News consumption and immigration attitudes: a mixed methods approach

ABSTRACT
 Existing research has shown that the media can influence public attitudes to immigration. The adoption of a mixed-methods approach to audiences of immigration news, combined with a comparative design and a focus on Eastern Europe – a region scoring lowest in the world in terms of migrant acceptance – can bring significant advances to knowledge in this area, leading to a more rounded understanding of how media come to shape immigration attitudes. To demonstrate this, we draw on a comparative, mixed-methods data set comprising representative population surveys (N = 4,092), an expert survey (N = 60), and qualitative interviews (N = 120) conducted in four Eastern European countries. In contrast to existing research on Western Europe, we found significant variation in the links between attitudes to immigration and use of Public Service Media (PSM), with PSM consumption linked with more negative attitudes to immigration in some countries, and with more positive attitudes in others. Second, our results confirm that different attitudes to immigration are embedded in different qualitative understandings of immigration: while participants with more positive attitudes often adopted a more inclusive understanding of immigration, those with more negative attitudes adopted a narrower understanding. Third, we demonstrated the importance of family and acquaintances as trusted sources of information.

Immigration has become a key axis of political contestation and played an important role in the rise of populist movements and parties in several countries. Perceived threats posed by immigration have been a salient issue in the discourse of right-wing populist groups in the USA, Europe, and Australia (Hogan and Haltinner 2015) and immigration also played an important role in influencing the Brexit vote in the UK and the election of Trump in the US (Goodwin and Milazzo 2017;Joppke 2020). The media have been shown to play an important contributing role in shaping attitudes to immigration, and in influencing populist voting intentions (Meltzer et al. 2021;Walgrave and De Swert 2004;Štětka, Mihelj, and Tóth 2020). The growing recognition of the role of the media has been reflected in the rapid growth of research on the issue, with more than a five-fold increase in the number of studies on media and immigration between 2009-10 and 2017-18 (Eberl et al. 2018, 208).
A large amount of existing research focuses on media coverage, often showing that immigration is negatively framed (e.g. Heidenreich et al. 2020;Lawlor and Tolley 2017) and revealing consistent differences in discursive patterns associated with negative vs. positive coverage Baker and McEnery 2005). Research on the role of media in shaping public attitudes has grown as well, showing significant correlations between public attitudes and news consumption (e.g. Štětka, Mihelj, and Tóth 2020;Jacobs, Meeusen, and d'Haenens 2016;Schemer 2012;Van Klingeren et al. 2015). While providing important insights, existing work on media and immigration attitudes also has some limitations. First, it is primarily focused on Western Europe and Northern America, and rarely adopts a comparative design (for exceptions see Esser et al. 2019;Meltzer et al. 2021;Van Klingeren et al. 2015). This limits our capacity to understand how the relationship between attitudes to immigration and news consumption is shaped by differences in political and media environments. This is particularly relevant in relation to the impact of Public Service Media (PSM) which are typically associated with more positive, or at least neutral, immigration coverage, and with more positive or neutral audience attitudes on the issue (Jacobs, Meeusen, and d'Haenens 2016;Vergeer and Scheepers 1998;Schulz 2019). However, these findings are largely based on data from countries where PSM enjoy relative independence, and where immigration attitudes are relatively positive. This is likely to be different in contexts where PSM is under government control and where the governing party uses anti-immigration rhetoric to galvanise public support, as is the case in some of the countries examined in this paper.
Second, existing research on immigration and media audiences is based almost exclusively on quantitative methods, which means that it has limited capacity to understand how quantitative differences in attitudes and news consumption patterns relate to qualitative differences in audience understandings of immigration, and to the way audiences use media to articulate their attitudes. Research on media coverage of immigrationwhere qualitative methods are more commonhas identified interesting regularities in relation to both qualitative and quantitative patterns of coverage (e.g. Baker et al. 2008). However, it remains to be seen whether similar patterns can also be found among audiences. Furthermore, quantitative methods are usually employed in the context of a media effects framework, typically in combination with agenda setting, framing, and social identity theories (e.g. Eberl et al. 2018;Meltzer et al. 2021;Van Klingeren et al. 2015). Such a framework leaves little room for examining the active engagement of audiences with media coverage, and therefore for developing a more rounded understanding of how media shape immigration attitudes.
To fill these gaps, this study uses a mixed methods approach, combining population surveys, an expert survey, and qualitative interviews conducted between December 2019 and February 2020 in four countries: the Czech Republic, Hungary, Poland, and Serbia. The mixed-methods approach enables us to build a multifaceted understanding of the role of the media in shaping attitudes, which combines an appreciation of broad patterns of links between media use and attitudes with insights into how these links relate to people's understandings of immigration and use of information sources. The four countries we focus on are of particular interest for the study of media and immigration, because they belong to a part of Europe that scores lowest in the world in terms of migrant acceptance (Heath and Richards 2019), and experienced a marked decline in the quality of democracya development closely linked with the rise of ethnopopulism (Vachudova 2020). While sharing many similarities, the countries also differ in their media systems, and in the extent of politicisation of immigration, which enables us to examine the impact of systemic factors on the relationship between media and attitudes.
