It's Who You Know. Political Influence on Anti-Immigrant Attitudes and the Moderating Role of Intergroup Contact

Introduction

1.1 According to group threat theory (Blumer 1958), which is one of the leading theories in research on anti-immigrant attitudes, such sentiments will rise when competition over available resources intensifies and native populations experience that immigrants threaten what they perceive to be rightfully theirs. Over the last decades, European countries have witnessed a steady increase in immigration with a consequent rise in the foreign-born as a percentage of the population. In the same period of time, issues of immigration and immigrant presence have become increasingly politicized. In public and theoretical debates, concerns have been raised that the continued immigration will generate divided societies with disrupted social cohesion and a diminishing will to contribute to the common good (Alesina & Glaser 2004; Putnam 2007). There are political parties that have incorporated such fears in their rhetoric and fuel them further through talking about cultures as incompatible as well as actively pitting groups against each other. While we theoretically expect such developments to generate an increase in anti-immigrant attitudes, there are no indications that peoples' attitudes have changed substantially over the past decade. Although there are variations between countries, the general trend seems to be stability or a slight decrease in anti-immigrant attitudes (Meuleman et al. 2009).

1.2 The absence of strong attitudinal reactions among native populations in Europe suggests that the relationship between immigrant presence, hostile political rhetoric and anti-immigrant attitudes is more complex than proposed by group threat theory. As immigration also increases opportunities for intergroup contact and friendship, it cannot be assumed that natives necessarily interpret growing immigrant populations as a threat. Although a negative political climate makes such an interpretation more likely (Hopkins 2010), other factors – for example personal experiences of intergroup contacts – can influence to what extent political statements resonate among individual citizens. It is possible that out-group friends and acquaintances not only reduce anti-immigrant attitudes, as predicted by intergroup contact theory (Allport 1954; Pettigrew & Tropp 2011), but also make people less likely to be influenced by negative political rhetoric. The study tests this possibility through examining whether intergroup contact, either at the workplace or with friends, to some extent can inoculate people against a hostile political climate.

1.3 The overall aim is to examine how political rhetoric relates to anti-immigrant attitudes depending on experiences of intergroup contact. Drawing on the insights from both group threat theory and intergroup contact theory, this study approaches political representatives and social ties as sources of frames (Chong & Druckman 2007a), with the potential to influence how immigrants are perceived by native populations in 21 European countries. In contrast to political frames, interpersonal contact with immigrants implies an alternative source of information presenting alternative frames with the potential to undermine what political representatives convey.

1.4 Obviously, not all kinds of political rhetoric is relevant in relation to anti-immigrant attitudes. In this study, I use manifesto data from the Comparative Manifesto Project (CMP) to test the effect of a type of political frames that both theoretically (Triandafyllidou 1998) and empirically (Kunovich 2009) relates to anti-immigrant attitudes, namely nationalistic frames with the aim of defining and consolidating the national way of life. Indicators of intergroup contact as well as of the dependent variable anti-immigrant attitudes are gathered from European Social Survey (ESS).

1.5 Despite strong theoretical arguments linking attitudes to their political setting, it is only recently that such circumstances have attracted attention in comparative studies of anti-immigrant attitudes. Analyzing the effects of political factors such as integration policies (Schleuter et al. 2013), the institutional make-up of countries (Weldon 2006), and electoral outcomes for particular groups of parties (Sprague-Jones 2011), studies support political circumstances as important explanations of attitudes towards immigrants. The present study contributes to this growing literature and develops it further by examining the conditionality of political influences. Thus, the interest goes beyond the direct effects of political contexts to also explore how individual experiences of contact interact with political messages to produce an attitudinal outcome. As such, the study contributes to a better understanding not only of political influences on attitudes towards immigrants, but also of the consequences of intergroup contact.

