Are anti-immigrant attitudes the Holy Grail of populists?

In this paper we aim to discuss attitudes towards immigrants in a European context and analyse drivers of anti-immigrant attitudes such as the feeling of control, basic human values, political orientation and preferences related to right-wing populism. Based on data from the European Social Survey, we first describe how attitudes of people in Europe changed throughout a period of almost two decades (between 2002 and 2018). We will show that although attitudes are influenced by a number of demographic and subjective features of individuals, on the macro-level they seem to be surprisingly stable, yet hide significant cross-country differences. Then, we zoom in to the three most significant elements influencing attitudes towards immigrants: the feeling of control, basic human values, and political orientation. Applying a multilevel model we test the validity of three theories about factors informing attitudes towards immigrants—competition theory, locus of control, and the role of basic human values—and include time (preand post-2015 refugee-crisis periods) into the analysis. In the discussion we link ESS data to recent research on populism in Europe that categorizes populist parties across the continent, and establish that the degree to which anti-migrant feelings are linked to support for political populism varies significantly across European countries. We show that right-wing populist parties gather and feed that part of the population which is very negative towards migrants and migration in general, and this process is also driven by the significance awarded the value of security vis-à-vis humanitarianism.


Introduction
The inflow of refugees to Europe in 2015 caused deep political fractures within the EU and is still one of the key issues around which debates and ideological clashes in European politics crystalize. The mass arrival of people from the Middle East and Africa triggered a rise in political populism and became a key topic for populist political parties. But this is not the first time Europeans have experienced a mass inflow of asylum seekers. If we look at mass refugee flows in a historical context, we find that the most recent peak in the number of asylum seekers, which took place in the 1990s (triggered by the war in the Balkans and armed conflicts in the Middle East, Iran, Iraq, and Afghanistan, and the Horn of Africa [Somalia]), was of a similar size and composition. '…when we look at the total numbers per half a decade and compare the 1990s with the first half of the 2010s then the number and origin in recent years do not deviate from what Western Europe experienced two decades ago.
[…] Furthermore, the sudden increase in numbers in 2014 and especially 2015 had a clear cause, the civil war in Syria, and there were no signs that indifferent masses of poor migrants from the Global South had been unleashed' (Lucassen, 2018, p. 385). Still, the reception of this new wave of asylum seekers was met with an apocalyptic tone from mainstream politicians, and rising, untrammelled fear in public and political discourse, as well as in the media (Chouliaraki & Srolic, 2017;Gheorgiou & Zaborowski, 2017).
Fear of migration is embedded in the process of polarization and rising populism (Mudde, 2016;Kende & Krekó, 2020). Right-wing populist narratives are a crystallization of wider uncertainties within the population which are being brought about by global challenges, such as rapid technological transformation via digitalization and robotization, climate change, the increasing influence of social media and the rising significance of fake news sources, just to mention the major stressors. These phenomena are difficult for individuals to follow or explain, especially those with weak educational backgrounds or a low level of interest in grasping the complexities these issues entail. Quite a number of political forces in Europe (and the world) benefit from people's growing insecurity by offering simplistic explanations for extremely complex phenomena and polarizing their populations. Migration, a phenomenon that can be interpreted within a simplistic nationalist framework that divides society into groups of 'us' and 'them, ' of 'nationals' and 'foreigners, ' has become a focal point of right-wing populist political and public discourse in which these fears may condense.
In this paper, we aim to contribute to the discussion about the drivers of anti-immigrant attitudes and how they are linked to basic human values, political preferences, and an openness to political populism. We aim to arrive at a better understanding of how attitudes to migration evolved in the context of the mass inflow of immigrants (refugees) in 2015, and whether the role of factors identified as those shaping attitudes towards immigrants in former research became more or less influential in the four years following 2015. Based on data from the European Social Survey (ESS), we first analyse geographical and timeseries trends in attitudes and show their differences across countries as well as changes since 2002. Then we zoom out to two of the most important determinants of attitudes: basic human values, and political preferences. In the next step, we offer a multilevel model for investigating the intersecting effects of the factors that trigger pro-and anti-migrant attitudes and compare the significance of these factors in time. Finally, in the discussion we analyse how openness to populist parties is linked to anti-immigrant attitudes and what country-specific differences can be identified. Linking ESS data to recent research on populism in Europe (Roduijn et al., 2019) that categorizes populist parties across the continent, we try to establish the degree to which anti-migrant feelings are linked to support for political populism and whether this link has changed in the past five years.

eoretical baground and resear questions stemming from this
An extensive amount of theoretical and empirical literature discusses the origins of antiimmigrant attitudes. Without striving for a complete overview, we will highlight the three theories that most inspired this analysis, and that we seek to test, including the element of time, by comparing pre-and post-2015 snapshots. We will look into the strength of association of factors described by traditional group conflict or competition theory, those explained by theories that link basic human values and attitudes, as well as political preferences and attitudes towards immigrants. In this part of the paper, we briefly introduce the theories that have been widely tested in the scholarly literature. We aim to analyse how the strength of the explanatory models has changed in time and test the validity of the latter on data collected before and after the 2015 refugee crisis.