We focus on three questions. First, we ask whether different attitudes are associated with qualitatively different understandings of immigration and consider whether these patterns reflect the findings of existing research on media coverage of migration. Second, we ask whether the link between PSM consumption and immigration attitudes differs depending on political and media context. Third, we consider whether participants with different immigration attitudes also use information sources differently in articulating their attitudes.

The role of the media in shaping attitudes to immigration
Existing research on media coverage of immigration revealed consistent discursive and semantic patterns, from the pervasive use of metaphors associated with natural catastrophes and military conflicts (e.g. El Refaie 2001; Gabrielatos and Baker 2008), to the common conflation of terms 'immigrants', 'asylum seekers', and 'refugees' (Baker and McEnery 2005), and discursive differences between tabloids and broadsheets (e.g. Baker et al. 2008). Discursive and framing differences in immigration coverage have also been linked to different perceptions of immigration and different opinions on immigration policy (Blinder and Jeannet 2018;Merolla, Ramakrishnan, and Haynes 2013;Rucker, Murphy, and Quintanilla 2019). For instance, one study found that framing immigrants as 'highly skilled' made it less likely for immigrants to be perceived as asylum seekers or 'illegal' immigrants, and reduced the overestimation of the size of the 'illegal' foreign born population (Blinder and Jeannet 2018), while another showed the use of more negative labels (e.g. 'illegal aliens') prompted more prejudice and greater support for punitive policies than use of more neutral labels (e.g. 'noncitizens') (Rucker, Murphy, and Quintanilla 2019). What aspect of immigration is being framed can matter as well, with one study pointing out that avoiding negative labels in framing immigration policies is considerably more consequential than avoiding such frames in relation to immigrants themselves (Merolla, Ramakrishnan, and Haynes 2013). Of most direct relevance to our research, there is also survey-based evidence indicating significant differences in public understandings of 'immigration', and links between these understandings and attitudes. Namely, individual variations in 'imagined immigration'for example, the extent to which immigration is equated to asylum seekers and permanent arrivalshave been shown to be strongly associated with individual preferences for reduced immigration (Blinder 2015). Yet, the lack of qualitative research on audience responses to immigration news means that we do not know whether the qualitative patterns found in media coverage are reflected in the way audiences themselves understand immigration. Our first research question therefore is: Are quantitatively different attitudes to immigration associated with qualitatively different understandings of immigration, mirroring similar patterns found in media coverage?
In contrast to research focused on media coverage of immigration, which utilises both quantitative and qualitative approaches, research on the link between media and attitudes to immigration is based almost exclusively on quantitative methods, such as cross-sectional surveys, survey experiments, or survey and media content or digital tracking data (e.g. Blinder and Jeannet 2018;Boomgaarden and Vliegenthart 2009;Jacobs, Meeusen, and d'Haenens 2016;Schemer 2012;Schlueter, Masso, and Davidov 2020;Štětka, Mihelj, and Tóth 2020;Van Klingeren et al. 2015). Most of this research operates within the media effects framework, and examines the effects of media coverage in general, focusing primarily on the impact of salience (e.g. Van Klingeren et al. 2015;Walgrave and De Swert 2004) and the impact of valence (e.g. Schemer 2012; Schlueter, Masso, and Davidov 2020) with some studies measuring both (e.g. Boomgaarden and Vliegenthart 2009;Meltzer et al. 2021).
Research focusing on salience tends to assume that the simple increase in media coverage of migration, and hence an increase in exposure to such news, may lead to negative attitudes. The results of this line of investigation have been mixed. Early, single-country studies did find evidence of links between salience of immigration coverage and support for right-wing politics (Walgrave and De Swert 2004), as well as negative perceptions of immigration (Boomgaarden and Vliegenthart 2009). However, recent comparative studies concluded that the strength of the link differs across countries (Meltzer et al. 2021;Van Klingeren et al. 2015). Research on the impact of valence, which acknowledges the possibility of differential impact of positive and negative coverage, has produced more consistent findings. Several studies suggested clear links between positive coverage, lower levels of public concern about immigration, and less stereotypical attitudes to migrants; and between negative coverage, higher levels of concern, and more stereotypical attitudes (e.g. Boomgaarden and Vliegenthart 2009;Meltzer et al. 2021;Schemer 2012).
