Not as innocent as it seems? The effects of "neutral" messaging on refugee attitudes

Hillenbrand, Tobias; Martorano, Bruno; Siegel, Melissa Working Paper Not as innocent as it seems? The effects of "neutral" messaging on refugee attitudes UNU-MERIT Working Papers, No. 2025-011 Provided in Cooperation with: Maastricht Economic and Social Research Institute on Innovation and Technology (UNU-MERIT), United Nations University (UNU) Suggested Citation: Hillenbrand, Tobias; Martorano, Bruno; Siegel, Melissa (2025) : Not as innocent as it seems? The effects of "neutral" messaging on refugee attitudes, UNU-MERIT Working Papers, No. 2025-011, United Nations University (UNU), Maastricht Economic and Social Research Institute on Innovation and Technology (UNU-MERIT), Maastricht, https://doi.org/10.53330/YTID6699 This Version is available at: https://hdl.handle.net/10419/326940 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-sa/4.0/ #2025-011 Not as innocent as it seems? The effects of "neutral" messaging on refugee attitudes Tobias Hillenbrand, Bruno Martorano and Melissa Siegel Published 16 April 2025 DOI: https://www.doi.org/10.53330/YTID6699 Maastricht Economic and social Research institute on Innovation and Technology (UNU-MERIT) email: [email protected] | website: http://www.merit.unu.edu Boschstraat 24, 6211 AX Maastricht, The Netherlands Tel: (31) (43) 388 44 00 UNU-MERIT Working Papers ISSN 1871-9872 Maastricht Economic and social Research Institute on Innovation and Technology UNU-MERIT UNU-MERIT Working Papers intend to disseminate preliminary results of research carried out at UNU-MERIT to stimulate discussion on the issues raised. https://creativecommons.org/licenses/by-nc-sa/4.0/ Not as innocent as it seems? The effects of “neutral” messaging on refugee attitudes Tobias Hillenbranda,b,1, Bruno Martoranoa,b, Melissa Siegela,b aUnited Nations University-MERIT, Boschstraat 24, 6211 AX Maastricht, The Netherlands bMaastricht University, Minderbroedersberg 4-6, 6211 LK Maastricht, The Netherlands Abstract Immigration has become one of the most divisive political issues in Europe and around the world. In Germany, Europe’s largest refugee hosting country, public attitudes have reached a low point. Besides increased “real-life” exposure to immigrants, exposure to all sorts of messages centered around immigration and refugees may be behind this worrying trend. While prior research has investigated the effects of specific subjects of the immigration discourse, such as specific frames or statistical information, it remains unclear how “neutral” reporting on refugee migration impacts public attitudes. We fill this gap using data from an original survey experiment conducted in Germany in May 2023. The findings suggest that a sober (neutral) video providing basic background information on Syrian refugees reduces humanitarian concerns for this refugee group, increases the perception of security threats and lowers the willingness to support refugee camps abroad. The results are driven by West German residents. Qualitative data reveals that, although the video is indeed perceived as “neutral”, it triggers security-related associations among West Germans, seemingly eroding concerns for refugees’ wellbeing. Conversely, East Germans, while starting from a slightly more negative base level, more frequently express indifference. Finally, merging our survey data with administrative data on the foreign population in respondents’ counties reveals that larger percentage increases in real-life immigration exposure mitigate the treatment effect. JEL classifications: A13, D63, D83, J15 Keywords: Migration attitudes, migration discourse, refugees; humanitarianism; threat perceptions; survey experiment 1 Corresponding author at: United Nations University-MERIT, Netherlands Email addresses: hillen[email protected]u.edu, [email protected], [email protected] 1 1) Introduction In many European countries, the issue of asylum immigration has shaped the political agenda and discourse more than most other issues in recent years. This was accompanied by a rise of far-right groups, such as the AfD party in Germany, which doubled its vote share in the 2025 legislative elections. Exit polls reveal that the voters in Europe’s largest refugee destination country perceived “immigration, asylum and integration” as the most important problem (ZDF, n.d.). This is in line with an analysis by Wieland (2024), suggesting that attitudes towards immigration are particularly negative in the aftermath of large inflows of asylum seekers and refugees as was the case in 2015/2016 and 2022/2023. Figure 1 illustrates the substantial fluctuations in the proportion of Germans skeptical of admitting more refugees since 2015. This raises questions about the drivers of attitudes towards (humanitarian) immigration, particularly in times of large-scale influxes. Figure 1: Percentage of Germans against the admission of more refugees Source: Authors’ illustration based on data from Wieland (2024) In light of the existing literature, we distinguish between two explanatory approaches: “reallife” exposure as well as messaging exposure. Real-life exposure relates to (changing) characteristics in an individual’s environment due to the presence or arrival of immigrants. This includes increased exposure to and opportunities for contact with immigrants as a direct consequence, but may also include more indirect implications such as increased competition Percent 2 for housing (Unal et al., 2024). Implications of immigration can relate to, for example, a country’s economy, cultural life and security situation and are not necessarily negative.2 A recent study by Dražanová & Gonnot (2023) points to differences in the relationship between real-life exposure and attitudes depending on how exposure is measured: On the one hand, a larger regional share of migrants is associated with more migration-friendly attitudes. Conversely, short-term increases in the share of the foreign-born population tend to fuel resistance. These dynamics can be linked to two lines of argument in the literature revolving either around positive effects of intergroup contact, at least under favorable conditions (Allport, 1954; Pettigrew & Tropp, 2006), or around perceptions of immigration as a threat and unwanted competition for scarce resources (Blalock, 1967; Bobo, 1999; Stephan et al., 2009). The alternative explanation is based on the insight that immigration attitudes are formed based on perceptions rather than hard facts (Wong, 2007). This suggests that the various messages around immigration that people are exposed to, for example via the public discourse, may also matter for how they think and feel about the topic. Typically, these studies examining messaging effects included specific elements, such as threat or humanitarian frames or information on immigrant numbers, that were intended to sway people’s attitudes in one direction or another. However, a lot of messages in the public discourse are not aimed at influencing people’s views but rather at, for example, providing basic background information on current or past developments, using a neutral tone. Indeed, calls for a calmer and more objective discourse have become commonplace. However, researchers have so far sidelined the question of how people respond to messaging that is designed to be neutral and sober. In fact, the literature suggests that messaging can impact people’s attitudes not only through its content but also by merely increasing the salience of the issue. Existing publications have largely translated issue salience into the prevalence of migration coverage in the media and produced somewhat ambiguous results with regard to effects of issue salience. These studies were predominantly based on observational data though, raising a number of endogeneity concerns3 and did not look at refugee and asylum migration in particular. However, this form of migration is especially politically sensitive and research suggests that the formation of 2 In particular, economic and cultural effects can also be (perceived as) positive contributions (Maurer et al., 2021). 3 A summary of existing findings and of the challenges to causal identification is discussed in Section 2. 