Theory

2.1 In this study, group threat theory serves as the theoretical point of departure. It is a theory that approaches prejudice as a matter of intergroup relations, and sees group competition and experiences of threat as the main triggers of negative out-group attitudes. Since it was originally introduced in the 1950s, it has received extensive support in studies on attitudes towards different minority groups (Bobo 1983; Strabac & Listhaug 2008; Raijman & Semyonov 2004). Since Quillian (1995) demonstrated its applicability with regard to anti-immigrant attitudes, the theory has been widely applied in research that seeks to understand attitudes to immigrants and immigration (McLaren 2003; Scheepers et al. 2002; Meuleman et al. 2009; Hjerm & Nagayoshi 2011). In these studies, experiences of threat have emerged as a strong and consistent explanation of anti-immigrant attitudes (Ceobanu & Escandell 2010).

2.2 According to group threat theory, an increase in anti-immigrant attitudes is expected when natives feel that immigrants threaten their interests. The threat can be tied either to economic and material resources (Scheepers et al. 2002) or to more symbolic values (McLaren & Johnson 2007). Although threats to material interests have been the main focus in much previous research (Quillian 1995; Kunovich 2004), recent findings suggest that symbolic and cultural threats are of equal, if not greater, importance to attitudes towards immigrants in Europe (Sides & Citrin 2007; Hjerm & Nagayoshi 2011).

2.3 Anti-immigrant attitudes also depend on the contextual circumstances. According to group threat theory, contexts where feelings of threat evolve have a central role in enhancing or mitigating such experiences. For example, in times of economic recession (Semyonov et al. 2008) or following a sudden influx of immigrants (Coenders et al. 2008), intensified competition implies a greater challenge to the position of the native population. Threats are also particularly pertinent in contexts of historically and culturally rooted racial hierarchies – reflected, for example, in white Americans' attitudes toward the Afro-American minority (Dixon 2005) – and when fueled by political actors (Hopkins 2010; Bohman 2011). Although the size of the immigrant population is generally considered as significant to feelings of threat (Quillian 1995; Semyonov et al. 2006), the documented difficulties of assessing the actual number of immigrants (Herda 2010) and inconsistent empirical results (Hjerm 2007; Schlueter & Wagner 2008) indicate a more complex relationship. Sides and Citrin (2007) suggest that as the immigration issue has become increasingly politicized, anti-immigrant attitudes have gradually detached from social demographic realities (i.e. the actual size of the immigrant population) and increasingly come to depend on how information is spread and framed by political representatives and media.

2.4 Prejudice in general revolves around ideas of hierarchy and separation in the sense that these attitudes include feelings that the other group is naturally inferior to, as well as fundamentally different from, one's own group (Blumer 1958; Bobo & Hutchings 1996). This suggests that consolidated boundaries (actual or perceived) between immigrants and natives provide a breeding ground for anti-immigrant attitudes as they limit intergroup mobility and interactions. Wright (2011) demonstrates that political factors play an important role in influencing the rigidity of these boundaries. Examining the normative content of national identity, he finds that political arrangements and policies have bearing on whether or not immigrants are welcomed as part of the national community. Through framing immigrants as fundamentally different or as generally the same, politicians may influence the preconditions for difference and distance between groups in society and as a result also affect natives' perceptions of immigrants.

2.5 Already Blumer (1958) emphasized the importance of the political context, arguing that perceptions of out-groups are largely dependent on public space activities and statements made by influential actors. Prejudicial attitudes, in this sense, are formed in interactions between group members, interactions where the major influence …is exercised by individuals and groups who have the public ear and who are felt to have standing, prestige, authority and power (Blumer 1958:6). Such processes of influence can be described as framing effects (Chong & Druckman 2007a) where political representatives offer interpretative models that provide meaning, and help individuals make sense of their surroundings. A frame is a kind of schemata of interpretation that organizes people's experience (Goffman 1974).^[1] It helps define occurrences and objects and accordingly, influences how issues and phenomena are evaluated (Druckman 2001a). Its role in the formation of attitudes implies that political actors are keen to influence peoples' frames of thought through presenting and discussing issues in a way that benefits their own purposes. They use political frames to influence how individuals understand and interpret the issue or phenomenon at hand.^[2] An initial reluctance on behalf of natives may thereby find its explicit form, and tie in with a more cohesive set of ideas, when confronted with broader anti-immigrant frames. Political frames induce manifest xenophobia through offering ways to approach immigrant presence that are in line with more or less articulate sentiments among the native population (Rydgren 2003).^[3]