While research discussed so far studies the impact of media coverage in general, a smaller body of literature examines the impact of specific media types. Studies of media coverage have often found significant differences, typically between tabloid/commercial and broadsheet/public service media (e.g. Baker et al. 2008) or between media associated with different political leanings (Merolla, Ramakrishnan, and Haynes 2013). Given these patterns, one would expect that different types of media may also have differential effects on attitudes. Existing research most often focuses on the impact of tabloid/ commercial and broadsheets/public service media. The results for the impact of commercial media appear more consistent; studies conducted in different countries found that exposure to commercial television and tabloid press is positively linked with negative attitudes to immigrants (Jacobs, Meeusen, and d'Haenens 2016;Vergeer and Scheepers 1998;Schulz 2019). The evidence for the impact of PSM is more mixed, with research suggesting that the link between exposure to PSM and more positive attitudes to immigration varies with context (Beyer and Matthes 2015;Štětka, Mihelj, and Tóth 2020;Strabac, Thorbjørnsrud, and Jenssen 2012). The reasons for such variation are yet to be systematically explored, although existing research suggests that the nature of the media and political system is key (Beyer and Matthes 2015). Importantly, existing research is limited to countries where PSM enjoy relative independence. This raises the question of how the impact might differ in countries where PSM are subjected to government control, and where the governing party has been relying on anti-immigrant rhetoric as a means of harnessing popular support, as is the case in some of the countries covered here. Our second research question therefore is: Is the impact of PSM consumption on immigration attitudes dependent on the extent of PSM independence?
In countries where PSM enjoy independence we expect PSM consumption to be associated with more positive attitudes to immigration. Conversely, in countries where PSM are under government control, we expect PSM consumption to be associated with more negative attitudes. Additionally, we expect the link between PSM consumption and negative attitudes to be particularly strong in countries where anti-immigrant attitudes have been integral to government discourse.
In contrast to the well-developed body of literature utilising quantitative methods, qualitative research examining audience responses to immigration coverage is scarce, and mixed methods studies are virtually non-existent. The rare studies that employ qualitative methods often focus on very specific groups such a migrants or ethnic minorities (e.g. Ong 2009;Lünenborg and Fürsich 2014) and have limited capacity to provide insight into trends among the general population. Nonetheless, a couple of exceptions are worth noting. An interview-based study of responses to media coverage of asylum seekers among Australian audiences (Haw 2020) showed that participants were critical of media coverage and argued the sensationalist approach made them distrustful of such media stories. In a qualitative study examining user comments on Greek news websites, Milioni and Vadratsikas (2016) likewise found that users often challenged news media; however, this typically involved challenging positivelyframed news by drawing on negative attitudes to migration. Both studies point to the importance of switching attention from media effects to audience uses and practices or, to paraphrase an argument familiar from early audience research (Katz 1959)from the question 'What do media do to people's attitudes to immigration?' to the question 'What do people do with the media to articulate and justify their attitudes to immigration?' However, while early audience research focused on establishing the socio-psychological motivations of media use (e.g. Katz, Blumler, and Gurevitch 1973), our focus is on audience practices (cf. Bird 2010)that is, on the 'how', rather than the 'why'and on how these practices relate to immigration attitudes. Our final research question therefore is: How do people use information about immigration to articulate their attitudes to immigration, and does this differ depending on their attitudes?

Politicisation of immigration, public attitudes, and the media in Eastern Europe
There are two key reasons that make a focus on media and immigration in Eastern Europe particularly attractive. First, the region experienced a marked decline in the quality of democracy in recent yearsa development closely linked with the rise of illiberalism ad ethnopopulism (Vachudova 2020). These developments have been paralleled by a marked reliance on anti-immigration rhetoric among political elites and governing parties in several countries, exemplified particularly by Hungary's Prime Minister Viktor Orbán (Bandelj and Gibson 2020) and the Polish right-wing nationalist Law and Justice (PiS) party (Krzyzanowski 2018). Similar tendencies can be observed in other parts of the region. For instance, the Slovenian Democratic Party (SDS), led by the Prime Minister Janez Janša, has been explicitly emulating Viktor Orbán's hardline anti-immigration policies in recent years (Zielonka and Rupnik 2020). In Slovakia, the then-Prime Minister Robert Fico and his party SMER adopted an openly hostility rhetoric towards migrants after 2015 (Harris 2019) and in the Czech Republic, anti-immigration narratives have been increasingly used by the political mainstream (Wondreys 2020). Second, the politicisation of immigration has gone hand in hand with the worsening of public attitudes to immigration after 2015 (Bandelj and Gibson 2020). As a result, Eastern European countries score among the lowest in the world in terms of migrant acceptance (Esipova, Ray, and Pugliese 2020; see also Heath and Richards 2019, for comparison between Central Eastern and Western Europe).
