The Ukrainian Refugee Crisis and the Politics of Public Opinion: Evidence from Hungary

The 2022 Russian invasion of Ukraine was a watershed moment in European politics. The invasion prompted a massive in ﬂ ux of refugees into Central Europe, a region in which immigration has proven highly contentious and politically salient in recent decades. We study public opinion toward refugees in Hungary, a highly exclusionary political environment in which anti-migrant and anti-refugee sentiments are commonly invoked by the ruling government. Combining historical public opinion data from the past decade with two rounds of original survey data from 2022, we demonstrate that the Ukrainian refugee crisis was accompanied by a large increase in tolerance for refugees, reversing what had previously been one of the most anti-refugee public opinion environments in Europe. To explain this reversal, we use a series of survey experiments coupled with detailed settlement-level demographic data to investigate how con ﬂ ict proximity and racial, religious, and national identity shape openness to refugees. We ﬁ nd that the distinguishing feature of the 2022 refugee crisis was that refugees were mostly white European Christians driven from their home country by con ﬂ ict. We discuss the implications of our argument for Hungary, for European politics in times of crisis, and for the politics of public opinion in competitive authoritarian regimes.

characterizing refugees as an existential threat to Hungarian security-and to European identity (Juhász, Hunyadi,  and Zgut 2015). 1 Seven years later, the Russian invasion of Ukraine produced a second mass influx of millions of people into Central and Western Europe.Although most Ukrainians sought protection from conflict in the form of temporary protected status, they are described in most popular commentary as refugees-just like those who fled conflict in 2015-2016. 2 The number of civilians fleeing war in Ukraine far exceeded the total from 2015: as of September 2022, 2.5 million Ukrainians had entered Hungary alone (Erőss 2022; UNHCR 2022), with millions more having fled to Hungary's neighbors.
Much popular and political commentary has described the Ukrainian refugee crisis as unprecedented in recent European history.At the same time, much critical commentary on the 2022 refugee crisis has focused on Europe's "refugee hypocrisy" (Traub 2022) and the plainly different standards to which Ukrainians have been held in comparison to non-European, non-Christian, non-white refugees from countries like Afghanistan less than a decade previously.In highly exclusionary political environments such as Orbán's Hungary, in which anti-migrant and antirefugee sentiments are commonly invoked by the ruling government, how are mass publics responding to the largescale influx of foreigners from a conflict on its own borders?And how should scholars of public opinion toward refugees interpret changes in refugee support when the composition of the refugee population changes as well?
These questions are relevant beyond the specific case of Ukrainian refugee crisis, and their answers can help to guide researchers studying refugee politics around the world.Identifying and measuring refugee hypocrisy in response to real-world events such as the Ukrainian refugee crisis presents a difficult research design problem.To understand whether the racial or religious features of a refugee population explain public opinion in receiving countries, we require a comparative approach that can compare across different kinds of refugees, but there is very little variation in the racial and religious features of Ukrainians who have fled to Hungary.Comparing different refugee populations across time can alleviate these concerns but introduces new challenges regarding the comparability of refugee crises with different causes, refugee populations, migration routes, and economic implications.Changing economic and political conditions in the hosting countries might pose additional challenges for intertemporal comparison.Surveys that pose questions about hypothetical migrant populations can uncover subtle distinctions across populations, and allow for designs that estimate causal effects, but may be too unrealistic to be relevant in the context of a real-world refugee crisis.Yet as scholars we have a responsibility to engage with such difficult questions in order to understand how refugee crises have affected contemporary politics in national contexts such as Hungary.Attuned to those inferential challenges and the importance of studying difficult contemporary issues such as refugee hypocrisy in Europe, we introduce new data collected right at the onset of the Ukrainian refugee crisis to identify how it has affected public opinion in Hungary.Specifically, we combine original survey data with detailed settlement-level demographic data to describe a dramatic change in Hungarian public opinion toward refugees following the 2022 Ukrainian refugee crisis and to explain the sources of this change.Comparing multiple rounds of public opinion data across the past decade with newly collected data from April and November 2022, we demonstrate that the 2022 Ukrainian crisis was accompanied by a large increase in tolerance for refugees, reversing what had previously been one of the most anti-refugee public opinion environments in Europe.
To explain this difference, we combine survey experiments with data on respondents' local environments to investigate how conflict proximity and racial and religious identity shape openness to refugees.We find that the distinguishing feature of the 2022 crisis was that those arriving in Central Europe were mostly white European Christians driven from their home country by conflict.Using careful research designs that we adapted to the specifics of the Hungarian context in 2022, we further demonstrate that race, religion, and values are important for explaining aggregate patterns in Hungarian public opinion toward refugees in 2022.Consistent with existing work on Hungarian politics, they are particularly important among supporters of the ruling Fidesz party, and for some groups of religious voters.We find no systematic evidence that these individual patterns are explained by regional factors within Hungary, although we do find that settlement-level religious identity explains the individuallevel correlation between religious identity and support.By introducing original survey data and attending carefully to the inferential challenges of comparing across sending countries, across regions of Hungary, and across time, we are able to show the central importance of race, religion, and values in explaining Hungarians' changing opinions toward refugees in 2022.
Our findings make two main contributions to the literature on public opinion toward refugees and migrants, especially in times of crisis (Brader, Valentino, and Suhay  2008; Bansak, Hainmueller, and Hangartner 2016; Dinas  et al. 2019; Hangartner et al. 2019; Vachudova 2020;  Goodman 2021; Kustov, Laaker, and Reller 2021).Ours is the most rigorous quantitative evidence available that the 2022 Ukrainian refugee crisis actually shifted public opinion toward refugees in a country where anti-migrant and anti-refugee sentiments were widely expressed, strongly held, and politically valuable to the incumbent government.These changes may be the consequence of a change in Hungarians' attitudes about refugees in general, or of changing expectations about who refugees are: we find that the common perception of Ukrainians as white, Christian, European refugees is responsible for the favorable shift in Hungarian public opinion toward refugees in 2022.Our methodological approach reveals why interpreting survey evidence about public opinion toward refugees requires care, for respondents will have very different expectations about who refugees are based on the time and place that the data is collected. 3eparately, our findings also contribute new evidence on public opinion formation in Hungary, helping us to better understand contemporary politics in a country that has been a focal point for discussions of illiberal politics in Europe and around the world, including the United States (Kelemen 2015; Enyedi 2018; Haggard and Kaufman  2021; McLaren 2022; Parton 2022; Scheppele 2022).Public opinion in competitive authoritarian regimes can reflect regime preferences, but it can also constrain regime behavior.We cannot use our data to evaluate whether popular support for the Fidesz government's general hostility toward non-white, non-Christian immigrants and refugees reflects the "true" preferences of the majority of Hungarians.But our findings are consistent with the argument that Hungarian public opinion is responsive to events beyond the control of the incumbent government, and our data show that this holds even among Fidesz supporters.We find it unlikely that the dramatic shift in Hungarian refugee opinion that we document later merely reflects changes in the Fidesz government's policy and rhetoric.What changed, instead, was the nature of the refugee crisis, leading both government policy and popular opinion to change accordingly.