2.6 Taken together, this leads us to the first hypothesis of the study:

(1) The more political representatives use nationalistic frames, the higher the levels of anti-immigrant attitudes.

2.7 Meanwhile, there are theoretical reasons to expect that not everyone is equally open to such influences. Studies drawing on group threat theory show that contextual factors often interact with individual characteristics in influencing the degree of anti-immigrant attitudes (Bohman & Hjerm 2013; Scheepers et al. 2002). People react differently to certain circumstances since they are more or less personally affected by, or personally concerned with, what is going on. This is very much the case in regard to political rhetoric. Previous research demonstrates that political frames do not necessarily influence different people's frames of thought to the same extent. Rather, framing effects are dependent not only on contextual (Chong & Druckman 2007b; Druckman 2001b), but also on individual, factors (Brewer 2001; Haider-Markel & Joslyn 2001). People are, for example, more prone to adhere to frames if conveyed by trusted political elites that share their ideological position (Zaller 1992), something which has received empirical support in recent studies on anti-immigrant attitudes (McLaren 2001; Bohman 2011). In terms of individual factors, people with strong predispositions appear to be much less susceptible to political framing than those without any clear a priori formulated opinion (Matthes & Schemer 2012). The same holds for those who have access, compared to those with limited access, to alternative frames (Druckman & Nelson 2003). The study at hand highlights how a particular individual factor may moderate the effect of political frames - a factor associated both with developing predispositions and access to alternative frames - namely the individual's involvement in intergroup contact.

2.8 Previous research identifies intergroup contact as a central explanation of prejudice and other anti-minority attitudes (Pettigrew & Tropp 2006; Dovidio et al. 2003). The findings support the intergroup contact hypothesis, as originally described by Allport (1954). The hypothesis states that, if the right circumstances prevail, people who get involved in close intergroup contact will become more positively disposed towards the other group. By stressing the importance of the 'right circumstances', contact theory asserts that not all kinds of contact generate the anticipated effect. For example, if contact is to reduce prejudice, it is advantageous if the groups involved have equal status, at least in the situation where the contact takes place, and if the contact is sanctioned by the people in charge of the particular situation. While critics have argued that all of the conditions rarely are fulfilled outside the laboratory situation (Dixon et al. 2005), later developments of the theory claim that these circumstances should be treated as facilitating rather than essential (Pettigrew & Troop 2006). Still, one condition appears to be particularly important, namely that the contact provides opportunities to become friends (Pettigrew 1998). Close high-quality relationships have repeatedly been shown to be the kind of contact that most effectively reduces negative out-group feelings (Pettigrew & Troop 2011).

2.9 To better understand why contact should make people less prone to take impression by nationalistic rhetoric, we have to know more about how – and not only when – contact reduces anti-immigrant attitudes. Pettigrew (1998) discusses a number of processes through which contact can counter negative out-group attitudes, for example through changing emotions, acquiring knowledge about the other group, and through evoking new perspectives on the own group. To the extent that people learn more about other groups and discover intergroup similarities through intergroup contact, it is also likely to impact the effect of political frames. Getting to know members of the other group undermines the estrangement and social distance that Blumer (1958) identified as fundamental to prejudicial attitudes. Hence, in contexts where political parties convey stereotypes and images of immigrants as fundamentally different, those with immigrant friends will be less likely to accept these images as grounds for evaluating immigrants and immigrant presence. By discovering that the other group is not all that different, as well as by affective changes like reduced anxiety (Paolini et al. 2004), people with intergroup contacts become less likely to accept frames where immigrants are portrayed as a cultural threat. To the extent that contact encourages natives to view their own group in a different light, and (at least ideally) acknowledge that their own group's way of doing things is not the single, or even the best, one available, it will also counter the effect of nationalistic frames. That nationalistic sentiments often go hand in hand with skepticism towards outsiders (Ceobanu & Escandell 2008; Hjerm 1998) suggests that such reassessments will reduce anti-immigrant attitudes (Pettigrew 1997) as well as strengthen resistance to nationalistic political frames.