At the same time, it is important to note that Eastern Europe is not a complete monolith when it comes to the dynamics of elite rhetoric and public attitudes to immigration. Immigration has not been politicised to an equal extent across all countries in the region. In particular, the democratic deconsolidation in South-east Europe (including Serbia) was accompanied by forms of ethnopopulism that focused more on internal ethnic divisions, economic arguments, or international relations, rather than immigration (Bieber 2020). Levels of anti-immigration attitudes among the public vary considerably as well. For instance, recent polls suggest that in Poland, acceptance of migrants has increased in recent years (Esipova, Ray, and Pugliese 2020). This is also supported by recent data from Globsec (2020) which demonstrate that the percentage of people who believe that migrants are 'threatening their identity and values' is the lowest in Poland (41%), and the highest in the Czech Republic and Slovakia (72%), with Hungary positioned around the region's average (52%). Finally, a study based on Eurobarometer data from 2015 and 2017 suggests that in Serbia, the proportion of citizens considering immigration as one of the most important issues facing their country was considerably lower than in Poland and Czech Republic (Esser et al. 2019, 128)a result consistent with lower levels of politicisation of immigration in this country.
Despite these interesting developments, studies investigating the role of the media in shaping anti-immigration attitudes in this part of the world remain relatively rare (Eberl et al. 2018). Although the volume of work focusing on media and immigration in Easter Europe has grown in recent years, most studies are focused on media coverage rather than the link between media and attitudes, and limited to single countries (e.g. Jelínková 2019; Krzyzanowski 2018; Merkovity and Stumpf 2020; Tóth 2020; Vincze, Meza, and Balaban 2020), with a handful adopting a comparative perspective (Brosius, van Elsas, and de Vreese 2019;Radovanović Felberg and Šarić 2017;Georgiou and Zaborowski 2017). These studies suggest that coverage in South-East Europe, including Croatia and Serbia, tends to be more positive, especially in relation to the 2015 refugee crisis (Radovanović Felberg and Šarić 2017; Georgiou and Zaborowski 2017)a result consistent with intra-regional differences in public attitudes and elite use of anti-immigration rhetoric. Some of these studies also found notable differences between tabloid and broadsheet coverage (Jelínková 2019;Vincze, Meza, and Balaban 2020) or argued that the position and level of independence of PSM may be a factor in cross-country differences (Georgiou and Zaborowski 2017) Among the few studies investigating the impact of media coverage on audiences is Wenzel and Żerkowska-Balas's (2019) study of Polish voters, which found that consuming news outside of one's regular media outlets could influence political opinion on immigration. A study combining survey data and digital data tracking, conducted in the Czech Republic (Štětka, Mihelj, and Tóth 2020), found that consuming more public service television resulted in less negative attitudes towards immigration, while watching the two biggest commercial TV networks in the country led to a hardening of anti-immigration attitudes. Finally, a comparative study covering Serbia, Poland, and the Czech Republic, among other European countries, found that the proportion of populist messages in immigration coverage in Serbia was considerably lower than in the Czech Republic and Poland (Esser et al. 2019).

Country selection
The selection of the four countries covered herenamely, the Czech Republic, Hungary, Poland, and Serbiawas driven by the aim to include countries that differ along three main dimensions: (a) the relative role of anti-immigration rhetoric in the recent rise of populism and illiberalism, (b) public attitudes to immigration, and (c) the extent of media and especially PSM independence. As evident from the previous section, populist leaders and parties in Poland, Hungary, and the Czech Republic have all relied on anti-immigration rhetoric to galvanise public support, while such rhetoric was much less prominent in Serbia (Bandelj and Gibson 2020;Bieber 2020;Krzyzanowski 2018). This is to an extent reflected in cross-country differences in public attitudes to immigration, with existing research based on data from 2015 to 2017 suggesting that Serbian citizens hold more positive immigration attitudes than citizens of Poland and the Czech Republic (Esser et al. 2019, 128). However, research based on more recent polls suggest than patterns may have changed in recent years, as attitudes among Polish citizens became more positive (Esipova, Ray, and Pugliese 2020). The most recent wave of the European Social Survey (ESS) from 2018 confirms this, showing that the Polesalthough still ranking above the EU averagewere among the least likely of all CEE nations to agree that their country should allow no immigrants 'from poorer countries outside Europe' (20%), while the same was claimed by 29.3% of Serbs, 43.3% of Czechs, and 59.3% of Hungarians, which was by far the highest percentage in the sample (ESS Round 9, 2018). With regard to media independence, and specifically PSM independence, the Czech Republic stands apart from the other three countries as the only one where PSM remain relatively independent and also trusted by the public (Brogi et al. 2020;Newman et al. 2020). In contrast, PSM in Hungary, Poland, and Serbia are tightly controlled by the government, and have been transformed into instruments of government propaganda (Surowiec, Kania-Lundholm, and Winiarska-Brodowska 2020;Polyák 2019). Furthermore, in Hungary and Serbia, government control extends beyond PSM to commercial outlets, leaving only a small minority of independent mediaa situation reflected in the global rankings of media freedom, where Hungary and Serbia are ranked lower than Poland and the Czech Republic (Freedom House 2019).