3 attitudes differs systematically between the contexts of humanitarian migration and other migration forms (Abdelaaty & Steele, 2022; Fraser & Murakami, 2022). This research is meant to fill this gap by investigating the question of how exposure to a neutrally designed video on refugees impacts humanitarian concerns, threat perceptions and policy preferences towards refugees. This is particularly important to investigate given that some media coverage aims to be as neutral as possible in messaging with the hopes to “only show the facts”. Additionally, researchers and academics are often under the impression that all they need to do is give neutral facts to properly form opinions on migration. We approach this question by means of a survey experiment in Germany, in which we employed a short professionally produced video clip about Syrian refugees in Turkey whose content does not contain any elements to influence attitudes in one way or another. Thanks to follow-up data collection, we can provide additional insights into the cognitive and emotional reactions of respondents, which sheds light on the mechanisms underlying our treatment effects. Finally, we merge our survey data with administrative data on the foreign population in German counties (Kreise) to explore interactions between our treatment of exposure to migration messaging with a proxy for real-life migration exposure. Our results demonstrate that even exposure to a video designed to speak about refugee migration in a calm and neutral way leads to eroding concerns for the wellbeing of Syrian refugees and drives up perceptions of a security threat. Further, we find weak evidence that video exposure reduces willingness to sign a petition to support refugee camps abroad. Our data shows that the increased negativity through treatment exposure is driven by West Germans and less pronounced in counties that experienced large percentage increases in their foreign populations in the years prior to the data collection. Data from a second data collection round confirms that West Germans are more likely to make associations with the security situation in Germany, whereas East Germans appear to be more indifferent when seeing the video. Moreover, it provides evidence that the video was not just intended to be neutral and sober but was also perceived as such and that the impact on attitudes is due to increased salience rather than specific contents in the video. This research makes various contributions to the literature on migration messaging and migration attitudes. First, our interest in the effect of neutral messaging deviates from existing studies that were mainly centered around framing interventions or the updating of (biased) beliefs. Further, we speak to the limited literature on salience effects to which we add a new 4 angle by focusing specifically on humanitarian migration, and we advance this literature strand methodologically by providing experimental evidence to complement the existing observational studies. In addition, we provide a deeper understanding of the underlying mechanisms of our intervention through an additional data collection specifically tailored for this purpose. Finally, we further illuminate the interactions between messaging effects and actual migration patterns. 2) Literature review 2.1) Public discourse and migration attitudes in Germany Generally, refugees are considered somewhat ambivalent figures since they are at the same time associated with vulnerability as well as potential threats. This pattern applies to both public perceptions (Adida et al., 2019; Fraser & Murakami, 2022; Jeannet et al., 2021) and media depictions (Chouliaraki & Stolic, 2017; Maurer et al., 2021; McCann et al., 2023). In 2015, the salience of the immigration topic, particularly of irregular immigration, grew in tandem with the increasing number of asylum seeker arrivals in European countries (Heidenreich et al., 2019; Maurer et al., 2019). Various studies analyzing the European media landscape at the time document the prevalence of different frames for these new arrivals, such as threat or victim frames (Chouliaraki & Stolic, 2017; Greussing & Boomgaarden, 2017; Heidenreich et al., 2019). However, there were changes over time: for example, Chouliaraki et al. (2017, p. 2) state that the sympathetic climate in the summer months of 2015 was “replaced by suspicion and, in some cases, hostility towards refugees and migrants”. For Germany, the prevalence of a “human interest” frame was relatively widespread and expanding until the late summer of 2015, when it began to decline. Conversely, news coverage linking migrants with crime and terrorism was initially low but increased steeply from autumn 2015 onwards (Heidenreich et al., 2019). In an analysis of six German leading media outlets, both print and television4, Maurer et al. (2019) finds that only 2 percent of immigration-related news items made links with criminality in 2015. In January 2016 though, following the New Year Eve events in Cologne, where a large number of harassment incidents occurred, mostly involving suspects with migration backgrounds, this proportion skyrocketed to 26 percent. This also coincided with a tonality change in the coverage of immigrants, which was quite positive in the German media in 2015 but turned much more negative in January 2016 (Maurer et al., 4 The study did not cover all items related to immigration in general, but only those linked to immigration from countries that played a relevant role in the 2015/16 “migration crisis”, such as Syria or Afghanistan. 5 2019). And while the individual immigrants were portrayed largely positively in 2015, and particularly the vulnerability and need of Syrian refugees was widely recognized (McCann et al., 2023), the large-scale immigration as a more abstract phenomenon was mostly depicted as a challenge rather than an opportunity (Maurer et al., 2019). In line with these developments in the media, Czymara & Schmidt-Catran (2017) revealed a significant decline in the acceptance towards immigrants among Germans between April 2015 and February 2016. This decline was particularly pronounced for immigrants from the Middle East and Africa. At the same time though, they found that the acceptance towards refugees, who were overall regarded as more acceptable, remained high. The media analysis by Maurer et al. (2019) was extended for the period 2016 to 2020 in a later publication (Maurer et al., 2021). Besides a decline in issue salience, the authors describe a predominantly negative depiction of the topic of migration. This time, the rather negative tone applied to both the phenomenon of asylum immigration as well as to the immigrants themselves. The immigration consequences were portrayed mainly as a threat to Germany, with links to security implications (i.e., crime and terrorism) being much more prevalent than economic or social implications (Maurer et al., 2021). After Russia started its full-scale invasion of Ukraine in February 2022, causing the displacement of millions of Ukrainians, many of whom migrated to Germany, the “Welcome Culture” in Germany was revived. However, the simultaneity of the influx of approximately one million Ukrainians and surging asylum seeker numbers led to strained resources resulting in, for example, appeals for help from German municipalities (Wieland, 2024). In this context, asylum immigration from outside of Europe became particularly controversial. Studies show that attitudes were more positive towards Ukrainians than non-European refugees, such as Afghans and Somalis (Moise et al., 2024) or Syrians (Dražanová & Geddes, 2023). These studies also show a deterioration of refugee attitudes in the months following Russia’s invasion. Between May/June 2022 and April 2023, the share of German respondents that preferred to admit no or only few Syrian refugees surged by roughly 20 percentage points (Dražanová & Geddes, 2023). This is also in line with the latest data collection of the Bertelsmann Stiftung in October 2024, in which 60 percent of respondents agreed with the statement that Germany could not take in more refugees because it had reached a capacity limit. This proportion was higher than in any of the three preceding waves of data collection (Wieland, 2024). 