Group competition and control theory
A frequently applied and tested theory concerns the economic rationale behind attitudes about migration. Group conflict theory (Blalock, 1967;Quillian, 1995;Mueleman et al., 2009) postulates that negative attitudes are driven by perceived competition for scarce goods such as jobs, housing, welfare services or wealth in the host society. Blalock (1967) proposed that the level of perceived group threat-the subjective perception of competitionplays a crucial mediating role in the evolution of negative outgroup attitudes. A number of empirical studies have tested this theory but their conclusions are not unanimous. Some argue that the size of the immigrant population has a significant impact (Quillian, 1995;Semyonov et al., 2006) while others find limited support for such a correlation (for example, Hjerm, 2007;Messing & Ságvári, 2018). However, many studies recognize that different groups of immigrants can trigger a variety of attitudes: these studies differentiate between immigrants in terms of their labour market potential or religious background when investigating the extent to which the local population perceives them as a threat to the local economy or culture. Low-skilled native workers, for example, are more likely to think that immigrants arriving from poor countries represent competition for them on the labour market (Scheve & Slaughter, 2001). In terms of whether economic or cultural grievances play a greater role in developing anti-immigration attitudes, research is multifaceted: findings regarding the labour-market competition hypothesis are highly contested (Chandler & Tsai, 2001;Citrin & Sides, 2008;Malhotra et al., 2013) and economic explanations are often understood as secondary (Lucassen & Lubbers, 2012), while cultural concerns about immigration at the individual level seem to have greater predictive power (Trifandafyllidou, 1998).
A recent theory that explains pro-and anti-migrant attitudes on an individual level perfectly fits the above argument. The concept of the perception of control was important inspiration for this paper as it links attitudes towards immigrants and openness to rightwing populism to the same root cause: a feeling of a lack of control. Harell et al. (2017) argue that a feeling of control is one of the most important explanatory factors of attitudes towards and the acceptance of migrants. The latter use the concept of locus of control, which refers to a set of beliefs about the causes of events (for example, losing one's job) or conditions (for example, being poor) to either internal or external sources (Lefcourt, 1991;Rothbaum et al., 1982). The argument is as follows. Citizens who believe they are personally responsible for what happens in their lives, and are thus capable of effecting change in their lives and the wider society they live in, are less hostile towards immigrants. These citizens are less likely to feel threatened by a changing social milieu. The feeling of being 'in control' of one's own economic or social situation, in contrast to feelings of insecurity and unpredictability, leads to less fear of the unknown and thus more open attitudes to immigration. Another level of the perception of control relates to the wider community: people who feel that the government is in control of the social and economic processes in a country, including migration-both its inflow, and migrant inclusion-are likely to feel less threatened by migration. The third level of control refers to migrants themselves: if individuals feel that migrants are agents of their own social inclusion and (have the potential to) become contributing members of society, they will be less likely to feel threatened by migration and thus reject migrants in general. In short, perceptions of control-as applied to citizens, the government, and immigrants-have an important effect on attitudes toward immigrants. A lack of a feeling of control brings about anxiety, which also feeds into preferences for populist political parties that promise easy and simplistic answers to highly complex, diverse social phenomena, such as migration (Wodak, 2015;Rico et al., 2017).
By using ESS data we are able to test these theories not only in a static, cross-sectional setting by comparing attitudes across countries, but we can create a dynamic image of attitudes. Following the logic of group competition theory and control theory, we propose that sudden changes in minority group size (such as following the 2015 migration crisis) or economic conditions (such as the 2008 financial crisis) are likely to change aitudes. Also, individuals who lack a feeling of being in control are likely to have stronger anti-immigrant aitudes than those who feel more existential and physical security.

Basic human values as drivers of interethnic group attitudes
Several studies have noted that individual human values are of overarching importance for explaining negative feelings towards immigrants (Davidov et al., 2008;Davidov & Meuleman, 2012;Grigoryan & Schwartz, 2020). These researchers have demonstrated that basic human values have great potential to explain attitudes-both the (dis)preferences of certain groups and of political ideologies. Values are broad, abstract principles that guide individuals' behaviour and opinions (such as honesty, freedom, equality, beauty, wisdom, etc.), thus it is meaningful to presume that they are correlated strongly with attitudes (including attitudes towards immigrants). Values and attitudes have the same roots: both are beliefs, but while values are beliefs in some end-state goals and the conduct preferable for achieving these goals, attitudes are an evaluative sum of many beliefs about a certain object or a group. Values occupy a more central position than attitudes (Hitlin & Piliavin, 2004) as they are abstract principles that guide attitudes towards concrete groups (Rokeach, 1968). Values stem from the primary agents of socialization: first and foremost, from family, but also from peers and school, and thus are very stable, while attitudes may change more eas-ily as they depend on numerous beliefs that may be altered in many ways (Ball-Rokeach & Loges, 1994). Davidov et al. (2008), in a study that tested the relationship between basic human values and attitudes towards immigrants, concluded that 'values of conservation and self-transcendence remained the strongest predictors for attitudes toward immigration' after controlling for the effects of basic demographic and economic variables. He found that basic human values inform attitudes, rather than the other way round. Based on this set of literature, we formulate our second research question.