Methodology
The current study used a mixed-methods approach, combining quantitative survey questionnaires, and expert survey, and qualitative interviews. A mixed-methods design has two advantages. First, it helps offset the relative weaknesses of each individual method for instance, limited generalizability and greater subjectivity of qualitative research, as opposed to limited contextual understanding and in-depth insight of quantitative research (Creswell 2015, 5). Second, it helps us acknowledge both the existence of media effects and active audience involvement in shaping their media use and interpreting messages. In sum, mixed methods enable us to produce generalisable findings without losing depth and contextual understanding, and contribute to a more rounded understanding of the role media play in shaping attitudes to immigration.
Quantitative analysis is based on two surveys. The first is a population survey (N = 4,092) collected online (CAWI, 75%) and via telephone (CAPI, 25%) in the Czech Republic, Hungary, Poland, and Serbia in December 2019 and January 2020. Quota sampling was used, designed to be representative of the general population for key socio-demographic quotas: age, gender, education, region, size of municipality and frequency of Internet usage. Pilot testing started in November 2019 and the fieldwork was conducted in December 2019 and January 2020. The second is an online expert survey (N = 60) conducted in the four countries among media experts with academic, professional, and civil society backgrounds between January and May 2020. Participants were selected through a combination of purposive and snowball sampling, following our project partners' suggestions and experts' recommendations.
Analysis of the population survey involved descriptive statistics and ordinary least squares (OLS) regressions. The main dependent variable measures attitudes towards immigration based on an index taking the mean of respondents' agreement to three survey questions: (a) immigrants abuse the welfare system, (b) immigrants are a threat to our culture, and (c) immigrants cause the rise of criminality. 1 This continuous variable ranges on a scale of 1-7, where higher scores denote more agreement with the survey statements and therefore more anti-immigrant attitudes. The main independent variable measures frequency of media consumption for accessing news in the previous month, based on a list of 16-18 media brandsincluding TV, radio, print, and online media specific to each country (see Appendix A); we also examined participants' consumption of both commercial media and public service media (PSM). We also included variables measuring gender, age, education, domicile (town) size, income, unemployment and voting for governing party as controls in the regression analyses. 2 Analysis of the expert survey shows descriptive statistics of the means and standard deviations of all media and public service media 3 as indicated by experts' responses to a question asking about specific outlets' coverage of immigration (neutral, or biased either for or against immigrants and refugees). 4 Qualitative analysis is based on 120 semi-structured interviews (30 participants per country), conducted in February and March 2020. Participants were recruited from population surveys, using a combination of purposive and quota sampling, and from personal connections. The sample was limited to participants who, based on survey responses, were at least somewhat engaged with news and politics, meaning they follow news on politics on a regular basis (minimum weekly). Quota sampling was used to ensure the sufficient diversity on several demographic dimensions known to shape both media use and political behaviour, namely age, gender, and domicile size; political preference was also considered. Each country sample was divided equally between three age groups (18-34, 35-59, 60+), gender (male, female), domicile type (urban, rural), and political preference (conservative, liberal, and, to a lesser extent, centre).
Interviews lasted an hour on average, and covered everyday routines, political engagement, the media environment, and news consumption routines and preferences. Of particular importance for this analysis, participants were asked to explain their answers to survey questions on immigration mentioned above, dealing with criminality, economics, and culture; in these instances, participants were reminded of their survey answers and directly requested to elaborate on their answers. Those participants found through personal contacts were asked to fill out the survey questionnaire before the interview. All participants were asked the same questions, translated into local languages. All interviews were transcribed directly from local languages to English; all material was analysed using thematic analysis (Braun and Clarke 2006) with support of the NVivo package. After coding for attitudes to immigration and news consumption routines, relevant demographic variables, and political orientations, individual profiles were constructed for each participant: this allowed for the analysis of patterns of regularities, and links between attitudes and news consumption both within and between countries.
Both the qualitative and quantitative research was given ethical approval from Loughborough University's Ethics Approvals (Human Participants) Sub-Committee (dated 20 March 2019), and all participants provided informed consent.
Between strong acceptance and strong refusal: attitudes and understandings of immigration Figure 1 shows the distribution of attitudes toward immigration among survey respondents by country. It is apparent that anti-immigration attitudes are strong in all four countries, with those who 'moderately' or 'strongly refuse' migration making up between 54.4% (Poland) and 72.8% (Czech Republic) of the sample. Acceptance of migration is much lower, ranging from 20.9% in the Czech Republic to 35.6% in Poland. As with existing comparative data on the issue (ESS Round 9, 2018; Esser et al. 2019; Globsec 2020), these results confirm the existence of more accepting attitudes in Poland, and more negative attitudes in the Czech Republic, but surprisingly suggest considerably higher (relative) levels of refusal in Serbia, similar to those found in the Czech Republic.