Refugee Crises and European Politics
The 2015 European refugee crisis was a humanitarian emergency with social, economic, and political consequences for refugees fleeing conflict.It also fundamentally shaped politics in both sending and receiving countries.As our focus in this paper is on how European-and specifically Hungarian-public opinion has responded to recent refugee crises, we refer readers to existing work that explains the origins, details, and personal tragedies of the 2015 crisis (Prickett 2015; Kingsley 2016; McDonald-Gibson 2016; Barlai, Fähnrich, and Griessler 2017).The 2015 crisis is nevertheless a political milestone in Europe, one of a series of crises that has tested European governments and Europe's supranational institutions following the Global Financial Crisis of 2008-2009 (Jones, Kelemen, and Meunier 2021).
There is abundant evidence that the 2015 refugee crisis in Europe shaped public attitudes toward refugees, migrants, and policies governing refugees, asylees, and migrants more generally (Messing and Ságvári 2016; Sik,  Simonovits, and Szeitl 2016; Hangartner et al. 2019;   Stockemer et al. 2020; Brug and Harteveld 2021; Lutz  and Karstens 2021; Peshkopia, Bllaca, and Lika 2022).The refugee crisis heightened anti-immigrant attitudes among Europeans, with electoral consequences that strengthened anti-immigrant parties like Fidesz in Hungary, Golden Dawn in Greece, and Alternative für Deutschland in Germany.Although several studies have estimated the causal effects of exposure to refugees on antiimmigrant attitudes and voting patterns (Dinas et al.  2019; Hangartner et al. 2019), we emphasize that the refugee crisis is a contextual variable as well as an individual one.Even Europeans who never personally encountered a refugee during the crisis or in its aftermath live in countries in which the refugee crisis was a prominent news item and a subject of extensive political discourse.
In addition to the administrative, logistical, and ethical challenges that receiving countries like Hungary faced during the 2015 refugee crisis, the arrival of hundreds of thousands of refugees from the Middle East, South Asia, and Africa in Europe prompted new questions about European identity (Ammaturo 2019).Subsequent analyses focus on its implications for nationalism and national identities (Schenk 2021), for religious identity and the politics of religion (Schmiedel and Smith 2018; Peker  2022), and through a lens of racialization (Rexhepi  2018; Burrell and Hörschelmann 2019).
Given the importance attributed to race, religion, and European identity in shaping the discourse around the 2015 refugee crisis in Europe, we view the key distinction between the 2015 and 2022 crises to be the identity of those fleeing conflict.Whereas the refugees entering Europe from 2015 onwards were not Europeans, mostly not Christians, and racialized as non-white, those fleeing Ukraine were mostly white Christian Europeans. 4Of course, there are other differences between the refugee populations entering Central and Western Europe between 2015 and 2022, such as the nature of the conflict that drives the current crisis, its proximity to Europe, reasons for migrating, as well as their legal status (we discuss these differences later).These all might produce a more accommodating environment for Ukrainians than had been the case for Afghan refugees.
The ongoing 2022 Ukrainian crisis has not yet generated a significant body of academic research on its effects, although preliminary work has already identified some important contrasts between European responses to refugees from Ukraine versus Syria (Paré 2022; Pratt and  LaRoche 2022).Comparing general trends in public opinion requires post-February 2022 public-opinion data, and key sources like the European Social Survey have not yet released data that covers that period.Beyond the specific issue of Ukrainians in Central and Western Europe, though, early analyses have highlighted the oftensurprising degree of European solidarity with Ukraine since the outbreak of the crisis (Allin and Jones 2022;  Bosse 2022).They have also noted, however, that European supporters of Ukraine might not support resettlement within their own communities, implying that there are limits to such solidarity (see Clayton, Ferwerda,  and Horiuchi 2022).