e question we address in our paper is whether the strength of the relationship between human values and aitudes towards immigrants is the same across European countries. We also want to see whether, following 2015, the strength of the relationship between basic human values and aitudes towards immigrants remained unchanged, or if the arrival of masses of refugees strengthened the link between the value of security and of humanitarianism.

Political preferences, right-wing populism (RWP), and anti-immigrant attitudes
As indicated in the title, we hope to generate better understanding of how and under which circumstances anti-immigrant attitudes are linked to specific political attitudes, and, more specifically, to a preference/susceptibility to right-wing-populism (RWP) by comparing European countries in space and time. A vast literature shows a strong link between political preferences and attitudes towards immigrants. The focus of research is on the link between extreme right and anti-immigrant attitudes, and results about this are clear: farright party success depends on mobilizing grievances over immigration to a great extent (Bohman, 2011;Bohman & Hjerm, 2016;Ivarsflaten, 2008). Several authors argue, however, that anti-immigrant attitudes are a tool by which right-wing populist political forces increase their influence rather than a response to threats posed by immigration. Kende and Krekó (2020) offered an explanation for why right-wing populist parties became very successful in East-Central Europe, concluding that the efficient instrumentalization of immigrants as a threat by leading right-wing politicians supported anxieties embedded in long-term historical fears, such as fragile national sovereignty and vulnerable national identity. The influx of refugees in 2015 gave momentum to RWP parties which were able to capitalize on a centuries-long identity crisis and existing prejudice to stigmatize visible minority groups. It is actually the political exploitation of 'the immigrant threat' (despite the negligible size of the immigrant population) that led to the spectacular success of RWP parties in the region. In some countries, RWP used immigrants as a means of enhancing a sense of moral panic (Cohen, 2011), and, based on the fears of the population of unknown immigrants, consciously pressed the moral panic button to polarize the population to the extreme, helping them maintain power through extending the sense of a continuous state of emergency (Gerő & Sik, 2020). Based on the above stream of literature, we aim to address the following set of questions: Did the link between political preferences and aitudes towards immigrants become stronger following the 2015 mass inflow of refugees? Are anti-migrant aitudes more likely to be manifest in open rejection of immigrants in countries where right-wing populist parties are strong compared to countries in which these parties are weak or negligible?

Data and measurement tools
The analysis is built on individual-level data from the European Social Survey's (ESS) nine consecutive rounds from 2002 to 2019 in the first analytical chapter, and R7 (2014/15), R8 (2016/17), and R9 (2018/19) in the chapters that analyse the influence of the 2015 migration crisis. R7 (2014/15) offers data prior to the events in 2015, R8 (2016/17) represents the situation directly after the mass inflow of refugees to the EU, and R9 (2018/19) offers insight into a more consolidated situation-when several years had passed since the 2015 crisis.
The ESS is considered to be one of the most trustworthy cross-national datasets, and even though we are aware of the biases inherent to measuring attitudes in surveys comparatively , we believe that there is currently no better publicly available time-series dataset with which to measure and compare attitudes in Europe. An additional source of data is recent research by Roduijn and colleagues who compiled a database of political parties in Member States of the European Union, categorizing them as right-wing populist, left-wing populist, and moderate (Roduijn, 2019).
In this paper we use the most widely applied approach to understanding the construction of attitudes-namely, the ABC model, which differentiates between affective (A), behavioural (B), and cognitive (C) components of attitudes (van den Berg, 2006;Eagly & Chaiken, 1998). Based on ESS data, we have created two indicators as independent variables to reflect the behavioural and cognitive elements of attitudes.¹ (1) The behavioural component will be indicated by the Rejection Index (RI), which denotes the share of those who would reject the entry of any immigrants from poorer countries outside Europe without consideration.² We argue that by using only the extreme response to migration as a single indicator we are able to capture unequivocal attitudes.
(2) The cognitive component of attitudes incorporates both symbolic and material elements. The perception of the consequences of migration on material life is gauged by an item that measures the perceived impact of migration on the economy through the following question: 'Would you say it is generally bad or good for [country]'s economy that people come to live here from other countries?' Symbolic elements of attitudes are measured by the following question: 'Would you say that [country]'s cultural life is generally undermined or enriched by people coming to live here from other countries?' The third item focused on a more general evaluation of the effect of migration: 'Is your country made a worse or a beer place to live by people coming to live here from other countries?' The 0-10 scale responses given to the three questions were summed up and converted into a Perception Index (PI) that was then converted to a 0-100 scale in order to facilitate harmonization and comparability with the values of the Rejection Index. In general, smaller values indicate more negative perceptions of migration, while larger values depict the opposite.