2.10 The framing by political elites must resonate with what people observe and experience in their everyday life. To gain an audience, these frames must display sufficient empirical credibility and cultural resonance (Benford & Snow 2000) meaning that individuals must be able to relate to what the politicians say. Natives with limited personal contact with immigrants, who have not already developed any positive attitudes, are more likely to adhere to negative political frames. Any initial suspiciousness or unarticulated negative sentiments on their behalf are at risk of becoming more pronounced and surfacing to a manifest level when stimulated by political elites (Rydgren 2003). In accordance with contact theory, people involved in intergroup contact are less likely to hold such negative feelings. Their personal experiences provide them with alternative frames that make the negative political frames appear less valid.

2.11 The second hypothesis thus reads;

(2) Political representatives' use of nationalistic frames influences attitudes towards immigrants primarily among natives without personal contact with immigrants.

2.12 Previous studies show that contact can moderate the effect of a threatening environment. McLaren (2003) and Schneider (2008) both find that a large immigrant population triggers feelings of threat particularly among native born people without contact with immigrants. Those who have immigrant friends are significantly less likely to feel threatened by such a context, probably because contact reduces the social and cultural distance underpinning cultural threats. Through hypothesis two, this study set out to test whether contact has the same effect on threats when conveyed by political actors.

2.13 Before proceeding, a few further points must be made with regard to intergroup contact as a potential moderator of political frames. First, any empirical test of the contact hypothesis must attend to the imminent risks that causality, in fact, runs in the opposite direction. In other words, the risk that it is not contact that influences anti-immigrant attitudes as much as it is anti-immigrant attitudes that determine the degree of contact. People sympathetic to other groups are indeed more prone to seek their company, but longitudinal studies have convincingly demonstrated that contact has an independent effect on prejudice (Binder et al. 2009; Levin et al. 2003; Eller & Abrams 2004; Christ et al. 2014). Testing the effect of contact at time 1 on prejudice at time 2, while prejudice at time 1 is controlled for, they find that net of the prejudice-contact effect, contact also serves to reduce prejudicial attitudes. Further, the path from contact and prejudice is often found to be stronger (Pettigrew 1997; Powers & Ellison 1995), in particular when it is a question of valued contacts with 'friendship potential' (Van Dick et al 2004; Pettigrew 1998). The current study tests two different indicators of intergroup contact: immigrant friends and immigrant colleagues. While a focus on immigrant colleagues restricts the analysis to the working population, it has the benefit of reducing the risk of reversed causality, as the freedom to choose to enter the contact situation is significantly more limited with regard to work relations, as compared to friendships. Also, if the same pattern is visible with regard to both contact indicators it provides extra support for the reliability of the findings.