The Hungarian Case
The 2015 refugee crisis deeply affected Hungarian politics and society.Prior to the crisis, Hungary's increasingly authoritarian regime had undermined many of the pillars of liberal democracy (see Bánkuti, Halmai, and Scheppele  2012; Kelemen 2015; Krekó and Enyedi 2018; Bernhard  2021).Today, Hungary is best described as a competitive authoritarian regime (Levitsky and Way 2020).In this political context, with anti-immigrant rhetoric already a central feature of Hungarian right-wing politics (Horvath,  Fox, and Vidra 2011; Korkut 2014), the inflow of refugees was easily politicized by the incumbent Fidesz government.
Orbán and his supporters characterized refugees as threats to the Hungarian nation and to state stability (Cantat and Rajaram 2019; Stivas 2023).This was accomplished at the discursive level through such tactics as erecting signs in Hungarian-thus for a Hungarian rather than a refugee audience-that warned refugees about their obligation to respect Hungarian culture and not to take Hungarian jobs, and through push polls distributed on behalf of Orbán that contained leading questions about refugees.Additionally, the government launched a broader campaign against supranational institutions such as the European Union, lambasting their unwillingness to protect European civilization and culture, and emphasizing national sovereignty to protect Hungary as a Christian European nation (Fekete 2016; Majtényi, Kopper, and  Susánszky 2019; Scott 2020).
Government rhetoric also legitimized anti-immigrant public opinion.In 2018, in his annual state of the nation speech, Orbán addressed the issue of migration and claimed that they [Western countries in the EU] want us to adopt their policies: the policies that made them immigrant countries and that opened the way for the decay of Christian culture and the expansion of Islam.They want us to allow in migrants and to become a country with mixed populations.(Orbán 2018a)   A few months later, in his speech on the 170th anniversary of the Hungarian Revolution of 1848, he added that Europe is now under invasion … Brussels is not defending Europe and does not stop immigration, but supports and organizes the inflow of people.It wants to dilute the population of Europe and to replace it, to cast aside our culture, our way of life and everything which separates and distinguishes us, Europeans from the other peoples of the world.(Orbán 2018b)   Recent work has documented that during the election campaign in 2018, the framing of the refugee crisis made it a salient domestic issue that shaped voter opinion (Márton and Goździak 2018; Cantat and Rajaram 2019).Moreover, Hungarian settlements where refugees were present were subsequently more likely to vote for far-right candidates and to support anti-immigrant positions (Gessler, Tóth, and Wachs 2022).
Given the depth of the anti-immigrant sentiment in Hungary, a renewed influx of people from another foreign conflict might have been similarly politicized-the same government still holds power, and the 2022 Ukrainian crisis began just over a month before Hungary's 2022 elections.And yet there is no evidence of anti-refugee rhetoric following Russia's invasion of Ukraine.Following his visit to the humanitarian transit zone in March 2022, the Hungarian prime minister claimed that for them [refugees from Ukraine] fleeing war is a shocking experience, a traumatic experience.After fleeing war, the first good news in their lives comes here in Budapest … when they are provided with food and water-and also accommodation for those who need it.We are also providing special care for children, we have medical services, and soon there will be employment agency representatives.Some people-the majority-move on; but those who stay in Hungary not only need food and shelter, but they eventually need jobs.In Hungary, fortunately, today there are more jobs than people in their working age … we in Budapest offer a happier future for those in need.(Orbán 2022c)   In May 2022, shortly after his election victory, the prime minister once again made it clear that Hungary is devoted to help refugees from Ukraine: In this war, Ukraine has been attacked and Russia is the aggressor.We are supporting Ukraine, and we have launched the largest humanitarian aid operation in Hungary's history.Proportionally, we have allowed in the largest number of refugees, and we are providing help for those in need.We will help Ukrainian refugees … Ukrainians can count on Hungary and on the Hungarian government.(Orbán 2022a)   At the outset of the crisis, this lack of anti-refugee politicking might have been explained by the fact that some of the first refugees entering Hungary were from Ukraine's small Hungarian-speaking minority, many of whom already held Hungarian citizenship (Erőss 2022).But this number was small relative to the vast majority of refugees who were Ukrainian speakers without any ethnic, national, or linguistic connection to Hungary.
The Fidesz government emphasized the racial, religious, and cultural differences between refugees from Europe and non-European countries.Orbán claimed that assisting refugees from Ukraine is an "elementary human, Christian instinct" and added that one does not have to be a "rocket scientist" to see the difference between "masses arriving from Muslim regions in hope of a better life in Europe" and helping Ukrainian refugees who have come to Hungary fleeing war (About Hungary 2022).The prime minister framed the migration waves from outside of Europe as part of a great European population replacement programme, which seeks to replace the missing European Christian children with migrants, with adults arriving from other civilizations.(Orbán  2022b)   and warned the Hungarian population about the danger of people arriving from outside of Europe: There is a world in which European people are mixed together with those arriving from outside Europe.Now that is a mixedrace world.And there is our world, where people from within Europe mix with one another, travel around, work, and move to other places.So, for example, in the Carpathian Basin we are not mixed-race: we are simply a mixture of people living in our own European homeland … creating [our] own new European culture … we are willing to mix with one another, but we do not want to become peoples of mixed-race.(Orbán 2022b)   Noting that Hungary's nationalist approach to migration policy will have long-term implications for the European Union's approach to migration, refugees, and asylum (Trauner and Stutz 2021), Hungary's response to the humanitarian crisis caused by Russia's invasion of Ukraine has first-order implications for migration policy across Europe.

Data
To study the evolution of Hungarian public opinion toward refugees over the past decade, we conducted two original surveys in April 2022 (just as refugees began flowing into Hungary) and in November 2022 (Pepinsky, Reiff, and  Szabó 2024).We partnered with the Hungarian survey firm TÁRKI, one of the most well-established polling firms in Hungary.TÁRKI selects respondents via random selection sampling resulting in surveys that are representative of the Hungarian adult population. 5Our sample includes 1,023 Hungarian adults in April and 1,000 adults in November.We collected data on the demographic characteristics and political orientations of the survey respondents, among other variables.We merged these data with administrative data on local demographic and economic factors in order to situate our respondents in their local contexts.
We combined these original survey data with two existing sources of data on Hungarian public opinion.First, we use four surveys conducted by TÁRKI in previous years that include questions about refugees. 6These were conducted in April 2014, January 2016, October 2016, and January 2017.The timing of these surveys allows us to compare Hungarian public opinion prior to the 2015 crisis (April 2014) with subsequent public opinion changes, culminating in our surveys that follow the Russian invasion.Second, we combine our 2022 survey results with recent survey data from the European Social Survey (ESS), which is also administered in Hungary by TÁRKI.For these analyses, we use ESS data from the previous six rounds (2010 through 2020, at two-year intervals).Again, the timing of the ESS rounds allows us to compare Hungarian public opinion prior to the 2015 crisis to subsequent survey rounds. 7