In order to check the internal consistency of the indicator we ran a separate reliability analysis for the Perception Index. The three elements were highly reliable when calculated for the total of 126 subsets of the dataset that included data from Round 1 to Round 9 for  Figure SI1 in the Supplementary Material section). Another question to be clarified in the section on data and methods concerns the conceptualization of migrants. Migration is a very complex phenomenon involving a large variety of categories. Although the term 'migrant' is well defined in legal and policy contexts, it is still used in many senses, especially in non-scholarly public discourse. Questions which measure the cognitive element of attitudes in the survey refer to 'people coming to live here from other countries, ' thus 'migrants' are understood in the widest possible sense. Thus, we have no idea what image people had about immigrants when answering this question and it is very likely that a person in Berlin or London would have had a different image to one living in a Hungarian or Polish village, or in an Italian harbour town. The question measuring the behavioural element of attitudes is somewhat more specific: It inquires about 'people from the poorer countries outside Europe. ' Here again, though, respondents may have responded to the question with quite different conceptions in mind-with some visualising south-east-Asian IT workers, others Syrian/Afghani war refugees, and others desperate North Africans fleeing across the sea. This is definitely a weakness when measuring attitudes towards migrants with surveys and questionnaires, but the survey responses are nonetheless still the best sources for comparing attitudes across countries, regions, and time.

Description of attitudes and their anges
First, we provide an overview of attitudes based on the nine consecutive rounds of the ESS survey (2002-2019) concerning attitudes towards immigrants in Europe measured by the two indicators described earlier in the paper. Figure 1 summarizes the scores of the Perception (PI) and the Rejection (RI) indexes by country. The overall perception of migration in Europe is on average neutral (PI average =53). That is, people see as many advantages as disadvantages to worldwide mobility. Based on the sole values of the indicator, more countries have a positive (55+) perception of migration (thirteen countries altogether) than those which, on average, perceive the consequences of immigration negatively (five countries score below 45). As for the behavioural element of attitudes, the Rejection Index shows that 17 per cent of surveyed Europeans would unconditionally reject immigrants arriving from poorer countries outside Europe (RI average =17). Again, acceptance is a more common attitude than rejection: in fourteen countries out of twenty-seven, 10 per cent of the population or less would reject immigrants (from poorer countries outside Europe) settling in their countries, while in seven countries the share is greater than 20 per cent.
Another noteworthy finding is the distinctiveness of the cognitive and behavioural attitude components (Figure 1). Obviously, the two indexes are strongly correlated; nonetheless the relationship is non-linear and works quite differently in various countries (or country groups). While the cognitive element of attitudes (PI) fluctuates moderately between countries, the behavioural element (i.e. the rejection of migrants) does show significant outliers, such as Hungary, Bulgaria, the Czech Republic, and Slovakia. The relationship between RI and PI does not follow similar patterns in old and new EU Member States; ineecion. ea eopean jonal of ocie and poliic, 7(2): 100-127. compared to long-term democracies like Germany, Sweden, Norway, and Switzerland, in transition countries such as Hungary, the Czech Republic, Bulgaria, Estonia, and Lithuania, negative perceptions of the consequences of migration are more likely to be transformed into the unconditional rejection of immigrants. Perceptions of immigration (PI) in Italy (45) are more negative than in Lithuania, Estonia, or Latvia (52, 48, 55, respectively), but the upfront rejection of immigrants is less widespread in old EU Member States (such as Italy) than in the post-Soviet Baltic countries. Also, when comparing other post-communist countries with Austria and Italy, we see that although PI is slightly more negative in the former group of countries (Hungary, Bulgaria, and the Czech Republic) it does not explain the significantly stronger rejection in these countries compared to in Austria and Italy. We suspect that the strength of norms developed historically, but also influenced by ongoing political and public discourse is decisive in determining the degree to which the cognition of migration as having negative consequences for the host country is transformed into explicit rejection and exclusion. Rejection of immigrants is especially extreme in Hungary (57 per cent of Hungarians would reject the settling of migrants from poorer countries outside Europe). Such openly hostile attitudes in Hungary may be attributed to several intersecting factors: the small number of immigrants and consequent lack of personal experience and knowledge about them, together with the generally low levels of trust and social cohesion that characterise Hungarian society (Messing & Ságvári, 2018;. A society in such a state proved to be extremely fertile terrain for the manipulative, anti-migrant propaganda that the Hungarian government put into action in early 2015 and has kept operating since ineecion. ea eopean jonal of ocie and poliic, 7(2): 100-127. then in a de-pluralized media environment unparalleled in the EU (Bernáth & Messing, 2016;Barlai & Sik, 2017;Goździak & Márton, 2018;Bajomi-Lázár, 2019;Kende & Krekó, 2020).