12 In the first round of data collection, participants watched either the treatment or the control video, after which they answered questions about their humanitarian concerns – i.e., concerns for the well-being of Syrian refugees in Turkey – and their perceptions of threat regarding a (hypothetical) migration of these refugees to Germany. Specifically, we asked respondents to what extent they were concerned about the refugees’ safety, material well-being, health, and future prospects (humanitarian concerns) on a scale from 1 (not at all concerned) to 5 (very concerned). The threat perception items assessed fears related to labor market competition, welfare state implications, security risks, and cultural compatibility between refugees and natives (threat perceptions), using a 5-point scale as well. Additionally, respondents expressed their policy preferences by indicating their support (or lack thereof) for a policy petition. One petition advocated for increased financial support for refugee camps hosting Syrians in Turkey (camps petition), while the other called for the admission of Syrians from Turkey to Germany (admission petition). The inclusion of such high-stakes questions helps to elicit the true opinions of survey respondents, particularly in contexts where certain responses may be seen as more (or less) socially desirable (Stantcheva, 2023). In the second round, we aimed at understanding the associations respondents made while watching the video using an open-ended question. We also asked them to fill out a matrix with 32 emotions, covering anger, disgust, fear, anxiety, sadness, desire, relaxation, and happiness items. Specifically, we requested them to indicate the extent to which they experienced these emotions while watching the video on a 7-point Likert scale. The order of the associations question and the emotional scale were randomized to address potential order effects. English translations of the questionnaires are printed in Appendix A and Appendix B, respectively. The original German versions can be made available upon request. This data is used to empirically test the effects of our video treatment on respondents’ attitudes towards refugees. Specifically, our empirical approach follows this OLS regression equation: 𝑦= 𝛽+ 𝛽𝑉𝑖𝑑𝑒𝑜+ 𝛽𝑉𝑖𝑑𝑒𝑜∗ 𝐼𝑚𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛+ 𝛽𝐼𝑚𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛+ 𝛽𝑋+ 𝜆+ 𝑢 In this equation, 𝑦 represents the refugee attitudes of respondent i pertaining to humanitarian concerns, threat perceptions, and policy preferences. The variable 𝑉𝑖𝑑𝑒𝑜 stands for the respondents’ treatment status, i.e. exposure to the treatment or the control video, whereas 𝐼𝑚𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛 captures the extent of respondents’ real-life immigration exposure depending on the county of residence c. This is either measured as a stock variable, i.e. the share of foreign 13 citizens in 2022, or as a flow variable, i.e. the percentage increase in the foreign population between 2014 and 2022. 𝑋 is a vector with individual-level control variables, comprising measures of the respondents’ gender, age, income, education, political orientation and social trust. Finally, 𝜆 is a dummy variable for the respondents’ state of residence s. 4) Results We proceed with the presentation of our results as follows: First, we show the effects of our treatment on humanitarian concerns, threat perceptions and policy preferences of respondents and discuss the heterogeneity between East and West Germany. Regression tables in the main text include a set of control variables, whereas the regression results without controls can be found in Appendix G. Second, we shed light on the channels underlying our treatment effects by analyzing our data on the associations and emotions respondents reported when watching the video. Finally, we investigate whether and how the treatment effects are moderated by respondents’ real-life immigration exposure (tables in Appendix H). 4.1 Analysis of the main treatment effects The levels of the four types of humanitarian concerns we observed were, on average, all close to the middle category of three, with concerns for safety and material wellbeing slightly below and concerns for health and future prospects just above three. The average values of threat perceptions show slightly more variation. In particular, perceived job threat is less pronounced than welfare, security and cultural threat perceptions, which aligns with earlier research (Dražanová et al., 2024; Hainmueller & Hopkins, 2014). Support for the camps and the admission petitions amount to 25 percent and 23 percent, respectively. This is small compared to earlier research (Azevedo et al., 2021), which may have to do with the specific context in which the data was collected.11 Comparing treatment and control group, we observe that the average levels of humanitarian concern are at least slightly greater in the control relative to the treatment group across all four concern types. Similarly, control group respondents show slightly lower averages for our threat perception measures and a greater willingness to sign the 11 The combination of large numbers of Ukrainian refugees and increased asylum immigration to Germany resulted in a widespread feeling of being overwhelmed and in preferences for reduced refugee immigration (Wieland, 2024). 14 camps petition.12 This indicates that refugee attitudes were, on average, more positive in the control vis-à-vis the treatment group. This impression is confirmed in our regression analysis. Table 1 summarizes the effects of exposure to the treatment video on respondents’ humanitarian concerns for the refugees. We consistently find a negative effect of similar size on all four dimensions of humanitarian concerns.13 This suggests that exposure to a video, designed to speak about refugee migration neutrally, leads to a significant erosion in compassion towards the respective refugee group. Table 1: Treatment effects on humanitarian concerns Safety Material Health Future Treatment -0.369 *** -0.354 *** -0.341 *** -0.298 *** [0.099] [0.099] [0.105] [0.106] Control mean 2.974 3.170 3.264 3.253 Observations 620 624 624 620 R-squared 0.172 0.194 0.175 0.185 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. The treatment effects on respondents’ threat perceptions are summarized in Table 2. Our data suggests that perceptions of job, welfare and cultural threat are unaffected by our treatment. However, we identify a positive effect on respondents’ perceptions of refugees as a security risk. Security concerns have been found to be particularly important in shaping refugee attitudes (Lahav & Courtemanche, 2012; Landmann et al., 2019). While past research studying messaging effects on immigration-related fears included specific threat-related elements in the treatments, we demonstrate that messaging can amplify security threat perceptions even though the messaging content makes no connections to security risks. 12 The propensity to sign the admission petition is virtually identical across both groups. 13 The effect size of our treatment amounts to roughly one fourth of the standard deviation of the outcome variables. 15 Table 2: Treatment effects on threat perceptions Jobs Welfare Security Culture Treatment 0.076 0.007 0.195** 0.117 [0.085] [0.085] [0.085] [0.081] Control mean 2.481 3.393 3.259 3.452 Observations 642 642 642 642 R-squared 0.173 0.255 0.297 0.212 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Finally, we are interested in whether the treatment also translates into differences in people’s policy preferences, specifically in people’s willingness to sign a petition either to support Syrians in Turkish refugee camps or for admitting Syrians from Turkey. Table 3 reveals that we did not find significant effects on these behavioral outcomes, at least not at the conventional 5 percent significance level. Our point estimate of a 7.6 percentage point reduction in respondents’ willingness to sign the petition for camps support is, however, meaningful in size and significant at the 10 percent level. This is in line with other findings suggesting that changing perceptions is more easily achieved than policy preferences (Jørgensen & Osmundsen, 2022). Table 3: Treatment effects on policy preferences Camps Admission Treatment -0.076* -0.008 [0.046] [0.047] Control mean 0.292 0.232 Observations 319 323 R-squared 0.262 0.202 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. 