Results
We begin by examining trends in Hungarian public opinion over time.Figure 1 shows the results for six survey waves in which respondents were asked their views about refugees.The trends are clear.What do you think?Hungary should allow in...
Prior to the 2015 refugee crisis, a plurality of Hungarians favored admitting at least some refugees, but Hungarian public opinion trended in a steadily anti-refugee direction in subsequent years, resulting the majority of Hungarian respondents opposing all refugees by the end of 2016. 8With the onset of the war in Ukraine, public opinion toward refugees improved dramatically, with the result that nearly 90% of all respondents reported that Hungary should admit some or all refugees in April 2022. 9That number declined by November 2022, but still remained significantly higher than at any time in the past decade. 10t is helpful to compare these results to existing findings about the durability of migration attitudes (see Kustov,  Laaker, and Reller 2021). 11We find a major change in public opinion toward refugees in Hungary in 2022, whereas existing research based on cross-national panel data finds these attitudes to be stable. 12Our repeated cross-sections of public opinion data do not allow us to track individual opinions over time, but the sheer magnitude of this shift in public opinion means a substantial proportion of the Hungarian population must have changed its views about refugees between 2017 and 2022.The difference between our results may be attributed to one of three factors.First, it could be that Hungary's experience is not representative of other European contexts, owing to the sheer depth of the anti-immigrant and anti-refugee rhetoric nurtured by Orbán and Fidesz since 2015. 13econd, it could be that attitudes about refugees are distinct from general attitudes about migrants and migration policy. 14Third, perhaps the Ukrainian crisis has had a qualitatively different impact on public opinion that have previous migrant, economic, or other shocks due to historical memories, 15 due to geographic proximity, 16 or because there is a small ethnic Hungarian population in Ukraine. 17 To what extent are these changes driven by-or conditioned by-political developments within Hungary itself?Recall that the incumbent Fidesz government prevailed in national elections in April 2022, which suggests that these swings in public opinion must have also occurred among Fidesz supporters themselves.In figure 2, we break down opponents to admitting all refugees to Hungary by their partisan affiliation (Fidesz supporters, Opposition supporters, and other non-aligned voters). 18hese data reveal, first, that prior to the 2015 refugee crisis, Fidesz supporters were not particularly opposed to refugees; they turned decisively against refugees only after 2015.And yet even Fidesz supporters shifted decisively in a pro-refugee direction in 2022.Looking at respondents in the April 2022 survey only, we find that Fidesz supporters overwhelmingly supported admitting some refugees to Hungary, and were only slightly less open to admitting all refugees than were members of the opposition (see figure 3).This shift in Hungarian public opinion is surprising.Over the last decade, Fidesz has developed close relations with Russia as part of its Eastern Opening policy. 20The Russianfinanced Paks nuclear power plant and long-term gas contracts both provide evidence of close economic ties between the Fidesz government and Russia.Foreign relations are also closely linked: after Russia annexed Crimea in 2014, Hungary used its veto powers in NATO to block high-level NATO-Ukraine meetings and joint military exercises (Visnovitz and Jenne 2021).After the 2022 Russian invasion of Ukraine, Orbán described Ukrainian president Volodymyr Zelensky as his opponent, and blamed the EU's Russia policy for inflation and soaring energy prices.
Indeed, there is ample evidence that the Fidesz government rhetoric has moderated the public's historical aversion to Russia, with opinion polls indicating that the population's sympathy toward Russia has increased during the Fidesz era (Krekó 2016). 21A recent survey from May 2022 also indicates that 33% of the Hungarian population claimed that Hungary should be moving closer to Russia even after its invasion of Ukraine. 22We infer from these developments that the shift in public opinion that we have identified is not likely to be driven by popular perceptions of Russia as a security threat.If anything, a generally pro-Russian political environment should have decreased empathy for Ukrainians. 23krainians have not historically been characterized as part of Hungary's Christian, European heritage, and Hungarian political discourse traditionally has not emphasized any cultural similarity between Ukrainians and Hungarians.Prior to February 2022, Ukraine appeared in the Hungarian popular media for three main reasons.The first was in discussions of EU enlargement.Hungary generally supported Ukraine's membership in the EU, although this was justified on economic rather than cultural or religious grounds.24 The second focused on the Hungarian diaspora.In 2017, Ukraine introduced a language law that curbed minorities' access to education in their native tongues, which affected the Hungarian minority.In response, Hungary blocked Ukraine's membership in NATO until Ukraine restored ethnic Hungarian language rights (Magyarország Állandó NATO Képviselete Brüsszel 2017).The third was in the context of energy security.In September 2021, Hungary signed a 15-year natural gas supply agreement with Russia that guaranteed supplies through new routes via Serbia and Austria, bypassing Ukraine.Under this new deal, Ukraine lost millions of dollars in transit fees, leading Ukraine's foreign ministry to state that Hungary's gas deal was a "purely political, economically unreasonable decision" that was taken "to the detriment of Ukraine's national interests and Ukrainian-Hungarian relations" (Joly 2021).In response, the Hungarian news media was flooded with articles claiming that Ukraine's opposition to a new gas deal with Russia threatened both Hungary's economic sovereignty and its national security (Mandiner 2021).
Changes in Hungarian public opinion over time remain robust when we control for survey respondents' sociodemographic characteristics.We estimate the following linear probability model on a pooled cross-section dataset between April 2014 and November 2022: where y it is a dummy variable indicating that respondent i in wave t is opposed to admitting any refugees; Fidesz it is a Fidesz voter dummy; Wave t are wave dummies; and X 0 it is a vector of socio-demographic variables such as education, age, gender, settlement type, activity, and variables on religiosity (self-declared level of religiosity and frequency of participating in religious services) (refer to online appendix E).To understand the changing attitudes of Fidesz voters over time, we interact the Fidesz voter dummy and the wave dummies, while also allowing the wave dummies to control for time-specific factors such as the general economic situation of the country, which could confound these relationships.
Table 1 shows that on average, Fidesz voters are more hostile toward refugees than non-Fidesz voters (Model 1).
Model 2 allows this relationship to differ across survey waves, and shows that while in 2014 and in 2022 the probability that a Fidesz voter is opposed to admitting refugees was not larger than for non-Fidesz voters, during the first refugee crisis, it was significantly larger (by 11.3-17.6percentage points).We also note that general hostility toward immigrants was particularly high in 2016-2017, when migration was a salient domestic issue in Hungary, but dropped significantly by 2022-as the wave dummies show.Table 1 also shows that religious service participation and education are highly correlated with individuals' attitudes toward immigrants.More educated people, and people who participate in religious services, are significantly less likely to oppose the entry of refugees.
To provide further evidence on changes in Hungarians' attitudes toward immigrants during the two crises, we also analyzed data from European Social Survey (ESS) between 2010 and 2020. 25We find similar results based on the ESS dataset.Figure A26 in online appendix P shows that before the 2015 refugee crisis, Hungarians had a rather neutral opinion on whether Hungary became a worse or better place by people coming to live there, but during the first refugee crisis, Hungarian public opinion trended in an anti-refugee direction with a peak in the anti-immigrant sentiments in 2016.Following the invasion of Ukraine, public opinion toward refugees improved dramatically, especially among Fidesz supporters.Examining trends over time, we find that Fidesz voters had similar attitudes toward immigrants than non-Fidesz voters in 2010 and 2012, but were particularly opposed to admitting refugees to Hungary between 2014 and 2020.By April 2022, however, they were similar to non-Fidesz voters.While the ratio of respondents in support of immigrants declined by November, the pro-immigrant sentiments were still higher than at any time in the past survey waves. 26n the remainder of this section, we focus our analysis on data from April 2022, as analyses using data from November 2022 produce substantively identical findings. 27The exception is for analyses of gender and religion, which we analyze later in a separate section.To adjudicate between these possibilities, we embedded two experiments within our April 2022 survey that asked respondents about their receptivity to refugees fleeing conflict in a particular country.In the first, respondents were randomly assigned to respond to a question about either Afghanistan or Pakistan.In the second, they were randomly assigned either Ukraine or Belarus.Answers to these questions fall on a five-point Likert scale ranging from strongly disagree to strongly agree. 28he logic of our survey experiment is as follows.We suspect that Hungarian respondents' views on refugees are shaped by their understanding of the level of conflict they face in refugees' origin countries, and another is race, religion, and identity of the refugees.By asking respondents about refugees from Afghanistan versus Pakistan, we can hold roughly constant the racial and religious features of refugees while allowing the presence of conflict to vary.The same is true of a comparison of refugees from Ukraine and Belarus: at the time that our survey was fielded, it was an open question whether Belarus would send its own conscripts to fight alongside Russia in Ukraine, or if Russian troops would be stationed en masse in Belarus, so a future with Belarusian refugees was conceptually possible.Although we paired Afghanistan with Pakistan to distinguish between countries facing conflict and those that are not, our design does not require that respondents understand this difference.The conflict of Afghanistan from August 2021-surrounding the withdrawal of U.S. troops-and the war in Ukraine from February 2022 were highly salient events in the Hungarian popular media, while developments in Belarus and Pakistan were rarely present.Therefore, it is reasonable to assume that at the time of our survey experiments in April and November 2022, Hungarians were relatively more informed about conflict in Afghanistan and Ukraine than in Belarus and Pakistan. 29And yet even if Hungarians were not attuned to differences in the nature or scale of conflict between Afghanistan and Pakistan, the comparison of each with Ukraine and Belarus remains informative.Because our main objective is to understand how race, religion, and values shape preferences for refugees, our central assumption is that Hungarians were aware in 2022 that Ukraine had been invaded by Russia, but that Belarus had not been.
Figure 4 shows the distribution of responses across the four categories: the Hungarian mass public is more receptive to Ukrainians than to any other refugee population. 30o analyze these results further, we estimate an OLS regression that predicts the level of support for refugees (1 = lowest, 5 = highest) as a function of the interaction between presence or absence of conflict (present for Afghanistan and Ukraine, absent for Pakistan and Belarus) and whether or not the country is in Europe.This is equivalent to a difference-in-differences design, which we estimate via where Support ij measures support for admitting refugees on a 1-5 scale, Europe j takes the value of 1 for the survey question comparing Ukraine and Belarus and 0 otherwise, Conflict ij takes the value of 1 for respondents assigned Ukraine and Afghanistan and 0 otherwise, η i are respondent fixed effects, and ϵ ij is an error term, with standard errors clustered at the level of the respondent.We also estimated a fixed effects logistic regression model, where the outcome is 1 if the respondent agrees or strongly agrees that Hungary should welcome refugees from conflict in that country, and 0 otherwise.
The results appear in table 2. The positive and highly statistically significant coefficient on Europe × Conflict signifies that respondents were far more likely to agree to welcome refugees from Ukraine relative to refugees from any other country.The OLS model estimates an increase of 1.1 (on a 5-point scale, equivalent to a full standard deviation in magnitude) in support of refugees from a European country in conflict, compared to the increased support for refugees from a non-European country in conflict.
The coefficient on Europe is further evidence of the importance of race, religion, and values in explaining support for refugees in 2022, showing that respondents were more supportive of refugees from a non-conflict country in Europe (Belarus) than from a non-conflict country outside of Europe (Pakistan).On the other hand, Conflict alone is not significant in explaining more positive attitudes toward refugees-it only appears to be important if it affects Europeans.
To convey the magnitude of these relationships, figure 5 plots the predicted level of support, calculated from the OLS results in table 2, for each of the four countries as defined by the interaction of Conflict and Europe.The 2022 crisis has shifted Hungarian public opinion in favor of refugees, but overwhelmingly in favor of white Christian European refugees fleeing open conflict.