To investigate whether attitudes are sensitive to macro-level societal and/or economic changes, we need to look into longer-term trends in attitudes. Data show surprising stability in Europe ( Looking beyond the macro data that characterizes all countries, certain important cross-country differences become visible, but actual exposure to the inflow of asylum seekers in 2015-16 has limited explanatory power. Table 1 illustrates changes in the perception and rejections indexes and their statistical significance in 20 countries participating in both R7 and R9 of the ESS. We used these two rounds because we can compare pre-2015 and the consolidated post-2015 moments. When comparing data in 2014/15 (R7) and in 2018/19 (R9) (using independent samples t tests) we can distinguish between three types of countries: (1) those where both the perception and rejection indexes have changed significantly; (2) countries in which there was no significant change; and, (3) countries where the change in the two indicators is asynchronous.
In most-eight out of twenty-countries a statistically significant and robust increase in pro-immigrant attitudes was recorded concerning both indicators (Belgium, Switzerland, Spain, Finland, France, UK, Ireland and Portugal). Attitudes became more pro-immigrant irrespective of whether we look at the cognitive or the behavioural element of attitudes. Out of the 20 countries, there was only one-Hungary-in which both indicators measuring attitudes towards immigrants deteriorated significantly in statistical terms between 2014 and 2018/19, reflecting the pre-and post-2015 migration crisis situation. There was no statistically significant change in any of the indicators in Austria and in Germany. In the remaining nine countries, the change in the two indicators measuring attitudes was incoherent. Table 1. Change in PI and RI between ESS R7 and R9 with results of independent sample T tests

How are attitudes to migration linked to basic human values?
We use the theoretical and methodological framework developed by S. Schwartz (1994) to map connections between basic human values and attitudes to migration. Applying a linear regression model to the Perception Index (with 21 individual items measuring 10 basic human values) to data from 15 countries that participated in all ESS survey rounds highlights that five items belonging to two basic value categories crystalize at both ends of the chart: security and humanitarianism⁴ ( Figure 3). Our data support findings of previous research; namely, the importance of humanitarian and security-related values in developing attitudes towards immigrants (Davidov, 2008). Agreement with the first two statements-the importance of secure surroundings, and the importance of a strong government that ensures safety-signals that a strong value is attributed to stability and externally provided physical safety. The two statements at the bottom of the list signal strong attachment to humanitarian values, such as understand-⁴ The term 'transcendence' is used in the original theoretical model, but for this paper we have changed it to 'humanitarianism' because we think that this relates better to the topic of migration. ing, and respecting and treating one another equally. These two types of values are the strongest predictors of attitudes towards immigrants, but in a contrasting direction: those who attribute great significance to security tend to be more negative towards immigrants, while those who value equality and respect for other people are the least fearful of migrants. Other basic values measured by the remaining items (such as achievement, hedonism, power, self-direction, and stimulation) seem to have only a weak or non-existent relationship with how people think about migration, thus our analysis will focus on these two sets of values: universalism (humanitarian values) and security. It might occur that a single point of measurement does not represent sufficiently the link between values and attitudes. Therefore, we analysed the Perception Index using these two types of basic human values for 15 countries participating in all rounds of the European Social Survey from 2002 to 2017, and found that-depending on country and fieldwork rounds-on average 8 per cent of the variance of PI is explained. However, the countryspecific values indicate major differences across Europe in the role of basic human values in developing attitudes towards immigrants. The adjusted R² statistics are well below average in Hungary, Portugal, Slovenia, and Poland ( Figure 4A) meaning that in these countries human values play a less vital role in developing attitudes towards immigrants than in other countries. The country-specific regression analysis applied to all ESS rounds separately ( Figure  4B) shows that, generally, the importance of the security value stayed relatively stable across time, and most importantly did not-as one might have presumed-increase following the arrival of masses of refugees in 2015. Contrarily, the role of the value attached to humanitarianism on attitudes towards immigrants (PI), after staying fairly stable throughout the 12 years between Round 1 in 2002 and Round 6 in 2012, had increased by 2014/15 and remained high until 2016, thereby becoming a stronger predictor of attitudes (compared to security). Based on these data we may conclude that the refugee crisis activated the link between humanitarian values and attitudes towards immigrants (more specifically, its cognitive element measured by the perception index), and this might have contributed to the stability of attitudes and even lessened the rejection of immigrants in some countries.  The chart shows a very clear pattern: those who self-identify as left-wing have a generally positive attitude towards immigrants, those in the centre of the political scale are relatively neutral, and those on the right have a generally negative attitude towards immigrants. This is no surprise. However, the gradient of the curve is notable: those selfidentifying with the left were equally positive about migration (in 2016/17 and 2018/19), irrespective of how left-oriented they feel; i.e. people who position themselves on the extreme of the scale have very similar attitudes to those who think of themselves as moderately left-wing. At the right-wing end of the scale, we see a very different picture: the gradient of the curve is steep, meaning that political right-wing extremism correlates strongly with extreme anti-immigrant attitudes.

