 9.42 Poland 72.14 7.44 20.42 Spain 80.08 11.71 8.21 France 76.42 11.76 11.83 United Kingdom 60.79 11.51 27.70 Pooled 72.07 9.74 18.19 NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 30

Supplementary Table 23. Public Support for Various Allocations of Asylum Seekers, Information Treatment. Table shows the weighted estimates of percentage support for each of the three allocation options given only the information treatment, as diplayed in the top right panel of Figure 2 in the main text. Country Proportional Same for All Status Quo Germany 62.57 23.21 14.22 Hungary 74.03 5.78 20.19 Sweden 62.38 18.18 19.44 Austria 68.31 12.68 19.00 Norway 64.57 9.40 26.03 Switzerland 73.13 9.73 17.15 Denmark 65.44 9.85 24.71 Netherlands 75.04 9.42 15.54 Greece 86.91 11.49 1.60 Czech Republic 61.18 2.99 35.83 Italy 64.15 25.77 10.08 Poland 72.49 7.22 20.29 Spain 72.77 14.44 12.80 France 74.56 13.26 12.18 United Kingdom 57.58 15.44 26.98 Pooled 68.83 12.38 18.79 NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 31

Supplementary Table 24. Public Support for Various Allocations of Asylum Seekers, Consequences Treatment. Table shows the weighted estimates of percentage support for each of the three allocation options given only the consequences treatment, as diplayed in the bottom left panel of Figure 2 in the main text. Country Proportional Same for All Status Quo Germany 68.73 24.26 7.00 Hungary 91.27 4.68 4.05 Sweden 71.80 17.97 10.24 Austria 88.99 6.47 4.54 Norway 74.97 9.34 15.69 Switzerland 62.04 5.37 32.59 Denmark 49.37 4.51 46.12 Netherlands 45.24 35.21 19.56 Greece 85.04 2.43 12.54 Czech Republic 29.08 2.04 68.88 Italy 42.47 42.98 14.55 Poland 33.99 8.64 57.37 Spain 45.46 18.83 35.70 France 47.80 41.24 10.95 United Kingdom 31.03 15.80 53.17 Pooled 57.53 16.01 26.47 NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 32

Supplementary Table 25. Public Support for Various Allocations of Asylum Seekers, Both Treatments. Table shows the weighted estimates of percentage support for each of the three allocation options given both the information and consequences treatments, as diplayed in the bottom right panel of Figure 2 in the main text. Country Proportional Same for All Status Quo Germany 67.72 23.24 9.04 Hungary 87.62 3.92 8.45 Sweden 74.26 14.17 11.57 Austria 79.07 11.28 9.65 Norway 73.71 6.78 19.51 Switzerland 58.11 5.07 36.82 Denmark 54.75 5.06 40.19 Netherlands 42.84 40.34 16.82 Greece 71.99 4.14 23.88 Czech Republic 25.25 1.07 73.68 Italy 35.22 49.88 14.90 Poland 36.03 11.21 52.76 Spain 47.39 19.66 32.95 France 44.11 44.12 11.76 United Kingdom 27.95 20.21 51.84 Pooled 55.36 17.68 26.96 NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 33

Supplementary Table 26. Support for Proportional versus Status Quo Allocation of Asylum Seekers. Table shows weighted estimates (with standard errors) of difference in the percentage support for proportional allocation versus status quo allocation by country of first entry given random assignment to the consequences treatment, as displayed in Figure 3 in the main text. Countries are ordered such that the country at the bottom (top) would see the largest increase (decrease) in the number of asylum seekers when moving from the status quo to proportional allocation, and the line separates the countries that would see an increase versus a decrease. Estimates employ sample weights. Differences between the estimates with and without the consequences treatment are statistically significant at p<0.001 for all countries except Norway and Greece, which have p values of 0.012 and 0.002 respectively. Country Consequences not Shown Consequences Shown Germany 42.33 (4.13) 60.07 (3.38) Hungary 55.38 (4.99) 82.91 (3.14) Sweden 44.21 (4.03) 62.15 (3.17) Austria 47.73 (4.31) 76.94 (2.66) Norway 43.61 (3.76) 56.51 (3.57) Switzerland 61.34 (3.82) 25.64 (5.02) Denmark 46.70 (3.71) 8.82 (4.45) Netherlands 55.32 (3.35) 25.86 (3.37) Greece 84.03 (3.36) 61.32 (6.33) Czech Republic 31.74 (6.34) -44.43 (4.98) Italy 62.37 (3.74) 24.05 (3.97) Poland 51.95 (3.62) -20.17 (4.76) Spain 66.00 (3.06) 11.94 (4.57) France 63.41 (3.60) 34.52 (3.70) United Kingdom 31.87 (4.02) -22.97 (4.31) Pooled 51.96 (1.08) 29.71 (1.20) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 34