Additional Evidence on Refugee Preferences
In this section, we provide additional evidence that reveals how respondents' attitudes-especially Fidesz voters' attitudes-are affected by the demographic characteristics and ethnicity of the immigrants.Figure 6 shows the average support of refugees from different source countries by partisanship. 31While Fidesz voters are more supportive toward refugees fleeing conflict in Ukraine than the population average, they are slightly less welcoming toward refugees from the other three countries.
In online appendix S, we model the relationship between respondents' socio-demographic characteristics, partisanship, religious identity, and their attitudes toward immigrants.Figure 7 shows the relative support of Fidesz voters when we control for individuals' socio-demographic characteristics (thus, the bars represent the estimated coefficients of the Fidesz voter dummy in Equation A2). 32 To test the robustness of our estimates, panels include different sets of control variables (as in table A21). Figure 7 reveals that Fidesz voters (relative to non-Fidesz voters) are more open-by 3.1-4.5 points on a 100-point scale, depending on the exact specification-for refugees fleeing from Ukraine, while Fidesz supporters' attitudes toward refugees from the other three source countries are always more negative (although insignificant).
The April 2022 survey included four additional questions designed to measure the importance of various skills, race, religion, and values in shaping Hungarian public opinion.This allows us to disentangle among the different dimensions of identity previously captured through the comparison of European and non-European refugees.To probe more deeply into how culture and its different manifestations affect respondents' opinion on migrants, we included an experimental treatment within one of these questions, to compare the importance of two manifestations of culture: values and race.The wording of the questions was "How important should it be for refugees to have good educational qualifications/work skills that Hungary needs/ having the same values as Hungarians do/arriving from a country with white European heritage; and being Christian?" 33First, we test whether or not asking about the importance of white European heritage or common values with Hungarians affects respondents' views (see table 3).
We find no difference in the distribution of responses based on which of these questions we ask: χ 2 3 ð Þ = 5:7, p = 0:13.This is evidence that race and values are indistinguishable from one another as explanations for Hungarian public opinion on refugees.Treating each experimental group as its own question, we then compare them to the importance of refugees being Christian, asked of all respondents (see table 4).We find that among Hungarian respondents, views about the importance of race, religion, and values are strongly related to one another, but views about race and values are more closely aligned with one another than they are with views about religion.
Figure 8 examines how these views relate to respondents' party preferences, showing the average importance of these five characteristics across Fidesz, opposition, and other voters.We see lower importance attributed to Christianity than to race and values, a conclusion that holds across parties.
Figure 8 also indicates that Fidesz voters' opinions about the importance of the necessary work skills and education do not differ from non-Fidesz voters' opinion.But Fidesz voters have a much stronger preference for immigrants with the same values as Hungarians, who come from a country with white European heritage, and who are Christian.These results also hold in a multivariate context. 34 Finally, we also investigate how ethnicity affects mass public opinion (and how this differs across individuals' partisanship).To this end, we added the following question to our April 2022 survey: "Should Hungary welcome immigrants from these ethnic backgrounds, so long as they are entering the country legally and have no record of criminal activity?", with seven different ethnicities: Hungarian, German, Russian, Chinese, Arab, Piresian and Piresistani.The last two of these-Piresians and Piresistani-are fictional ethnic groups; we include them to measure the respondents' general hostility toward truly unknown people.Possible answers were on a 4-point Likert scale, with 1 corresponding to "should not at all be welcome", and 4 to "should be welcome." 35igure 9 shows the changing altitudes of Hungarians across immigrants' ethnicity by partisanship.Hungarians in general are very welcoming toward ethnic Hungarians and German immigrants.They are, however, rather opposed to Russians and Chinese, and mostly hostile to Arabs, Piresians, and Piresistani.Fidesz voters are more welcoming toward the ethnic Hungarian immigrants-who have the same national identity by definition-than any other groups of voters.Additionally, Fidesz voters have similar views as other voter groups toward Germans, Russians, Chinese, and even Piresians and Piresistani and they are rather opposed to Arabs.Again, these findings hold in a multivariate context. 36