How are attitudes about migration linked to political preferences?
Our initial question for the research described in the paper concerns the impact of the 2015 migration crisis on the strength of the relationship between political orientation and attitudes towards immigrants. We ask whether political orientation predicts attitudes towards immigrants in the same way after 2015 than before it. Figure 5 shows that there is one important change from 2014/15 to 2016/17 and 2018/19; namely, that those who selfidentify as strongly left-oriented became more tolerant towards immigrants after the 2015 refugee crisis.
The data also provides evidence for the hypothesis that in different countries the relationship between political orientation and attitudes towards immigrants may play out differently, and also, that the latter became slightly stronger after 2015. Looking at individual countries we see significant diversity in this respect (see Figure SI2 in Supplementary Information) Although the overall averages show a very clear pattern of correlation between political preferences and attitudes towards immigrants, on the country level these correlations are diverse and much less clear. In Austria, Germany, Belgium, Switzerland, Spain, France, and the UK the relationship between political orientation and attitudes towards immigrants is strong. There are however a few outliers: in the Czech Republic, Esto-nia, and Lithuania the relationship is reversed: left-oriented respondents express stronger anti-immigrant attitudes than those in the centre or the right of the political spectrum.
There are a few countries where the relationship is extremely weak or does not exist at all: Finland, Hungary, Ireland, and Poland.
As to how migration processes in 2015 influenced this relationship we see similar diversity. In Finland, France, Netherlands, and Slovenia it stayed unchanged. But in some important cases we see a polarizing effect: those on the right became more anti-immigrant and those who felt closer to the left had become more accepting of immigrants by 2018/9 compared to 2014/5 and 2016/7. Such countries include Austria, Germany, the UK, and Sweden most distinctively. In several East European countries-Hungary and Poland, but also Estonia and Czechia to some extent-a similar phenomenon is visible: a strong polarization of attitudes along political lines between 2014/5 and 2016/7 and a bounce back to rather homogeneous(ly negative) attitudes in 2018/9.

Overview of factors that shape attitudes to migration: A multilevel model
In order to understand the composite effects of various factors that define one's overall perception of migration rather than looking at these factors individually, we developed a complex model that includes four groups of explanatory variables and takes into account the effect of country and time (pre-and post-2015) (Fairbrother, 2014). The data behind the model are limited to the last three rounds of the ESS (R7 to R9). Since respondents in the ESS are nested within higher level units (namely, countries), we used a hierarchical linear model for the analysis. As already discussed earlier in this paper, the Perception Index varies considerably across countries and time. In terms of numbers, the variability of the PI that is directly attributable to countries (interclass correlation, ICC) is .0.11 per cent for the three rounds overall, which is assessed as rather high (Model 0). This implies that the context of a given country (its historical, cultural, economic, political and media trajectories) immensely influences how people perceive migration. Based on our previous research (Messing & Ságvári, 2018; and the works of other authors (and of course considering the limits of the ESS data) we included the following groups of explanatory variables that reflect our theoretical framework into the model:⁶ -Basic demography: gender, age, education, foreign-born status -Locus of control: subjective financial status, institutional trust, interpersonal trust, feeling of lack of physical safety -Human values: humanitarianism-related values, security-related values -Political orientation: placement on left-right scale The model was built in stages by including groups of variables step by step (Table 1). In the simplest model (Model 1) basic individual demographic characteristics were entered, and the results showed that all variables significantly predict attitudes towards immigrants (more specifically, is cognitive element). Without taking into consideration the effect of country and time, female respondents generally tend to have somewhat lower PI values (however, the estimate is very small), and with an increase in age there is also a small decrease in PI. The total number of years spent in education seems to have a strong influence on attitudes towards immigrants. In modelling terms, each extra year in school increases ineecion. ea eopean jonal of ocie and poliic, 7(2): 100-127.
PI by 1.2 point, on average. Not surprisingly, not being born in the current country of residence or having parent(s) born in a different country also have a strong positive effect on perception (+5.4) Models 3 to Model 5 include explanatory variables that are connected to our theoretical frameworks. The first group of variables in Model 3 includes variables related to people's general feeling of control. The results of the model suggest that both institutional and interpersonal trust indexes (measured on a 0 to 10 scale) are strongly related to our dependent variable. In general, higher levels of trust result in more positive perceptions about migration. The strength of the two trust indexes in altering perceptions is almost equal, indicating their dual effect in the model. On the other hand, a deprived financial status and feeling of a lack of physical safety are factors that lead to more negative perceptionproviding evidence of the importance of the role of control in the lives of individuals.
Security and humanitarianism-two of the human values (measured on a 1 to 6 scale) that proved to be significantly associated with attitudes towards immigrants ( Figure 3)were also included in the model (Davidov et al., 2008).⁷ Both have very strong effect on the PI, but in the opposite direction (Model 4). Those who place strong value on security are more negative about migration and those who think that humanitarian values are important have generally more positive attitudes. This association remains strong, even if we discount the effect of demographic characteristics and proxies of the feeling of control.