4.2 Heterogeneity of treatment effects between East and West Overall, the treatment effects appear quite similar across a wide range of socioeconomic characteristics, such as respondents’ gender, education, political and humanitarian orientation or social trust levels. However, we find interesting geographical variations. For example, we identify significant differences in how East and West German residents respond to the 16 treatment (Table 4). While East Germans show, on average, slightly lower levels of humanitarian concerns than West Germans in the control group14, there is no decline in East Germans’ humanitarian concerns following our treatment. In other words, the negative overall effect of our treatment is driven by the reactions of West rather than East German respondents. Table 4: Treatment effects on humanitarian concerns – by East vs. West Safety Material Health Future Treatment -0.445 *** -0.415 *** -0.397 *** -0.399 *** [0.107] [0.108] [0.114] [0.115] Treatment*East 0.685 ** 0.544 * 0.624 ** 0.670 ** [0.288] [0.292] [0.306] [0.314] East -0.353 * -0.341 -0.368 -0.323 [0.212] [0.217] [0.226] [0.231] Control mean 2.974 3.170 3.264 3.253 Observations 592 596 596 592 R-squared 0.159 0.179 0.170 0.172 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Similarly, with respect to respondents’ security threat perception, we again find evidence that it is the West Germans who are responsible for the treatment effect. Even though the interaction term is not significant here15, when computing the treatment effects separately, the coefficient for the East German subsample is virtually zero. This contrasts with the much larger and significantly positive effect among West Germans (Table 5). Table 5: Treatment effects on threat perceptions – by East vs. West Jobs Welfare Security Culture Treatment 0.079 -0.024 0.220 ** 0.131 [0.092] [0.092] [0.092] [0.088] 14 These differences are not statistically significant though. 15 The lack of significance may also be due to the relatively small number of East Germans in the sample. 17 Treatment*East -0.090 -0.006 -0.233 -0.194 [0.244] [0.245] [0.244] [0.235] East 0.022 0.112 0.101 0.110 [0.176] [0.177] [0.177] [0.170] Control mean 2.481 3.393 3.259 3.452 Observations 613 613 613 613 R-squared 0.170 0.246 0.294 0.197 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Turning to the petition questions, we find, once more, that the video treatment reduces support for on-site assistance among West Germans, whereas the coefficient for East Germans is positive, albeit insignificant. We do not detect significant effects or differences with respect to admission preferences. Table 6: Treatment effects on policy preferences – by East vs. West Camps Admission Treatment -0.108 ** -0.029 [0.050] [0.050] Treatment*East 0.298 ** 0.113 [0.139] [0.138] East -0.200 * -0.122 [0.113] [0.087] Control mean 0.292 0.232 Observations 306 307 R-squared 0.215 0.160 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. 18 4.3 In-depth analysis of associations and emotions We have provided a comprehensive picture of the effects of “neutral” refugee migration messaging on various forms of refugee attitudes and analyzed the extent to which the effects differ between residents of East and West Germany. To understand better why and how our treatment affects people’s views, we followed up with a second round of data collection in which we asked respondents explicitly what the video made them think of (associations question) and which emotions they experienced when watching it (emotions question). While all 327 responses on the emotional experiences were analyzed, we removed responses to the associations question from 29 participants due to insufficient quality. The emotion with the highest average was “sad”, followed by “worry” and “grief” ranked third. While “sad” and “grief” likely express some form of empathy with the refugees, “worry” could be linked either with concerns for the refugees or concerns for oneself or for Germany as an important host country of Syrian refugees (even though this was not explicitly mentioned). Indeed, disaggregating this emotion by subgroups, we find that average values are highest among those on the very left (4.21) and those on the very right (4.19) of the political spectrum, whereas the lowest average is found for people identifying as centrist (3.52). We interpret this as suggestive evidence that respondents’ worry was directed at different groups of people. The emotions can be linked to a large number of empathetic statements people made answering the associations question. Among these responses, we found 242 quotations distributed across 186 responses that included some kind of a “pro-refugee” message. By far the most common statements were linked to a recognition of the adversity the refugees had to cope with and/or an expression of sympathy, with 64 percent of the pro-refugee items entailing such a message. Other relatively common types of pro-refugee statements were characterized by a change of perspectives (19 percent), i.e. when respondents put themselves in the shoes of the refugees, or by a positive attitude towards support for refugees (14 percent). On the other hand, quotations entailing some kind of anti-refugee message were also quite frequent (161 quotations), yet these were spread across only 88 respondents. In contrast to the pro-refugee category, there is no clearly dominating theme among the anti-refugee statements. That said, more than 17 percent of these quotations showed a link to the theme of security and crime, making this the most common negative association. These quotations range from responses with a rather understanding tone, maintaining, for example, that the refugees’ 19 challenging living conditions made them more prone to delinquency, to labeling refugees as “murderers, thieves, rapists”16 (no. 48). Another relatively common group of statements doubted the legitimacy of the motives of the migrants (14 percent), e.g. arguing that they were moving rather for economic than humanitarian reasons or demanding harsher admission and return policies (12 percent). With regards to further threat perceptions, statements regarding economic threat were entailed in 10 percent of anti-refugee statements, whereas culture-related fears were hardly mentioned (1 percent). These anti-refugee remarks can be linked with another set of emotions with relatively high averages, such as “rage”, “mad”, and “anger”.17 Respondents’ associations reveal that these emotions are in part directed at refugees, in particular at “those who exploit the help” (no.39). In other instances though, these anger items were apparently linked with a disapproval of the Syrian Civil War or wars in general.18 Taken together, our data suggests that reporting pro-refugee – relative to anti-refugee – associations and emotions is more widespread. At the same time, data from the survey experiment indicated somewhat higher averages for threat perceptions vis-à-vis humanitarian concerns. One reason may be that social desirability effects are particularly strong when asking respondents specifically for their associations and emotions when watching the video. Moreover, it is common for both sentiments to co-exist within the same individual, even though one perception may be more pronounced than the other (Jeannet et al., 2021). In line with that, 30 responses included expressions of ambivalence. For example, one respondent speaks of “mixed feelings, concerns for the security of my country, for of my future and those of my children, as well as empathy with the Syrian population” (no. 70). This illustrates the attempt to balance the two dominant ‘logics’, a humanitarian assistance and a national interest ‘logic’, that typically govern the formation of attitudes towards humanitarian migration (Jeannet et al., 2021). The fact that the security issue ranks first among the anti-refugee statements and that other forms of threat perception were much less common, backs our finding that exposure to the treatment video amplifies the perception of security related fears. 16 Since the data collection was conducted in German, all quotations have been translated into English by the authors. 17 They ranked fifth, sixth, and seventh, respectively. 18 Overall, the condemnation of wars and warmongers was mentioned in 74 responses, making it one of the most frequent themes in participants’ responses. 