The analysis of interview data revealed significant differences in the understanding of immigration, which were linked to differences in attitudes. Participants in the 'strong acceptance' category, and to a somewhat lesser extent those in the 'moderate acceptance' category, often adopted a very inclusive understanding of the category 'immigrant', including themselves (or their nation as a whole) in this category or emphasised the similarities between themselves and immigrants, either with reference to the wider category of 'humans' or by mentioning their friendship ties with immigrants. Also common was the slippage between the term 'immigrants' and the semantically wider term 'migrants.' The following quote from one of our Hungarian participants is illustrative of these tendencies: … And it's dumb anyway because Hungarians are immigrants too. If I go back a thousand years to 978, chief Árpád came here. But others lived here too, for God's sake. Romans, Pannonian Avars lived here … Slavs, Slavs are also immigrants! Everyone is an immigrant … We migrate up and down the Earth, wherever we can. male,52) Another Hungarian participant with strongly accepting attitudes explained that she herself has been an immigrant: 'In London I was an immigrant too' female,36). Similarly, a Polish participant expressed that immigrants are also 'human beings': 'A human being is a human being … I believe every human being should be given a chance' male,38). Characteristically, participants in this category also tended to reject the view that immigrants had any more propensity for criminality than others, argued that immigrants did not take advantage of any economic or social systems, or that immigrants would not adversely affect the nation's culture.
'Strong refusal' was reserved for those participants who clearly stated that they are against immigration or spoke ill of immigrants. In contrast to participants in the 'strong acceptance' category, these participants typically adopted a very narrow understanding of immigration, and often explicitly or implicitly equated immigrants with Muslims or non-white immigrants, and sometimes also with refugees. The following extract, from one of our Polish interviews, provides a good example of this tendency: 'They don't want to work. They don't come here for work … they come here to settle, build mosques … ' female,63).
In a similar vein, another Polish participant quickly jumped from talking about immigration to talking about the need to keep Poland 'white', implying that immigrants are not only cultural but also racial 'others': 'I wouldn't want Poland to turn multi-culti as our neighbours for example. I'm very proud that Poland is white and pure' (Pol-08, female, 36). The emphasis on racial and/or cultural purity was sometimes combined with a perception of immigrants as dirty and as bearers of disease. For instance, one of our Czech participants, went as far as to say migrants bring disease into the country: 'I do not like at all that half of Prague is Ukrainian. They are not clean, they have illnesses, tuberculosis, they do not behave themselves, or hardly anybody does, I just do not like it that we let so much of them come here' (Cze-13, female, 39). Others, like one Serbian participant explained that migrants are dirty and do not respect the country: 'You can't ignore the fact that by their coming here, they don't respect the places they stay. They leave them dirty, sloppy … ' (Srb-23, female, 60+). In contrast to participants in the 'strong acceptance' category, participants in the 'strong refusal' category also tended to believe immigrants had a higher propensity for criminality and had a negative effect on both the economy and national culture.
The attitudes of other participants, outside of the 'strong acceptance' and 'strong refusal' categories, were less clear-cut, and generally combined elements of both types of labels and understandings of immigration outlined above. For instance, a participant in one of the 'moderate' categories may jump from talking about immigrants to talking about Muslims, but then also acknowledge some similarities between immigrants and the host population. Likewise, a participant in this category may not believe immigrants to be criminals or to have a negative impact on the country's economy, but may believe that immigrants will negatively affect the culture of the country. As one of our Serbian participants in the moderate refusal category stated: ' … it's tragic they [migrants] are given more rights than, let's say, you and I. In every regard' (Srb-03, male, 46).
The patterns we highlighted here display notable affinities with some of the findings of qualitative and mixed-methods analysis of media coverage of immigration. For instance, the conflation of 'immigrants' with Muslims, non-white immigrants, and to a lesserextent refugees in Serbia, evident among our 'strong refusal' participants, is consistent with the conflation of terms 'immigrant,' 'refugee,' and 'asylum seeker' found in (negative) media coverage of immigration (Baker and McEnery 2005). This parallel suggest that such conflation may be intrinsically associated with more negative attitudes to immigration. In one further example, a Hungarian participant (Hun-30, male, 38) specified who they believe to be migrants: 'We also have Chinese people in our building and I have no problem with them … here I meant migrants'. Furthermore, the slippage between 'immigration' and 'migration', common among our participants in the 'strong acceptance' category, seems consistent with the finding that the use of the term 'migrants' in media coverage tends to be associated with more positive terms than the use of terms 'immigrants,' 'asylum seekers,' and 'refugees' (Baker et al. 2008, 288). In sum, our research suggests that the qualitative patterns found in media coverage of immigration are indeed replicated in the way audiences understand and talk about immigration.

Links between media consumption and attitudes to migration: the role of PSM
Having established the broad patterns of attitudes to migration, we now move on to consider how these related to specific news consumption practices. Figure 2 5 shows the results of the OLS regression estimating the relationship between consumption of different types of media and attitudes toward immigration; any tail crossing the vertical red line indicates an insignificant relationship. Accordingly, the figure shows that when taking together all media types (TV, radio, print, and online), there is no significant correlation between people's attitudes towards migration and consumption of media. Contrastingly, when focusing specifically only on PSM (TV and radio), higher consumption of PSM media is associated with more anti-immigrant attitudes in Hungary and Poland, but with more pro-immigrant attitudes in the Czech Republic. In Serbia, there is no significant effect.