Finally, we also included political orientation measured on a 0 (left) to 10 (right) scale (Model 5). The coefficient estimate of -1.2 in the model indicates that people on the right side of the political spectrum have lower PI values in general, again discounting the effect of basic demographic characteristics-of the proxies of feeling of control and identification with humanitarian and security-centred human values.
The last model (Model 6) includes information on time and the interaction of time and non-demographic variables. The initial time point in our model is ESS Round 7, for which the fieldwork took place during 2014 and 2015. We have argued that this period captured the last 'moment' before the 2015 migration crisis, so the next two rounds refer to the 'postcrisis' period: 2016/7 (R8) indicates the short-term while 2018/9 (R9) the more consolidated impact of the refugee crisis on public attitudes towards migration and immigrants. This change is reflected by the 'essround' variable in our model. Overall, this has a small positive effect, implying that at macro-level the perception of migration became more positive after 2015. However, the effect of the interaction terms is either non-significant or very small, supporting the idea that deeply rooted attitudes do not tend to change within a short period, while nor do relationships with the perception of migration show considerable stability if all potential intersecting factors (demography, human values, feeling of being in control, and political orientation) are discounted for. However, there are important country-level differences-as we have shown in the previous section-which are the consequence of a more complex web of interrelated factors stemming from politics, media discourse, historical and cultural traditions, and more. ⁷ See Section 2.3 for an explanation of the use of these two specific value dimensions.

Discussion: How are attitudes towards migration linked to openness to populism?
In the introduction to this paper, we raised the idea that anti-immigrant attitudes and rising populism may be strongly connected in Europe, and are transmitted through changes in the importance attributed to values of humanitarianism and security, as well as the feeling of control. A socio-psychological explanation of this link postulates that when people perceive an increase in disorder-something that they feel they (or their governments) are not in control of-they feel anxiety (Harell et al., 2017). Such anxiety leads to a need for security. Individuals strive to bring safety and stability back into their lives and are more likely to vote for right-wing populist political parties, which promise strength and order. Fear and frustration linked to migration have become symbols for this general feeling of uncertainty, and populist political forces have always fallen back on a fear of immigrants and the consequences of immigration. Analysis of populist politicians' speeches in Europe and the USA shows that migration is high on their agenda. The victory of Brexit, of Donald Trump, Viktor Orbán, Matteo Salvini, the Freedom Party in Austria, and the Law and Justice Party in Poland can all be attributed to this increased desire for stability and order, and all of these movements had anti-immigration at the top of their agendas (Yilmaz, 2012;Hooghe & Dassoneville, 2018;Wirz et al., 2018).
Using recent research on populism in Europe⁸ to categorize populist parties in nine European countries, we tried to establish the degree to which anti-immigrant feelings are linked to support for political populism.⁹ For this analysis, we used data from the eight round of the ESS, harmonized with party categorization in time. This analysis of the connection between voting for a populist party and attitudes towards immigrants resulted in a very obvious conclusion: those with negative attitudes are much more likely to vote for right-wing populist parties. Figure 6A illustrates the association between party preferences and attitudes towards immigrants (perception and Rejection indexes), while 6B highlights the differences between both cognitive (PI) and behavioural (RI) elements of attitudes between supporters of right-wing populist parties and others.
Data on attitudes of party supporters ( Figure 6B) show very clearly that supporters of right-wing populist parties (such as AfD in Germany, the Front National in France,¹⁰ the League in Italy, and the Swedish Democrats) have significantly more negative and exclusionary attitudes towards immigrants than supporters of any other parties on a national level. To put it plainly, these parties gather and feed that part of the population which is very negative towards immigrants and migration in general. Right-wing populist parties seem to provide terrain on which to openly express the rage fuelled by uncertainty and to ⁸ https://www.theguardian.com/world/2018/nov/20/measuring-populism-how-guardian-charted-risemethodology ⁹ The following definition was used for the concept of populist parties: 'Parties that endorse the set of ideas that society is ultimately separated into two homogeneous and antagonistic groups, "the pure people" versus "the corrupt elite, " and which argue that politics should be an expression of the volonté générale, or general will, of the people' (Mudde, 2014). ¹⁰ Rassemblement National (National Rally) since June 2018. In almost all countries, one or two such parties exist: the difference lies rather in how powerful they are. They are tiny in Sweden and Norway, small but significant in Germany and France, large in Italy and Austria, and even form a supermajority government in Hungary. While Figure 6 demonstrates that in most countries right-wing populist parties attract the population with extreme anti-immigrant attitudes, in Hungary and Italy supporters of centrist or even left-wing political parties may hold very negative attitudes towards immigrants in European comparison. There is another notable lesson from Figure 6: although supporters of right-wing political parties perceive consequences of migration (PI) quite similarly (very negatively) across countries, these negative perceptions translate into very different levels of rejection of immigrants: rejecting any kind of migration is most explicit in Hungary, while supporters of right-wing populist parties (FPÖ, FN, LN) in other countries, even if they may perceive migration more negatively, are more moderate in their rejection of immigrants. This data shows the degree to which dominant norms, defined by historical-political traditions and current mainstream politics, matter in terms of transforming aversion into the extreme rejection of immigrants. We would argue that the political power such parties wield, whether in government or in opposition, plays a critical role in determining the degree to which anti-migrant narratives are allowed to become the norm within society.