20 Besides our overall results, the different reactions among East and West Germans stood out. An analysis particularly of the anti-refugee associations between these groups provides some insights:19 First, almost all of the references to security implications were made by West Germans, making it the clear frontrunner with a share of almost 20 percent of anti-refugee quotations in this sub-group. In line with this finding, all of the fear and anxiety emotions that we measured were, on average, higher among West than East Germans. The difference is particularly pronounced for “Dread” (Angst), where it is also statistically significant at the 10 percent level despite the modest sample size. Conversely, among East Germans, indifference towards the refugees is the most common antirefugee item. For example, one respondent writes “Far away - only concerns me to a limited extent, actually not my problem, I can’t and don’t want to help with that” (no. 26). This is corroborated by differences among relaxation-related emotions that East and West Germans reported. The average for East Germans is higher among all four of the respective emotions (easy-going, chilled out, calm, and relaxation). For the “chilled out”-item, the difference is especially large and significant at the 5 percent level. In summary, the data suggests that East Germans are more likely to be unmoved when watching the video, which could explain the lack of a treatment effect in this part of the sample. In contrast, video exposure is more likely to drive up security fears among respondents from West Germany, which is exactly what we found in our regression analysis. Given that the West German subsample is also responsible for the observed erosion of humanitarian concerns following our treatment, it seems likely that the increased concerns about one’s own security or the security of one’s co-citizens comes at the cost of concerns for more distant others. This pattern has been discussed in the Intergroup Threat Theory, according to which increased feelings of threat may cause the erosion of empathy towards outgroup members, whereas empathy towards ingroup members may be amplified (Stephan et al., 2009). Finally, our follow-up data also sheds some light on whether specific aspects of the video’s content are responsible for the observed impact on respondents, or whether the treatment effects are more likely to be based on salience manipulation. Several remarks made by respondents suggest that the video was not only intended to be neutral but that this is also how it was perceived. For example, one respondent wrote “I took it as an informational video that, for me, 19 Data from respondents from Berlin are excluded for this analysis as they could not be clearly assigned to either camp. 21 contained neither new or surprising facts, nor did it evoke any noteworthy emotional response” (no. 235). Another saw it as “a sober, calm video without sensationalism or polemics”. The term “sober” (sachlich) was also highlighted by three other respondents as well. Multiple respondents mentioned explicitly that there was nothing new they learnt from the video, as illustrated in this quote: “It didn’t convey anything new, just what’s already known from the news.” Among those who reported having learned something through the video, the majority referred to the importance of Turkey as a host country for Syrians, which has no evident link to our attitude variables. In addition, the average scores for the emotions questions were all below the middle category, suggesting that the video did not evoke strong emotional reactions of any kind. Overall, our data makes it seem unlikely that it was the content of the video that impacted the attitudes of respondents. 4.4 Interactions between treatment and real-life immigration exposure We have shown that reactions to the treatment differ between the eastern and western regions of the country. In this subsection, we take a closer look at the role of people’s place of residence by adopting a more fine-grained approach, focusing on how real-life exposure to immigration moderates these reactions. To this end, we have merged our survey experiment data with administrative data on the foreign population by county (Kreise) for the years 2014 and 2022. Theory suggests that living in a more or less diverse environment affects people’s immigration attitudes, yet, it is ambivalent with regards to the direction of this effect (see Section 2). As pointed out by Dražanová & Gonnot (2023), exposure to immigration can be operationalized in two different ways, either capturing immigrant stocks, measured by the share of the foreign population at a specific point in time, or immigration flows, i.e. the change in the foreign population share in a specific time period. In our data, there is a strong negative correlation20 between those two because places with high levels of diversity in 2022 were usually quite diverse in the baseline year of 2014 already. Therefore, the percentage increase in these counties has been lower than in initially more homogenous counties, where even modest inflows of foreign citizens resulted in relatively high percentage increases in the foreign population. The data shows that diversity has gone up in all German counties, without exception, with a median increase of 63 percent. The foreign population share in 2022 is correlated with slightly 20 The correlation coefficient is -0.67. 28 Dražanová, L., & Gonnot, J. (2023). Attitudes toward immigration in Europe: Cross-regional differences. Open Research Europe, 3, 66. https://doi.org/10.12688/openreseurope.15691.1 Dražanová, L., Gonnot, J., Heidland, T., & Krüger, F. (2024). Which individual-level factors explain public attitudes toward immigration? A meta-analysis. Journal of Ethnic and Migration Studies, 50(2), 317–340. https://doi.org/10.1080/1369183X.2023.2265576 Facchini, G., Mayda, A. M., & Puglisi, R. (2017). Illegal immigration and media exposure: Evidence on individual attitudes. IZA Journal of Development and Migration, 7(1), 14. https://doi.org/10.1186/s40176-017-0095-1 Ferwerda, J., Flynn, D. J., & Horiuchi, Y. (2017). Explaining opposition to refugee resettlement: The role of NIMBYism and perceived threats. Science Advances, 3(9), e1700812. https://doi.org/10.1126/sciadv.1700812 Fraser, N. A. R., & Murakami, G. (2022). The Role of Humanitarianism in Shaping Public Attitudes Toward Refugees. Political Psychology, 43(2), 255–275. https://doi.org/10.1111/pops.12751 Getmansky, A., Sınmazdemir, T., & Zeitzoff, T. (2018). Refugees, xenophobia, and domestic conflict: Evidence from a survey experiment in Turkey. Journal of Peace Research, 55(4), 491–507. https://doi.org/10.1177/0022343317748719 Green, D. P., Strolovitch, D. Z., & Wong, J. S. (1998). Defended Neighborhoods, Integration, and Racially Motivated Crime. American Journal of Sociology, 104(2), 372–403. https://doi.org/10.1086/210042 Greussing, E., & Boomgaarden, H. G. (2017). Shifting the refugee narrative? An automated frame analysis of Europe’s 2015 refugee crisis. Journal of Ethnic and Migration Studies, 43(11), 1749–1774. https://doi.org/10.1080/1369183X.2017.1282813 Hainmueller, J., & Hopkins, D. J. (2014). Public Attitudes Toward Immigration. Annual Review of Political Science, 17(1), 225–249. https://doi.org/10.1146/annurev-polisci-102512194818 29 Hangartner, D., Dinas, E., Marbach, M., Matakos, K., & Xefteris, D. (2019). Does Exposure to the Refugee Crisis Make Natives More Hostile? American Political Science Review, 113(2), 442–455. https://doi.org/10.1017/S0003055418000813 Heidenreich, T., Lind, F., Eberl, J.-M., & Boomgaarden, H. G. (2019). Media Framing Dynamics of the ‘European Refugee Crisis’: A Comparative Topic Modelling Approach. Journal of Refugee Studies, 32(Special_Issue_1), i172–i182. https://doi.org/10.1093/jrs/fez025 Hillenbrand, T., Martorano, B., Metzger, L., & Siegel, M. (2024). The power of narratives: How framing refugee migration impacts attitudes and political support towards refugees. MERIT Working Papers 2024-007. Jeannet, A.-M., Heidland, T., & Ruhs, M. (2021). What asylum and refugee policies do Europeans want? Evidence from a cross-national conjoint experiment. European Union Politics, 22(3), 353–376. https://doi.org/10.1177/14651165211006838 Jørgensen, F. J., & Osmundsen, M. (2022). Correcting Citizens’ Misperceptions about nonWestern Immigrants: Corrective Information, Interpretations, and Policy Opinions. Journal of Experimental Political Science, 9(1), 64–73. https://doi.org/10.1017/XPS.2020.35 Lahav, G., & Courtemanche, M. (2012). The Ideological Effects of Framing Threat on Immigration and Civil Liberties. Political Behavior, 34(3), 477–505. https://doi.org/10.1007/s11109-011-9171-z Landmann, H., Gaschler, R., & Rohmann, A. (2019). What is threatening about refugees? Identifying different types of threat and their association with emotional responses and attitudes towards refugee migration. European Journal of Social Psychology, 49(7), 1401–1420. https://doi.org/10.1002/ejsp.2593 Liu, J. (2023). Framing Syrian refugees: Examining news framing effects on attitudes toward refugee admissions and anti-immigrant sentiment. Newspaper Research Journal, 44(1), 6–25. https://doi.org/10.1177/07395329221077251 30 Manzoni, E., Murard, E., Quercia, S., & Tonini, S. (2024). News, Emotions, and Policy Views on Immigration. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.4844524 Maurer, M., Jost, P., Haßler, J., & Kruschinski, S. (2019). Auf den Spuren der Lügenpresse. Publizistik, 64(1), 15–35. https://doi.org/10.1007/s11616-018-00466-y Maurer, M., Jost, P., Kruschinski, S., & Haßler, J. (2021). Fünf Jahre Medienberichterstattung über Flucht und Migration. Johannes Gutenberg-Universität Mainz Institut für Publizistik. McCann, K., Sienkiewicz, M., & Zard, M. (2023). The role of media narratives in shaping public opinion toward refugees: A comparative analysis (N° 72; Migration Research Series). International Organization For Migration (IOM). Moise, A. D., Dennison, J., & Kriesi, H. (2024). European attitudes to refugees after the Russian invasion of Ukraine. West European Politics, 47(2), 356–381. https://doi.org/10.1080/01402382.2023.2229688 Newman, B. J., Hartman, T. K., Lown, P. L., & Feldman, S. (2015). Easing the Heavy Hand: Humanitarian Concern, Empathy, and Opinion on Immigration. British Journal of Political Science, 45(3), 583–607. Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90(5), 751–783. https://doi.org/10.1037/0022-3514.90.5.751 Putnam, R. D. (2007). E Pluribus Unum: Diversity and Community in the Twenty-first Century The 2006 Johan Skytte Prize Lecture. Scandinavian Political Studies, 30(2), 137–174. https://doi.org/10.1111/j.1467-9477.2007.00176.x Schemer, C. (2012). The Influence of News Media on Stereotypic Attitudes Toward Immigrants in a Political Campaign. Journal of Communication, 62(5), 739–757. https://doi.org/10.1111/j.1460-2466.2012.01672.x Schneider-Strawczynski, S., & Valette, J. (2023). Media Coverage of Immigration and the Polarization of Attitudes. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.4673964 31 Stantcheva, S. (2023). How to Run Surveys: A Guide to Creating Your Own Identifying Variation and Revealing the Invisible. Annual Review of Economics, 15(1), 205–234. https://doi.org/10.1146/annurev-economics-091622-010157 Stecker, C., & Debus, M. (2019). Refugees Welcome? Zum Einfluss der Flüchtlingsunterbringung auf den Wahlerfolg der AfD bei der Bundestagswahl 2017 in Bayern. Politische Vierteljahresschrift, 60. https://doi.org/10.1007/s11615-019-001513 Steinmayr, A. (2016). Exposure to refugees and voting for the far-right:(Unexpected) results from Austria (No. 9790). IZA Discussion Papers. Stephan, W., Ybarra, O., & Rios, K. (2009). Intergroup threat theory. In T. D. Nelson (Ed.), Handbook of prejudice, stereotyping, and discrimination (pp. 43–59). Lawrence Erlbaum Associates Publishers. Unal, U., Hayo, B., & Erol, I. (2024). The Effect of Immigration on Housing Prices: Evidence from 382 German Districts. The Journal of Real Estate Finance and Economics. https://doi.org/10.1007/s11146-024-09988-x Vertier, P., Viskanic, M., & Gamalerio, M. (2023). Dismantling the “Jungle”: Migrant relocation and extreme voting in France. Political Science Research and Methods, 11(1), 129– 143. https://doi.org/10.1017/psrm.2022.26 Wieland, U. (2024). Willkommenskultur in Krisenzeiten. Wahrnehmungen und Einstellungen der Bevölkerung zu Migration und Integration in Deutschland. Bertelsmann Stiftung. https://www.bertelsmannstiftung.de/de/publikationen/publikation/did/willkommenskultur-in-krisenzeiten Wong, C. J. (2007). “Little” and “Big” Pictures in Our Heads: Race, Local Context, and Innumeracy About Racial Groups in the United States. Public Opinion Quarterly, 71(3), 392–412. https://doi.org/10.1093/poq/nfm023 ZDF. (n.d.). Https://wahltool.zdf.de/wahlergebnisse/2025-02-23-BT-DE.html. Retrieved April 7, 2025, from https://wahltool.zdf.de/wahlergebnisse/2025-02-23-BT-DE.html 32 Appendix Appendix A: Questionnaire of survey experiment In the following the English translation of the online questionnaire is printed. The survey in the original German can be made available upon request. We report the answer options in italic below. They are separated by semicolons. Q1.1 Welcome! You will be asked to take part in a survey. This survey is conducted by researchers from Maastricht University. All the information you get in this survey is verified. There are no right or wrong answers. We are interested in knowing your personal views about yourself and the world. The survey covers society-related questions that also concern the role of the state and politics. Participation is voluntary. You can refuse to participate in this survey. If you start the survey, you can leave the study at any time, in which case you will not receive any financial compensation. Apart from the time you spend completing the survey (10-12 minutes), there is no cost to you. You still have the option of withdrawing your consent after completing the survey by contacting Prof. Dr. Melissa Siegel ([email protected]). Your study-related information will be treated confidentially. This study has been approved by the Ethics Review Committee of Inner City Faculties (ERCIC) at Maastricht University. The collection of data is confidential. Data analysis and reporting are anonymous. Your data will be kept separate from your Bilendi identification number. Upon completion of the study, you will receive financial compensation for your time. This is done in accordance with Bilendi's guidelines. If you have any questions or comments about this survey, please contact Prof. Dr. Melissa Siegel. At the end of the survey, you will have the opportunity to give us feedback on your experience. ____________________ Declaration of consent I agree that my data will be used for scientific purposes. I had sufficient time to decide if I wanted to participate in the study. I know that participation in the study is voluntary, and I know that I can choose to cancel the survey and withdraw my consent at any time. I do not have to give reasons for such a decision. In this case, I will not receive any financial compensation. I had the opportunity to connect with Prof. Dr. Melissa Siegel, a researcher involved in this study, and ask questions. I am aware that the data is stored anonymously and therefore only published anonymously. I agree to participate Yes; No Q2.1 The Survey During the survey, you will be asked to watch a short video. For this, you will need working speakers or headphones. The information contained in the video is genuine and comes from 33 one or more publicly available and verified sources. The survey will take about 10-12 minutes to complete. Your financial compensation After completing the survey, you will receive financial compensation for your time. The survey is considered complete once you reach the last page thanking you for participating. You will not receive any compensation if you cancel the survey early, but you can do so at any time. Q3.1 How would you identify yourself? Man; Woman; Non-binary / third gender; Prefer to self-describe; Prefer not to say Q3.2 What is your age in years? Younger than 18; 18-27; 28-37; 38-47; 48-57; 58-69; Older than 69, Prefer not to say Q4.1 In which state do you live? List of all German states (Bundesländer); Other; Prefer not to say Q4.2 In which German county do you live? List of all German counties (Landkreise); Other; Prefer not to say Q5.1 What is your highest educational degree? If you are currently still in education, select the highest degree you have already earned. No formal education; Completion of primary school; Completion of Hauptschule or Realschule; Fachabitur or Abitur; Bachelor’s degree; Master’s degree or Diplom; Doctoral degree; Other; Prefer not to say Q5.2 How would you describe your current employment status? Employed (full-time or part-time; Self-employed/freelance; unemployed, jobseeker; unemployed, no jobseeker; Retired; In full-time education; Other, Prefer not to say Q5.3 What is your approximate annual household income after deduction of taxes and social security contributions? Please select the appropriate category. 11,999 or less; 12,000-19,999; 20,000-26,999; 27,000-33,999; 34,000-40,9999; 41,00049,999; 50,000-59,999; 60,000-74,999; 75,000-99,999; 100,000 or more; Prefer not to say Q5.4 What is their religious affiliation? Christianity, Roman Catholic; Christianity, Protestant; Other Christian Church; Islam, Judaism; Hinduism; Buddhism; No religious affiliation; Other; Prefer not to say Q6.1 Many people use the terms "left" and "right" to denote different political views. When you think about your own political views, how would you rank those views on that scale? Left, Center-left; Center; Center-right, Right, Prefer not to say Q6.2 Generally speaking, would you say that most people can be trusted, or that you can't be too careful in dealing with people? 