2.14 Second, understanding how positive contact leads to positive attitudes towards the entire out-group is important, particularly as any lack of, or limitation in, such a generalization would interfere with the moderation of frame effects. Put differently, if no parallels are drawn between individual experiences and the content of political frames, the prospect of contact inoculating individuals against political rhetoric will be highly limited. Along with the question of causality, generalization of effects is one of the main problems associated with the original formulation of the contact hypothesis (Pettigrew 1998). Also, it is particularly relevant to the current study as this study focuses on attitudes towards 'immigrants', which in Europe is a very heterogeneous category that encompasses people of different origin with a multitude of languages, ethnicities, religious denominations etc. A friend or colleague, even if she/he has immigrated, is not necessarily perceived as a representative of the category 'immigrants', or more specifically, not perceived to be the same 'kind' of immigrant as the immigrants referred to in political frames. While such a discrepancy would significantly restrict contact as a potential moderator, there is significant evidence that generalization occurs even beyond the groups involved in the contact situation (Tausch et al. 2010; Pettigrew 2009). Through promoting new perspectives on one's own group and making people less provincial (Pettigrew 1997), intergroup contact reduces prejudice in a manner that improves attitudes also towards other out-groups. In other words, it is reasonable to expect some degree of moderation of the framing effect also in the case of discrepancies between the groups in question.

2.15 To summarize, this study tests two hypotheses about the relationship between political frames, intergroup contact and anti-immigrant attitudes. The type of political frames tested in the analysis is nationalistic frames to define and consolidate the national identity. As such, they touch upon important issues tied to the boundaries of the national community which relate specifically to the relationship between natives and immigrants. The exclusive side of national identities, manifested in the close relationship between nationalism and anti-immigrant attitudes (Ceobanu & Escandell 2010), leads us to expect that anti-immigrant attitudes will increase the more political representatives use nationalistic frames, but also that this relationship will be weaker if people have personal contacts giving them access to alternative frames.

Data and method

Dependent variable and intergroup contact

3.1 To be able to examine political frames' impact on anti-immigrant attitudes and the moderating role of intergroup contact, it is necessary to collect data from several data sources. Data on individual level variables is retrieved from the European Social Survey (ESS). The ESS is a comprehensive cross-sectional European survey carried out every two years. In this study, I use data from ESS 2002 as this round includes the most extensive set of questions concerned with views on immigration. Also, it is the only round covering questions relating to intergroup contact.

3.2 Twenty-one of the countries that participated in ESS 2002 are included in the study; Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Luxembourg, Netherlands, Norway, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland, United Kingdom.^[4] In the analysis, all non-natives were excluded from the sample, leaving a total number of 36,566 respondents.

3.3 For the dependent variable I use an index of attitudes towards immigrants. It is based on six items from the ESS: 1) Would you say that people who come to live here generally take jobs away from workers in [country], or generally help to create new jobs? 2) Most people who come to live here work and pay taxes. They also use health and welfare services. On balance, do you think people who come here take out more than they put in or put in more than they take out? 3) Would you say it is generally bad or good for [country]'s economy that people come to live here from other countries? 4) Would you say that [country]'s cultural life is generally undermined or enriched by people coming to live here from other countries? 5) Is [country] made a worse or a better place to live by people coming to live here from other countries? 6) Are [country]'s crime problems made worse or better by people coming to live here from other countries? Together these items load on one single factor, with high Cronbach's alpha scores in all 21 countries.^[5] As this indicates that the items, although focusing on different aspects of immigrant presence, still capture a more general attitude towards immigrants, they were merged into an additive index. For respondents who answered at least four out of the six questions, missing values were imputed from mean values on their valid responses. The index was then standardized to range between 0-100 with higher values representing higher levels of anti-immigrant attitudes.

3.4 Intergroup contact is measured by two separate items. The first addresses the closer kind of contact: whether natives have immigrant friends, and is captured by the survey question: Do you have any friends who have come to live in [country] from another country? The second kind of contact, contact at the workplace, is captured by the question: Do you have any colleagues at work who have come to live in [country] from another country? When answering the questions, the respondents were given the alternatives; Yes several, Yes a few, No, none at all. With regard to immigrant colleagues, the respondents who stated that they were not currently working were excluded from the analysis.^[6]