Supplementary Table 27. Number of Asylum-seeker Applications Implied by Each Allocation Method Country Status Quo Same for All Proportional Germany 476510 43223 213416 Hungary 177130 43223 21162 Sweden 162455 43223 34025 Austria 88233 43223 30705 Norway 31115 43223 30330 Switzerland 37729 43223 39749 Denmark 19931 43223 23928 Netherlands 44975 43223 50622 Greece 12976 43223 22268 Czech Republic 1500 43223 36376 Italy 84085 43223 137483 Poland 12185 43223 66528 Spain 13727 43223 107192 France 70680 43223 164177 United Kingdom 38700 43223 159613 Note: Data for status quo is based on Eurostat 2015 Asylum and first time asylum applicants by citizenship, age and sex (accessed 02/05/2016). The same for all numbers and proportional allocation numbers are computed by the authors using, respectively, an equal weight for each country and country-specific weights proposed by the European Commission [4]. NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 35

Supplementary Table 28. Percent to Correctly Answer Knowledge Questions Syria Question Dublin Question Tariff Question Austria 73 60 72 Czech Republic 57 16 96 Denmark 77 27 39 France 74 8 67 Germany 77 43 60 Greece 71 42 75 Hungary 85 36 79 Italy 73 28 79 Netherlands 77 14 30 Norway 73 38 17 Poland 63 11 65 Spain 73 13 73 Sweden 83 39 83 Switzerland 78 50 71 United Kingdom 70 11 53 Pooled 74 29 64 NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 36

Supplementary Table 29. Estimates of Interaction between Information Treatment and Knowledge, by Country. 95% confidence intervals are based on heteroskedasticy-robust regression estimates of the variances. Country Interaction Estimate 95% CI Germany -0.036 [-0.142, 0.07] Hungary 0.085 [-0.05, 0.219] Sweden -0.036 [-0.165, 0.094] Austria -0.018 [-0.133, 0.096] Norway 0.081 [-0.021, 0.183] Switzerland 0.024 [-0.084, 0.131] Denmark -0.014 [-0.109, 0.082] Netherlands -0.001 [-0.097, 0.095] Greece -0.051 [-0.19, 0.088] Czech Republic 0.012 [-0.172, 0.195] Italy 0.025 [-0.083, 0.133] Poland 0.005 [-0.098, 0.108] Spain 0.025 [-0.083, 0.132] France 0.082 [-0.047, 0.21] United Kingdom 0.020 [-0.096, 0.137] NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 37

Supplementary Figures Supplementary Figure 1. Public Support for Increasing the Number of Asylum seekers (Unweighted). Percent of respondents that support increasing the number of asylum seekers in each of the fifteen surveyed countries. Corresponding normality-based 95% CI are shown. Pooled N = 17, 883. Austria (n=1,202) Czech Republic (n=1,200) Denmark (n=1,192) France (n=1,199) Germany (n=1,130) Greece (n=1,200) Hungary (n=1,195) Italy (n=1,196) Netherlands (n=1,195) Norway (n=1,190) Poland (n=1,196) Spain (n=1,203) Sweden (n=1,183) Switzerland (n=1,201) United Kingdom (n=1,201) Pooled (n=17,883) 0 25 50 75 100 Want More Asylum Seekers in Own Country (%) NATURE HUMAN BEHAVIOUR DOI: 10.1038/s41562-017-0133 www.nature.com/nathumbehav 38