1 You can’t be too careful; 2; 3; 4; 5 Most people can be trusted; Prefer not to say Q7.1 Please indicate to which extent you agree with the following statements Strongly disagree; Somewhat disagree; Neither agree, nor disagree; Somewhat agree, Strongly agree 34 - One should always find ways to help others less fortunate than oneself - A person should always be concerned about the well-being of others - It is best not to get too involved in taking care of other people's needs - People tend to pay more attention to the well-being of others than they should Q8.1 Please watch the following approximately one-minute video carefully and in its entirety, and also make sure that the sound is working well. Q14.1 A large number of Syrians are living in refugee camps in Turkey. When you think about their situation, how concerned are you about... Not at all concerned; A bit concerned; Somewhat concerned; Quite concerned; Very concerned - their safety? - their provision with basic material goods? - their health situation? - their prospects for the future? Q14.2 Please indicate how you agree with the following statements. If Syrians from Turkish refugee camps were to come to Germany to live here… Strongly disagree; Somewhat disagree; Neither agree, nor disagree; Somewhat agree; Strongly agree - they would take jobs away from the German population - they would, in the long run, benefit more from the welfare state than they contribute - the security situation in Germany would deteriorate - their values and beliefs would be at odds with those of the Germans Q15.1 Do you think the German government should increase financial support for Turkish refugee camps? Definitely not; Rather not; Maybe; Rather yes; Definitely Q16.1 Would you be in favor of Germany taking in Syrians from Turkish refugee camps? Definitely not; Rather not; Maybe; Rather yes; Definitely Q16.2 Would you be in favour of your hometown providing housing for Syrians from Turkish refugee camps? Definitely not; Rather not; Maybe; Rather yes; Definitely Q16.3 Would you be willing to privately accommodate Syrians from Turkish refugee camps for a few days? Definitely not; Rather not; Maybe; Rather yes; Definitely Q17.1 You can become politically active by signing a petition. We will send the petition to the Commissioner of the Federal Government for Migration, Refugees, and Integration. Your 35 name will not be mentioned. Instead, we report how many participants in our study supported the respective petitions. Would you like to sign the petition below? I would like to sign a petition calling for more financial support from the German government for Turkey's refugee camps.; I do not want to sign this petition. Q18.1 You can become politically active by signing a petition. We will send the petition to the Federal Government Commissioner for Migration, Refugees and Integration. Your name will not be mentioned. Instead, we report how many participants in our study supported the respective petitions. Would you like to sign the petition below? I would like to sign a petition calling for the admission of Syrians from Turkish refugee camps to Germany.; I do not want to sign this petition. Q19.1 As a participant in this survey, you will be provided with an additional euro. You can choose how much of this amount you want to keep for yourself or how much you want to donate to one of the following organizations. 1) You can donate up to 50 cents to a certified international non-governmental organization (NGO) A that works to improve living conditions in Turkish refugee camps. 2) You can donate up to 50 cents to another certified international non-governmental organization (NGO) B that is committed to hosting Syrians from Turkish refugee camps in Germany. Please indicate how many cents you would like to donate to the respective organization. Note that you can donate a maximum of 50 cents per organization. The amounts will be allocated to you or the named organizations once the data collection process has been completed. We will inform you of the names of the organizations once you have completed the survey. How many cents would you like to donate at a time? Organization A: No donation; 5 cents; 10 cents; 15 cents; 20 cents; 25 cents; 30 cents; 35 cents; 40 cents; 45 cents; 50 cents Organization B: No donation; 5 cents; 10 cents; 15 cents; 20 cents; 25 cents; 30 cents; 35 cents; 40 cents; 45 cents; 50 cents Q20.1 What is your relationship status? Single, never been married; In a relationship, not married; Married, remarried; Divorced, Single; Widowed, single; Other; Prefer not to say Q20.2 Were you born in Germany? Yes, No, Prefer not to say Q20.3 In which country were you born? Q20.4 Was your mother born in Germany? Yes, No, Prefer not to say Q20.5 In which country was your mother born? Q20.6 Was your father born in Germany? Yes, No, Prefer not to say Q20.7 In which country was your father born? 36 Q20.8 Have you ever lived in another country for at least six consecutive months? Yes, No, Prefer not to say Q20.9 In which country(s) have you lived? Q21.1 You are now reaching the end of the survey. As announced, we would like to inform you to which organizations the donations will be forwarded to: Organization A: Médecins Sans Frontières / Organization B: Amnesty International You now have the opportunity to give feedback on this survey. Please note that we do not accept hateful and hurtful messages and will contact Bilendi if necessary. Thank you very much. 37 Appendix B: Questionnaire of the follow-up data collection In the following the English translation of the questionnaire of the second data collection round is printed. Again, the survey in the original German can be made available upon request. Q1.1 Welcome! You will be asked to take part in a survey. This survey is conducted by researchers from Maastricht University. All the information you get in this survey is verified. There are no right or wrong answers. We are interested in knowing your personal views about yourself and the world. There are no right or wrong answers. We are interested in hearing your personal views on societyrelated issues. For this purpose, you will also be asked at one point to write a short text of a few sentences. Participation is voluntary. You can refuse to participate in this survey. If you start the survey, you can leave the study at any time, in which case you will not receive any financial compensation. Apart from the time you spend completing the survey (10-12 minutes), there is no cost to you. You still have the option of withdrawing your consent after completing the survey by contacting Prof. Dr. Melissa Siegel ([email protected]). Your study-related information will be treated confidentially. This study has been approved by the Ethics Review Committee of Inner City Faculties (ERCIC) at Maastricht University. The collection of data is confidential. Data analysis and reporting are anonymous. Your data will be kept separate from your Bilendi identification number. After completing the study, you will receive financial compensation for your time. This is done in accordance with Bilendi's guidelines. If you have any questions or comments about this survey, please contact Prof. Dr. Melissa Siegel. At the end of the survey, you will have the opportunity to give us feedback on your experience. ____________________ Declaration of consent I agree that my data will be used for scientific purposes. I had sufficient time to decide if I wanted to participate in the study. I know that participation in the study is voluntary, and I know that I can choose to cancel the survey and withdraw my consent at any time. I do not have to give reasons for such a decision. In this case, I will not receive any financial compensation. I had the opportunity to connect with Prof. Dr. Melissa Siegel, a researcher involved in this study, and ask questions. I am aware that the data is stored anonymously and therefore only published anonymously. I agree to participate 44 Appendix F: Descriptive Results Table A2: Mean values of refugee attitudes by experimental condition Outcome variable Control group Treatment group Safety Concern 2.97 2.60 Material Concern 3.17 2.80 Health Concern 3.26 2.92 Future Concern 3.25 3.01 Jobs Threat 2.48 2.60 Welfare Threat 3.39 3.40 Security Threat 3.26 3.40 Cultural Threat 3.45 3.55 Camps