Gender and Religion: Additional Results from November 2022
As noted earlier, nearly all results using April 2022 data are substantively identical when using November 2022 data. 37There are two notable exceptions.First, our November survey contained a new item designed to adjudicate how the anticipated gender composition of refugees affects Hungarian public opinion.Second, the relationship between religion and refugee support differs dramatically between April and November.We discuss these two findings in turn.
In addition to framing refugees with reference to their race, religion, and values, Orbán has also noted repeatedly that arriving Ukrainians are mainly women and children, while refugees coming from Africa, the Middle East, and South Asia were young men.He argued that everyone can see the difference between the frightened women fleeing from the fighting in our neighbouring country with their bags and children, and the migrants from thousands of kilometres away besieging our borders.Hungary helps refugees, but continues to reject migration.(Orbán 2022d)   Indeed, in 2015, Orbán claimed that 80% of immigrants are male and that they [male immigrants] look like an army rather than a group of refugees … even if other European countries deal with their demographic issue with allowing in young, warrior-like males, we cannot accept this.(Híradó 2015)   Thus, a natural concern is that our results may be driven by the Hungarian government's framing of the gender composition of Ukrainian refugees. 38If true, Hungarians are more welcoming of Ukrainian refugees than Afghan refugees because they assume that Ukrainian refugees are mostly women and children, whereas they assume that Afghan refugees are young men.To address this concern, we designed another survey experiment in the November wave that asked respondents about their receptivity to male versus female and children refugees fleeing from Afghanistan versus from Ukraine. 39e predict refugee attitudes as a function of refugee gender/age (males versus females/children) and the source country of refugees (Ukraine versus Afghanistan) using the following specification where Support i measures support for admitting refugees, Europe takes the value of 1 for the survey question comparing Ukraine and 0 for Afghanistan, Female takes the value of 1 for respondents assigned female and children refugees and 0 for male refugees, and ϵ i is the error term.In these regressions, X 0 i captures respondents' sociodemographic characteristics, their party preferences, and religiosity.
Table 5 indicates that Hungarian respondents are significantly more welcoming toward women and children than men (a difference of 11-14 points on a 0-100 scale).But as shown in figure 10, Hungarians are still more receptive to Ukrainians, revealing the continued importance of race, religion, and values.
Turning now to religion, we observe a meaningful difference between April 2022 and November 2022 in how religious participation relates to refugee support. 40To put these differences in context, we collected historical data on the anti-immigrant sentiments of various religious and non-religious groups from the first half of the 2010s, i.e., from a period when immigration was not an important or a salient issue.Figure 11 shows the proportion of respondents who say that no immigrants should be allowed to Hungary, by the frequency of service participation, in ten survey waves between 2011 and 2022. 41rior to 2015, frequent religious participation is associated with lower opposition to admitting refugees.As anti-refugee sentiment increased in 2016 and afterwards, differences in refugee support generally disappeared, suggesting that religious Hungarians were particularly receptive to rhetoric about race, religion, and values.April 2022 saw the dramatic drop in opposition to refugees that we identified previously.But by November 2022, the rise in anti-refugee sentiments was much larger among those who never participated in religious services than among religious participants. 42o investigate the sources of the November 2022 changes, we model support for refugees as a function of the interaction between individual religiosity and survey wave. 43Figure 12 predicts the probabilities that respondents oppose the admission of all refugees, and shows that even when we control for individual's socio-demographic characteristics, religious service participants were generally less opposed to refugees prior to the first refugee crisis.We conclude that Russia's invasion of Ukraine generated strong and unanimous support for refugees in its immediate aftermath, but over time the relationship between religion and refugee support in Hungary has returned to its pre-crisis baseline pattern (see also Kustov, Laaker, and  Reller 2021).

Contextual Factors and Refugee Support
In this section, we complement our individual-level results with information about respondents' local environments.This extends our argument about race, religion, and values to a different level of analysis, to examine how factors such as the local strength of Fidesz, local demographic composition, geographic factors, or local economic conditions shape respondents' views.Importantly, none of the analyses in this section overturn the substantive conclusions we have drawn in previous sections.Although we will show that respondents' local environments explain additional variation in refugee support, our argument that race, religion, and values explain the sharp increase in support for Ukrainian refugees in 2022 remains unchanged, as do our empirical findings about the individual-level predictors of public opinion toward refugees across survey waves.
We study the contextual determinants of public opinion by merging our April and November 2022 surveys with settlement-level data compiled from the Hungarian Central Statistical Office.Settlements are the smallest administrative units in the country; there are a total of 3,177 settlements in Hungary, including the 23 districts of Budapest.In our survey, data are drawn from 81 settlements and 23 districts from Budapest in April and 82 settlements and 23 districts in November.We collect data on local demographic factors like Christian population share, Roma population share, and income per capita, as well as other local factors such as Fidesz vote share and the distance to Hungary's border with Serbia and with Ukraine. 44dopting a multilevel modeling approach (Steenbergen and Jones 2002), we begin with a simple variance decomposition (as specified in Equation A6 in online appendix AA) to estimate the relative importance of settlement-level factors in explaining individual attitudes, and then model contextual factors directly using the following model: where y ij is the attitudes toward immigration (on a 0-100 scale), Z 0 j is a vector of settlement-level explanatory variables, X 0 ij is a vector of individual explanatory variables, α 00 is the average level of support, α 0j is the settlement-level  error term with variance σ 2 α , and ϵ ij is the individual-level error term.
The main results appear in tables 6-9.Our first result is that settlement-level factors matter: a substantial proportion (32%-54%) of total variation is due to variation between settlements (see the last rows of each Panel A).Approximately 8%-28% of this between-settlement variation is explained by the settlement-level explanatory variables that we included.We also find that, consistent with our expectations, the settlement-level Christian population share is negatively correlated with support for non-Ukrainians (table 6) and non-Hungarians (table 8) in the April survey wave.By contrast, settlement-level Roma population share is negatively correlated with support for non-Ukrainians (table 7) and non-Hungarians (table 9) in November.We also find in November that respondents in settlements with a larger foreign population share are more welcoming of non-European refugees and immigrants.
Finally, in online appendix AA we investigate whether the effect of residential exposure to religious majority is larger for religious individuals by allowing the effect of individual religiosity on attitudes toward immigrants to vary across settlements with different Christian share. 45e find that settlement-level Christian population share explains anti-immigrant attitudes primarily among      A34). 46Religion not only is correlated with individual anti-immigrant sentiments, a religiously homogeneous context strengthens this correlation. 47Table A33 and table A34 show that the correlation between settlement-level Christian population share and anti-immigrant attitudes is larger for those supporting the incumbent government.
Although our main conclusions about the dramatic prorefugee shift in Hungarian public opinion remain unchanged in the multilevel analysis, the results in this section further reveal how our argument about race, religion, and values fares when examining individual public opinion in its local context.Respondents in settlements with large Christian majorities and large Roma populations are particularly opposed to non-European refugees, which is consistent with our argument about the primacy of identity factors for explaining refugee opinion.Further research can explore these and other dimensions of local context in shaping public opinion toward immigrants and refugees. 48