Summary and conclusion
Looking at longer-term trends in attitudes toward immigrants we found notable stability over a period of 16 years. The overall perception of migration, as well as the share of those supporting the explicit rejection of immigrants from poorer countries outside Europe, has not changed radically. Attitudes may have changed within shorter periods of time in certain countries, but in the longer term, and in general, they have remained rather stable across the continent. The first research question-whether attitudes towards immigrants have become more negative following the 2008 economic crisis and the 2015 refugee crisis-was thus answered in the negative. Contrary to expectations based on group conflict theory, the 2015 refugee crisis actually brought about slightly more openness towards immigrants: the share of those who would refuse the settlement of immigrants-on average, from the population of 20 countries-decreased from 15 to 10 per cent between 2014 and 2016. The multilevel model that discounts the effect of changes in demographic characteristics, changes in basic human values, or political attitudes, supports the above statement. However, intergroup conflict theory cannot be refuted either, as short-term, country-specific changes in attitudes occurred in spaces and times marked by uncertainty brought about by large-scale political changes, such as the enlargement of the EU in 2004. We claim therefore that group conflict theory on its own is not sufficient to explain all the differences in attitudes across time and location. A noteworthy finding of the analysis of the impact of time and geographic trends on attitudes is that, despite similar perceptions about the consequences of migration, people living in different countries reject immigrants in radically different levels. Data suggest that the level to which negative perceptions of migration result in (unconditional) rejection is a function of the general norms characteristic of a country, and is brought about by political and media discourse, historical experiences, and dominant social values.
The second set of research questions about the link between basic human values found that, contrary to our expectations, the strength of the link between values and attitudes towards immigrants is not homogeneous across countries or time. We found that the strength by which values of humanitarianism may explain attitudes towards immigrants slightly increased in 2015 and afterwards. The strength of this relationship also varies across countries: in countries with a communist authoritarian heritage this link is weaker, while in long-term welfare states, such as the Nordic countries and Germany, values predict attitudes much better. This finding suggests that certain conditions (i.e. the historical experiences of a country, or sudden changes such as an inflow of refugees) may activate or enhance the link between basic human values (more specifically, the value attributed to humanitarianism) and attitudes towards immigrants. In our conclusion we emphasize that security, a basic human value, is one of people's basic needs and that humanitarian values and related tolerance may come to the fore in an environment where people feel secure; most countries in Central and East European countries do not seem to have reached that stage.
The third part of our analysis looked into the relationship between attitudes towards immigrants and political preferences. There is nothing new about the finding that individuals' subjective position on the political scale correlates strongly with attitudes towards immigrants. This correlation, however, shows very different patterns across Europe: in many countries there is a linear link, while in a significant number of countries there is hardly any association between political self-identification on the left-right scale and anti/pro-migrant attitudes. The multi-level model also showed that the strength of the association between political preferences and attitudes towards immigrants has not become significantly stronger since the 2015 refugee crisis.
Based on theories and earlier literature, we presumed that there is strong correlation between anti-immigrant attitudes and a preference for right-wing populism. Our analysis confirmed this correlation and found that anti-immigrant attitudes are more likely to manifest in open rejection of immigrants in countries where populist parties are strong or have even taken power. In all countries, right-wing populist parties gather and feed that part of the population which is the most negative towards immigrants and migration in general. However, a novel finding of the analysis is that it demonstrates cross-country variety concerning the level to which negative cognition of the consequences of migration are transformed into negative behavioural expectations among supporters of rightwing populist parties. Although the perception indexes of this subgroup are quite similar Europe-wide, the strength of rejection is very different: rejecting any kind of migration is most explicit in Hungary, while in other countries even the more negative perception of migration by supporters of right-wing populist parties (FPÖ, FN, LN, etc.) results in a smaller share of those rejecting immigrants. This data shows the degree to which dominant norms, political and historical traditions, and mainstream politics matter in terms of transforming aversion into the extreme rejection of immigrants. To answer the initial question of the paper, it may be assumed that anti-immigrant attitudes are indeed the Holy Grail of right-wing populism in Europe, and are activated by the importance attached to basic human values of security and humanitarianism. In countries where people feel insecure and are striving for stability (and where there is little presence of immigrants), right-wing populists may gain momentum by using anti-immigrant narratives. Figure SI1. Reliability analysis of the Perception Index (PI) ineecion. ea eopean jonal of ocie and poliic, 7(2): 100-127.