Petition 0.29 0.21 Admission Petition 0.23 0.23 45 Appendix G: Main effects without controls Table A3: Treatment effects on humanitarian concerns – no controls Safety Material Health Future Treatment -0.371 *** -0.369 *** -0.340 *** -0.240 ** [0.093] [0.096] [0.100] [0.101] Control mean 2.974 3.170 3.264 3.253 Observations 769 774 777 771 R-squared 0.020 0.019 0.015 0.007 Standard errors in brackets. * p < .1, ** p < .05, *** p < .01 Table A4: Treatment effects on threat perceptions – no controls Jobs Welfare Security Culture Treatment 0.121 0.010 0.136 0.103 [0.080] [0.083] [0.086] [0.078] Control mean 2.481 3.393 3.259 3.452 Observations 807 807 807 807 R-squared 0.003 0.000 0.003 0.002 Standard errors in brackets. * p < .1, ** p < .05, *** p < .01 Table A5: Treatment effects on policy preferences – no controls Camps Admission Treatment -0.084 * -0.002 [0.043] [0.042] Control mean 0.292 0.232 Observations 404 403 R-squared 0.009 0.000 Standard errors in brackets. * p < .1, ** p < .05, *** p < .01 46 Appendix H: Treatment effects moderated by real-life immigration exposure Table A6: Treatment effects on humanitarian concerns – by immigrant stocks Safety Material Health Future Treatment -0.053 0.092 0.142 0.160 [0.251] [0.251] [0.265] [0.269] Treatment*Stocks -1.891 -2.573 * -2.793 * -2.687 * [1.371] [1.376] [1.450] [1.474] Immigrant Stocks 0.340 0.736 1.295 1.097 [1.136] [1.141] [1.203] [1.219] Control mean 2.974 3.170 3.264 3.253 Observations 611 614 614 610 R-squared 0.177 0.199 0.180 0.189 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Table A7: Treatment effects on threat perceptions – by immigrant stocks Jobs Welfare Security Culture Treatment -0.127 -0.448 ** -0.411 * -0.153 [0.217] [0.217] [0.215] [0.207] Treatment*Stocks 1.280 2.798 ** 3.651 *** 1.675 [1.190] [1.192] [1.182] [1.135] Immigrant Stocks -1.354 -1.357 -2.592 *** -1.032 [0.987] [0.989] [0.981] [0.942] Control mean 2.481 3.393 3.259 3.452 Observations 631 631 631 631 R-squared 0.177 0.266 0.314 0.222 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. 47 Table A8: Treatment effects on policy preferences – by immigrant stocks Camps Admission Treatment 0.195 * 0.130 [0.115] [0.121] Treatment*Stocks -1.657 *** -0.834 [0.633] [0.662] Immigrant Stocks 1.234 ** 0.786 [0.535] [0.535] Control mean 0.292 0.232 Observations 316 315 R-squared 0.289 0.208 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Table A9: Treatment effects on humanitarian concerns – by immigration flows Safety Material Health Future Treatment -0.703 *** -0.686 *** -0.660 *** -0.690 *** [0.180] [0.180] [0.191] [0.193] Treatment*Flows 0.441 ** 0.459 ** 0.440 ** 0.527 ** [0.198] [0.199] [0.210] [0.213] Immigration Flows -0.196 -0.197 -0.208 -0.335 * [0.187] [0.185] [0.195] [0.198] Control mean 2.974 3.170 3.264 3.253 Observations 611 614 614 610 R-squared 0.180 0.201 0.181 0.194 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. 48 Table A10: Treatment effects on threat perceptions – by immigration flows Jobs Welfare Security Culture Treatment 0.185 0.184 0.443 *** 0.261 * [0.155] [0.156] [0.155] [0.148] Treatment*Flows -0.127 -0.215 -0.315 * -0.175 [0.170] [0.170] [0.169] [0.162] Immigration Flows 0.069 0.172 0.241 0.064 [0.160] [0.161] [0.160] [0.153] Control mean 2.481 3.393 3.259 3.452 Observations 631 631 631 631 R-squared 0.175 0.261 0.307 0.220 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Table A11: Treatment effects on policy preferences – by immigration flows Camps Admission Treatment -0.300 *** -0.101 [0.081] [0.090] Treatment*Flows 0.288 *** 0.118 [0.089] [0.101] Immigration Flows -0.114 -0.075 [0.086] [0.088] Control mean 0.292 0.232 Observations 316 315 R-squared 0.296 0.206 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. 49 Table A12: Treatment effects on humanitarian concerns – by immigrant stocks (West only) Safety Material Health Future Treatment -0.567 * -0.240 -0.249 -0.192 [0.307] [0.312] [0.330] [0.327] Treatment*Stocks 0.719 -0.845 -0.764 -1.072 [1.611] [1.643] [1.733] [1.722] Immigrant Stocks -0.489 0.273 0.677 0.535 [1.237] [1.263] [1.334] [1.321] Control mean 2.974 3.170 3.264 3.253 Observations 501 504 503 502 R-squared 0.173 0.177 0.171 0.193 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Table A13: Treatment effects on threat perceptions – by immigrant stocks (West only) Jobs Welfare Security Culture Treatment -0.284 -0.522 * -0.541 ** -0.131 [0.267] [0.266] [0.263] [0.253] Treatment*Stocks 2.147 2.886 ** 4.293 *** 1.600 [1.408] [1.401] [1.382] [1.332] Immigrant Stocks -1.390 -1.541 -2.965 *** -1.252 [1.083] [1.078] [1.063] [1.025] Control mean 2.481 3.393 3.259 3.452 Observations 515 515 515 515 R-squared 0.184 0.258 0.326 0.221 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. 50 Table A14: Treatment effects on policy preferences – by immigrant stocks (West only) Camps Admission Treatment 0.029 0.069 [0.142] [0.145] Treatment*Stocks -0.857 -0.614 [0.746] [0.758] Immigrant Stocks 0.913 0.884 [0.584] [0.575] Control mean 0.292 0.232 Observations 261 254 R-squared 0.303 0.208 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Table A15: Treatment effects on humanitarian concerns – by immigration flows (West only) Safety Material Health Future Treatment -0.531 ** -0.663 *** -0.538 ** -0.646 *** [0.233] [0.236] [0.250] [0.248] Treatment*Flows 0.150 0.452 0.255 0.436 [0.342] [0.345] [0.364] [0.361] Immigration Flows -0.217 -0.252 -0.081 -0.150 [0.256] [0.253] [0.268] [0.265] Control mean 2.974 3.170 3.264 3.253 Observations 501 504 503 502 R-squared 0.173 0.179 0.172 0.195 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. 51 Table A16: Treatment effects on threat perceptions – by immigration flows (West only) Jobs Welfare Security Culture (1) (2) (3) (4) Treatment 0.313 0.234 0.706 *** 0.237 [0.201] [0.200] [0.198] [0.190] Treatment*Flows -0.360 -0.407 -0.806 *** -0.143 [0.295] [0.294] [0.291] [0.279] Immigration Flows 0.078 0.110 0.380 * -0.168 [0.217] [0.216] [0.214] [0.205] Control mean 2.481 3.393 3.259 3.452 Observations 515 515 515 515 R-squared 0.183 0.255 0.322 0.222 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. Table A17: Treatment effects on policy preferences – by immigration flows (West only) Camps Admission Treatment -0.179 * -0.138 [0.105] [0.114] Treatment*Flows 0.093 0.164 [0.153] [0.170] Immigration Flows 0.026 -0.088 [0.125] [0.109] Control mean 0.292 0.232 Observations 261 254 R-squared 0.299 0.203 Notes: Controls include gender, age, marital status, university education, income, being unemployed, being Muslim, migration background, overseas experience, humanitarian orientation, political orientation, social trust, state (Bundesland) of residence. Standard errors in brackets. Significance levels: * p < .1, ** p < .05, *** p < .01. The UNU-MERIT WORKING Paper Series 2025-01 Development strategies for the green hydrogen economy in emerging economies by Fabianna Bacil, Anthony Black, Marina Domingues, Jun Jin, Rasmus Lema, Glen Robbins and Sören Scholvin 2025-02 Do global value chains and local capabilities matter for economic complexity in EU regions? by R. Boschma, E. Hernández-Rodríguez, A. Morrison and C. Pietrobelli 2025-03 Innovation and productivity: the recent empirical literature and the state of the art by Jacques Mairesse, Pierre Mohnen and Ad Notten 2025-04 Digital Technologies, Hiring, Training, and Firm Outcomes: Evidence from AI and ICT Adoption in Indian firmss by Sneha Marydas, Nanditha Mathew, Giordano De Marzo and Carlo Pietrobelli 2025-05 New technologies and employment: the state of the art by Marco Vivarelli and Guillermo Arenas Díaz 2025-06 Formal-Informal Supply Chain Linkages and Firm Productivity in Sub-Saharan Africa: The Role of Human Capital by Gbenoukpo Robert Djidonou, Neil Foster-McGregor and Nanditha Mathew 2025-07 Internet Access and Youth’s Mental Health and Well-being: Evidence from Ethiopia by Trang Thi Pham and Pui-Hang Wong 2025-08 A Twin Transition or a policy flagship? Emergent constellations and dominant blocks in green and digital technologies by Linnea Nelli, Maria Enrica Virgillito and Marco Vivarelli 2025-09 Good for business, not so much for the environment? Entry into importing and the energy intensity of Indian plants by Michele Delera, Nanditha Mathew and Tania Treibich 2025-10 The role of business visits in fostering R&D investment by Marco Vivarelli, Mariacristina Piva and Massimiliano Tani 2025-11 Not as innocent as it seems? The effects of "neutral" messaging on refugee attitudes by Tobias Hillenbrand, Bruno Martorano and Melissa Siegel