Conclusion
Our paper has used new survey and administrative data from Hungary to study how the Ukrainian refugee crisis of 2022 has shaped public opinion toward refugees in a highly illiberal political environment in which antimigrant rhetoric has been a mainstay of political discourse for over a decade.The 2022 crisis produced an overwhelming shift in public opinion in favor of accepting refugees in Hungary, countering a trend of growing antirefugee public opinion.This finding is substantively important on its own, as it reveals the power of external events to shape public opinion on refugees in profound ways.
But our main finding is that this shift in public opinion is specifically driven by changes in the composition of refugees, specifically by reactions to Ukrainian refugees, and does not apply generally.The Hungarian mass public remains opposed to refugees entering from countries that are outside of Europe, even those facing violent conflict that drives their citizens to seek refuge abroad.Ours is the most comprehensive quantitative evidence available that Europe's alleged "refugee hypocrisy" (Traub 2022) is widely felt among the mass public.We have documented that race, religion, and values are important predictors of attitudes toward migrants, and this is particularly true among Fidesz supporters.We have also tested whether additional differences between the refugee populations entering Europe between 2015 and 2022 have played an important role in guiding Hungarians' anti-immigrant attitudes such as the nature of the conflict that drives the current crisis, its proximity to Europe, reasons for migrating as well as the legal status of immigrants.In all cases, the results remain robust and the findings highlight the important roles that race, religion, and values had in shaping Hungarians' attitudes.Looking at the interaction between individual and contextual factors reveals that in general, the negative relationship between settlement-level religiosity and support for non-Ukrainian refugees and non-Hungarian immigrants is particularly strong among religious respondents and Fidesz supporters.
Although our findings explain aggregate shifts in public opinion across Hungary toward refugees, a more finegrained empirical approach is needed to address several outstanding questions about how race, religion, and values interact in the context of refugee crises in Hungary.For example, we lack data on Hungarian opinion toward nonwhite refugees arriving from Ukraine into Hungary, or on Hungarian-speaking Roma from Ukraine.Similarly, we lack data on how the two crises affected Hungarian opinion toward the country's own Roma minority within its borders.Understanding Hungarian reactions to nonwhite minority communities such as the Roma can reveal further insights about how race and values shape public opinion during times of crisis.Separately, our data do not allow us to dismiss the possibility that anti-refugee sentiment in the mid-2010s followed from the belief that these refugees should have remained in the first countries that they entered, rather than continuing their journey to Hungary before seeking asylum.
Future research may also examine race, religion, and values in more depth.When Hungarians are asked about the desirable characteristics of refugees entering their country, views on race, values, and religion are all closely aligned with one another.In addition to exploring how these patterns vary across Europe, future research may build on these results-following the approach of Helbling and Traunmüller (2020) or Adida, Lo, and Platas  (2019)-to distinguish among the various facets of identity in contemporary European politics.Relatedly, future scholarship may probe more deeply into the role of religion in shaping attitudes toward refugees.In the Hungarian context, these effects might be explained by the politicization of churches, or by the fusion of religion and national identity in government rhetoric, or by the direct role that Fidesz politicians play in Hungarian religious organizations.Differences between the organization of Protestant and Catholic churches might also lead to different responses by Christian denomination.
Scholars of public opinion in competitive authoritarian regimes can build on our findings to characterize how public opinion shapes-and is shaped by-Fidesz's governing strategy.The Ukrainian refugee crisis of 2022 shortly preceded legislative elections that returned Fidesz to power with a strong majority, meaning that a highly anti-immigrant party won an election in the midst of a serious refugee crisis.We have shown that Fidesz voters did indeed follow other Hungarians in becoming more open to refugees in the wake of the 2022 refugee crisis, but our analysis of how race, religion, and values affect Hungarian public opinion toward refugees is consistent with the Fidesz government's emphasis on European civilization as defined in racial and religious terms.These results from Hungary suggest that even as the war in Ukraine has upended politics as usual in Central Europe, it may not have fundamentally changed the logic of illiberal politics in Europe's authoritarian regimes.