These cross-country differences can be explained with reference to systemic differences between countries outlined earlier in the paper, specifically levels of government control over PSM (and other parts of the media landscape) and the extent to which immigration is a politicised issue. This is clearest for Hungary and the Czech Republic. In Hungary, immigration highly politicised, and the combination of the government's anti-immigration campaign and government control over most of the country's media (including PSM) led to a situation whereby anti-immigrant content is hard to escape, and more positive or neutral coverage is limited to a handful of anti-government outlets. In contrast, the Czech media sector is characterised by higher levels of media independence, including PSM independence. While the proportion of negative immigration coverage in Czech media is larger than the proportion of positive coverage (Štětka, Mihelj, and Tóth 2020), PSM reporting on the subject is less negative, and consuming PSM was also shown to reduce anti-immigration prejudices among Czech citizens (Štětka, Mihelj, and Tóth 2020). Poland and Serbia are slightly more complex. In Poland, PSM is under government control, and the Polish government has been advancing an anti-immigration agenda, which helps explain our result. However, it is also important to note that government control over the media is less extensive than in Hungary, and general levels of anti-immigration sentiments in Poland are lower compared to the other three countries. Finally, in Serbia, government control over the media is similarly extensive as in Hungary. However, immigration is not as politicised, which helps explain the lack of correlation between PSM consumption and anti-immigrant attitudes, despite the fact that public attitudes to immigration are rather negative.
The results of our expert evaluation of media bias against immigrants support this interpretation. Figure 3 shows the average scores of bias in favour or against immigrants or refugees in the media, according to the experts survey. All averages fall to the right side of the neutral line, indicating that on average the media in all four countries are biased against immigrants or refugees; though PSM in the Czech Republic and to a lesser extent in Serbia is close enough to the neutral line that these can be taken to be 'truly' neutral. These two are in stark contrast with PSM in Hungary and Poland, which are significantly more biased against immigrants or refugees.

Using information sources to justify attitudes to immigration
In the final step, we considered what participants did with different sources of information about migration, and how they used these sources to articulate or justify their attitudes. Two key findings are worth highlighting: one relating to the general perceptions and uses of media as sources, and the other relating to perceptions and uses of family and friends as sources. Across all countries, participants were often critical of news coverage on the subject or claimed it did not influence thema finding aligned with conclusions of existing qualitative studies of audience engagement with immigration news (Haw 2020;Milioni and Vadratsikas 2016). However, the extent to which participants questioned news coverage varied in line with their attitudes to immigration, with this pattern being particularly clear in relation to social media. That is, participants with positive immigration attitudes tended to question negative information by either cross-checking with other sources or taking a critical view, while those with negative attitudes did not question it. For example, one Czech participant, who had strong pro-immigration attitudes, stated that he often heard news about migrants from their neighbour and news media; he then described checking various media as he wanted 'to verify the information' male,39). In contrast, participants with antiimmigrant attitudes seemed to take information gathered from social media for granted. For example, one strongly anti-immigrant Serbian participant described finding news about migrants on Facebook, without raising any doubts about its veracity: Well, I was reading how a migrant had cut a kid with a knife. First, he asked some boy for directions and then, when the boy tried to tie his shoes, he attacked him with a knife. I didn't really have the time to check whether it was true or not. [laughs]. female,42) It is unclear how exactly we should interpret these patterns. It is possible that people with differing attitudes to immigration use news sources in different ways, with participants holding anti-immigrant attitudes less likely to question the veracity of information on social media. It is also possible, however, that these patterns are a result of people reacting to information which contradicts their pre-existing attitudes; because media coverage of immigration tends to be negative, it is to be expected that those holding more positive attitudes are more likely to question it. This suggests that the nature of critical reactions to media coverage, and their links with attitudes to immigration, depends on the nature of migration coverage in a particular context.