Supplementary Materials
To view supplementary material for this article, please visit http://doi.org/10.1017/S1537592724000410. rely on the question of is "Hungary is made a worse or better place to live by people coming to live here from other countries?"that was also asked in our April as well as in our November 2022 waves.We used the same wording as well as the same response category: a 0-10 scale, where 0 is the much worse and 10 is the much better end of the spectrum.We re-scaled these, and all subsequent answers, to a 0-100 scale.26 See the estimated equations and the regression outputs in table A17, in online appendix P. 27 In online appendix N, all subsequent figures and tables are replicated for the November 2022 survey wave.28 The precise wording of the questions is included in table A18 of online appendix Q, both in English and in Hungarian, with italics highlighting the manipulation, while table A19 presents the characteristics of randomly assigned respondents across the different questions.29 Online appendix R justifies this assumption using Google search data from Hungary.Figures A27 and A28 plot the absolute search volume for the words "Ukraine" and "Belarus" in 2022 as well as for "Afghanistan" and "Pakistan" in 2021 respectively.The uptick in searches on Afghanistan in August 2021 as well as the peak in searches on Ukraine in February 2022 provide clear evidence that these countries were salient in Hungary.Figures A29 and A30 show the popularity of the terms Ukraine and Belarus as well as Afghanistan and Pakistan over time, which reflects how many searches have been done for the particular term relative to the total number of searches on Google in Hungary.Afghanistan was more salient than Pakistan in 2021 and Ukraine was more searched than Belarus in 2022.Additionally, "conflict" and "war" were among the most frequently searched words with Ukraine and Afghanistan, while Google searches on Belarus or Pakistan were typically not combined with conflict-related expressions (see table A20).Even in 2022, the most frequently searched word with Afghanistan was "war".30 While the legal status of the people who have fled Ukraine and Afghanistan is different (see the definition in online appendix A), it is unlikely that this difference explains these results.First, the vast majority of Hungarians never personally encountered a refugee during the 2015-2016 crisis given the small number of people staying in Hungary (see tables A2 and A3).Second, public opinion is unlikely to be driven by any meaningful difference in the social costs associated with having refugees or people with temporary protection status (see table A5).In sum, Hungarians' attitudes are not primarily affected by individual contact or by rational cost calculus.Instead, the refugee crisis is a contextual factor that affects public opinion responses in the aggregate.31 We transformed all ordinal scales to a scale of 0-100, so that we are able to compare the strength of the effects across different questions with different ordinal scales.32 See the estimated equations and the regression outputs in online appendix S. 33 Possible answers ranged on a scale of 1-4 (with 1 being "not at all important" and 4 being "very important"), which we again transformed to a 0-100 scale.The third and fourth of these options were assigned randomly, so we have a total of four questions with five outcomes.Table A22 in online appendix T lists the questions used to capture subjects' opinion on the importance of different cultural, educational, and religious background of immigrants, while table A23 provides descriptive evidence for the successful randomization.34 See the estimated equations in online appendix V and the regression outputs in table A24.35 A similar question about "Piresians" was asked in previous TÁRKI survey waves over the past two decades."Piresistani" is our invention: their ethnicity should also be unknown for the respondents, but their name sounds more Muslim than "Piresians."We randomized these two questions so that a random half of our sample obtained the question with Piresians, and the other half obtained the Piresistani.Table A25 in online appendix W lists the questions in English and in Hungarian, while table A26 shows that randomization was successful.36 See the estimated equation in online appendix X and the regression outputs in table A27.We note that Hungarians are strongly opposed to admitting Piresians and Piresistanis, but the difference between the attitudes toward the two are not statistically significant.37 Online appendix N replicates all results on the survey wave in November and compares them with the results in April.38 Online appendix Y breaks down refugees and Ukrainians with temporary protected status staying in Hungary by their age and gender.The number of accepted refugees (male and female alike) in Hungary was very low during the first refugee crisis: this rules out the concern that our results are driven by Hungarians' personal encounters with male refugees (tables A28 and A29).It is true that most Ukrainians (66%) with temporary protected status are female (table A31).39 The question was worded as follows: "To what extent do you agree or disagree with the following statement?
The Hungarian government should allow the entry of [adult men]/[adult women and children] fleeing conflict in [Ukraine/ Afghanistan]"? 40 Tables A21, A24 and A27 in online appendices S, V, and X show that in April 2022, individual service participation was not correlated with anti-immigrant sentiments toward refugees with different skills and characteristics.By November, this changed dramatically: more frequent service participants were significantly more supportive toward all refugees.See tables A12, A15, and A16 of online appendix N. 41 All surveys were conducted by TÁRKI, and the wording of the question was identical.42 This trend also holds for changes in public opinion toward refugees by source country (figure A33) and by ethnicity (figure A34).43 The model is Equation (A5) in online appendix Z, and table A32 shows the estimation output.44 We also included the share of Catholics and Protestants separately, net income per capita (instead of gross), and unemployment rate.All are insignificant in all specifications, so we do not report them here; see online appendix AB. 45 We estimate Equation A7 as specified in online appendix AA.We only do this analysis for the April wave, when the settlement-level share of Christians is strongly significant.46 This heterogeneous effect is significant for attitudes toward immigrants with different ethnic background except toward Ukrainians and ethnic Hungarians.47 We tested three aspects of individual religiosity: identity, practice, and affiliation.All measures suggest similar results.48 Online appendix AC extends the analysis and estimates the effect of settlement-level Roma and Christian shares on the support of refugees during the precrisis period of 2011-2014.

Figure
Figure 1 Trends in public opinion toward refugees, 2014-2022 Future research can help to disentangle these possibilities, although the November 2022 results suggest a reversion toward earlier patterns in Hungarian public opinion as Russia's war in Ukraine continues. 19

Figure
Figure 2 Opposition to refugees by party, 2014-2022 Figure 3 Public opinion toward refugees by party, April 2022 If respondents are equally open to all potential refugees, we may conclude that the pro-refugee shift in Hungarian public opinion is unconditional.If they are more open to European refugees from Ukraine or Belarus than to non-European refugees from Afghanistan or Pakistan, we can conclude that Hungarian respondents strongly prefer European refugees who presumably have similar values to them.If they are more open to refugees from Ukraine and Afghanistan than from Belarus and Pakistan, we can conclude that the existence of conflict is the key feature explaining shifts in public opinion.If they are open only to refugees from Ukraine, we can conclude that the change in preferences is driven by the fact that the 2022 crisis affected mostly white European Christians fleeing conflict.Our survey design allows us to distinguish among these four possibilities.
Figure 4 Public opinion toward refugees by source country, April 2022 Strongly disagree Figure 5Predicted support for refugees, difference-in-differences design

Figure 7
Figure 7The relative support of Fidesz voters (to non-Fidesz voters) toward refugees from different source countries (estimated coefficients), April 2022 Figure 8The importance of immigrant characteristics and skills by party, April 2022 Figure 10Predicted support for refugees by source country and by gender, November 2022 Figure 12Opposition to refugees by religious service participation, 2011-2022 2 Opposition to refugees by party, 2014-2022 6 Perspectives on Politics What explains the decisive shift in Hungarian public opinion toward refugees in 2022?On one hand, it could be that the existence of a refugee crisis within Europe has shifted Hungarian public opinion about all refugees, showing that ordinary civilians may face political conditions that are not of their own making.But on its face, this appears less plausible than an alternative interpretation-commonly invoked to explain not just Hungary's responses to the Ukrainian crisis, but those across Europe more generally(Pratt and LaRoche 2022;  Traub 2022) -that the distinctive feature of Ukrainians in 2022 relative to the 2015-2016 refugee crisis is that the latter involved non-white, non-European Muslims, and the former affected mostly white European Christians.
Notes: Robust t statistics in parentheses.*,**and *** denote significance at 10%, 5% and 1% level, respectively.The dependent variable is a dummy variable indicating that respondents are opposed to admitting any refugees.Control variables are included (see online appendix E).8 Perspectives on PoliticsRefugee Preferences: Experimental Evidence

Table 3
Experimental results comparing race and values, April 2022 Notes: This table compares the distribution of responses to a question about the importance of refugees have a specific characteristic, where two options were assigned randomly to respondents: arriving from a country with white European heritage versus having the same values as Hungarians.Responses of "Don't know/refuse to answer" are excluded.The table shows the weighted distribution across the share of the responses.

Table 4
Race, values, and religion compared, April 2022The panels compare the distribution of responses of the importance of refugees being Christian (column variable) with the importance of coming from a country with a white heritage or the same values as Hungarians (row variables).Responses of "Don't know/ refuse to answer" are excluded.Columns of the table show the weighted distributions across the share of the responses.In online appendix U, figureA31shows the distribution of responses for Panel A, while figure A32 presents the distribution for Panel B. Notes:

Table 5
Refugee gender and source country, November 2022

Table 6
Variance decomposition and MLM estimation, refugees by source country, April 2022

Table 7
Variance decomposition and MLM estimation, refugees by source country, November 2022 In online appendix O, we quote Viktor Orbán himself in explaining his pro-Russian politics prior to the February 2022 invasion of Ukraine.24 See selected speeches of the prime minister on Ukraine joining the EU in online appendix O. 25 The ESS surveys were conducted by TÁRKI.Here, we