In contrast to media sources, which were often contested, several participants mentioned friends and family as important and trustworthy sources of information about migrants. Contrary to information from media sources, which was often questioned, information from family and friends was presented as trustworthy, and used to reinforce one's claims. This practice was particularly common among people with more antiimmigrant attitudes. For example, one Polish strongly anti-immigrant participant, when discussing incidents in Germany, stated she does not 'follow German media, but I have family in Germany, meaning my fiancé's family. Whenever his aunt visits, she talks about it a lot' (Pol-08, female, 36). Information about immigrants often came from friends, family or acquaintances working abroad, especially in Germany and Sweden. For instance, one strongly anti-immigrant Czech participant, when asked if she had heard of news about immigrants in the media, responded: While the tendency to place more trust in the experiences of friends and family was more common among participants with anti-immigrant attitudes, it could occasionally be found also among those with pro-immigrant attitudes. For instance, for one strongly proimmigrant Polish participant, a testimony from a family member lends credibility to more positive attitudes: … but when I talk to my family living in Germany, or to my parents, who live in Paris, my dad works in construction, and works with what you could call 'fresh arrivals', he spends his free time with them, and I can't figure out when would they have time to commit all those crimes. So, I hear a lot about it but when I try to verify this information, I cannot find its source. male,25)

Conclusion
This study brought three key insights into audience engagement with immigration news, each with significant implications for further research in this area. First, it showed that different attitudes to immigration are intertwined with different understandings of what 'immigration' entails. While participants with strongly positive attitudes interpreted immigration in broader terms, drawing an equation between 'immigrants' and 'humans' or people,' and often replacing the terms 'immigration' and 'immigrant' with 'migration' and 'migrant,' participants holding strongly negative attitudes adopted a considerably narrower interpretation, associating 'immigration' with specific subgroups of immigrants, such as Muslim immigrants or non-white immigrants. These links between attitudes and understandings of immigration appear consistent with discursive patterns highlighted in existing qualitative and mixed methods research on media coverage of immigration Gabrielatos and Baker 2008) as well as with patterns of association between attitudes to immigration and understandings of immigration established in survey-based research (Blinder 2015). Given that existing, survey-based research suggests that subtle changes in the choice of labels in immigration news have capacity to influence audience attitudes to immigration (Blinder and Jeannet 2018;Merolla, Ramakrishnan, and Haynes 2013;Rucker, Murphy, and Quintanilla 2019) future research could fruitfully explore this aspect of audience engagement with immigration news further, using mixed methods approaches.
Second, our data show that links between attitudes to immigration and PSM consumption differ significantly from country to country, in line with the level of PSM independence, with PSM consumption significantly correlated with negative attitudes in Hungary and Poland, and with positive attitudes in the Czech Republic (and no significant correlations either way in Serbia). This result shows that the scope for cross-country variation in this area is even wider than existing research suggests (Beyer and Matthes 2015;Strabac, Thorbjørnsrud, and Jenssen 2012), and demonstrates that systemic differences in political and media environmentsin particular, the level of media independence,can shape the relationship between attitudes to immigration and news consumption. Once we broaden the scope of comparison to countries where the media lack independence, it becomes apparent that PSM can also contribute to negative attitudes to immigration, rather than having either a moderating or neutral effect. Whether or not such a negative effect will materialise depends on several contextual factors, including the government stance on immigration, the relationship between PSM and other parts of the media system, and public attitudes to immigration. On a more general note, these findings also highlight the importance of examining the impact of government-controlled media on the migration-media nexus, especially in parts of the world experiencing democratic backsliding. Our results indicate that government-controlled media in such countries can further strengthen anti-immigrant attitudes, while also confirming that independent PSM can help counteract them.
Third, our data suggest a difference between audience uses of media and nonmedia sources of information on immigration. Across all countries, participants were either critical of news coverage of immigration, or denied its impact on their own attitudes. In contrast, information about immigration gained from family, friends, and acquaintances was presented as more trustworthy, and was regularly used to justify one's attitudes. Furthermore, these divergent perceptions of media and non-media sources of information were associated with different attitudes to immigration: negative information about immigration was more often questioned by participants with positive attitudes toward immigration, who either crosschecked with other sources or viewed news with a critical lens; in contrast, participants with negative attitudes were less likely to question news about immigrants. These results invite us to broaden our investigation of audience engagement with immigration news beyond the media, and investigate how public attitudes to immigration are formed through interactions with multiple sources of information, including non-media sources.
It is also important to mention some limitations. First, our observations about the impact of systemic media and political factors are based on a small sample of only four countries. A wider range of countries would enable us to develop a more fine-grained understanding of how the three systemic factorsnamely levels of media (and PSM) independence, public attitudes to immigration, and elite reliance on antiimmigration rhetoricinteract in different ways to shape the role media play in informing immigration attitudes. Second, our assessment of media bias was reliant on expert assessment, while conducting a media coverage analysis would allow for greater reliability and nuance. And finally, our measures of media use were based on selfreported use, whichwhile still standard and widely used in the filedis known to suffer from misreporting and exaggeration.
Despite these limitations, our study brings important additions to existing knowledge on media and immigration attitudes and helps direct attention to new avenues for future research. First, the study highlights the importance of benefits broadening the study of media and immigration beyond long-established liberal democracies of Western Europe and Northern America, particularly at a time when immigration acts as a key object of political contestation. Future research on media and immigration attitudes in other parts of the world would enable us to further refine our understanding of how the key features of media and political systems condition the way media intervene into the formation of immigration attitudes. Second, the research presented here illustrates the benefits of adopting a mixed methods approach to the analysis of media and immigration attitudes and using it to combine the analysis of media affects on immigration attitudes with a consideration of audience uses of the media. Future research could usefully adopt such an approach to study the interaction between media use and attitudes to immigration in other socio-political contexts, or take into account the contribution of entertainment genres and non-media sources of information.