Independent variable

3.5 In order to capture the main independent variable; political frames, I use data from the Comparative Manifesto Project (CMP) 1990-2003 (Klingemann et al. 2006). The CMP data set is based on content analysis of the political manifestos launched by political parties before national elections. While expert surveys often focus on party positions in a single year and cover a limited number of countries, the CMP, in full, includes manifestos from 1945 to 2003 in over 50 countries. Acknowledging these advantages, a growing number of studies use the CMP data to capture party positions and issue saliency, both with regard to immigrant presence (Hjerm & Schnabel 2012; Arzheimer & Carter 2006) and in the case of other political issues (Netjes & Binnema 2007; Kulin & Seymer 2014). In the CMP data set, the parties are scored based on the percentage of statements they devote to a particular issue in each manifesto.^[7] For the purpose of this study, I treat this score as an indicator of the extent to which political parties use a certain kind of frames in their broader political rhetoric. More specifically, the interest is in their use of nationalistic frames.

3.6 The content of each party manifesto is categorized based on 56 different topics. To capture the party's use of nationalistic frames, I use manifesto articulation with the aim of defining and consolidating the national identity, captured by the category 'positive references to the national way of life' (per601).^[8] This category includes appeals to patriotism and nationalism, support for established national ideas, and protection of the state from subversion. While representing the inclusive side of national identification, such references also imply the existence of out-groups not included in the national 'we'. As experiences of national sameness become meaningful only when contrasted with others, processes of demarcation and exclusion are always present in ideas of national identity (Triandafyllidou 1998). Hence, the relative importance assigned to the national way of life is also an indication of the extent to which political parties seek to differentiate the national in-group from those not defined as nationals. In Europe today, the typical negative reference category in this regard is immigrants (De Figueiredo Jr & Elkins 2003), reflected for example in a close relationship between anti-immigrant attitudes and nationalism at individual level (O'Rourke & Sinnott 2006).

3.7 Directly applying the score provided in the CMP data set would give all political parties equal weight, regardless of whether it is a small party or a party dominating the national political landscape. As this study uses manifesto emphasis as a proxy for frames as they contribute to the general political climate, it is necessary to take the size of the political party into consideration. To better account for the fact that larger political parties often are more influential compared to smaller parties, each party's manifesto score is multiplied by their size, given by the percentage of votes gained in the election.^[9]

3.8 Also, the broad time coverage of CMP makes it possible to create a measurement based on framing over time. Using the relative importance political parties devoted to the issue during the period from 1990 to 2002, a national mean is calculated, standardized for the number of parties and elections in each country. This generates a framing measurement that allows for country comparisons and that is less sensitive to temporary fluctuations. It stretches from 1.59 to 86.61^[10], where higher values indicate that more emphasis is put on the issue on the country's national political arena.

Control variables

3.9 As manifesto data is gathered partly during, or immediately after, periods of dramatic political change in parts of Europe, I have included a variable controlling for the transfer countries (i.e. countries previously tied to either the Soviet Union or Yugoslavia). This is considered necessary to control that the findings are not due to any political convulsion or political instability in these countries during the time when the political manifestos were formulated. I also control for a number of contextual factors that may impact both anti-immigrant attitudes and the focus of political party rhetoric. These include indicators of the economic conditions in each country (GDP per capita, unemployment level) as well as the size of the foreign population. Drawing on group threat theory, a constrained economy and a large foreign population are expected to intensify competition and threat, at the same time as they may induce political representatives to resort to scapegoating and welfare chauvinism. With regard to the economic measurements, data are gathered from the World Bank.^[11] Data for the size of the foreign population are gathered from OECD.^[12] At individual level, a number of additional variables previously identified as important with regard to anti-immigrant attitudes are included the models. These are; age, years of education, gender, unemployment and political interest. Traditionally in Europe, right-wing parties have taken a tougher stance on immigration than their left-wing counterparts (Van der Brug & Van Spanje 2009). To control for this, the models include a variable that captures the individuals' ideological position (left/right).

The models

3.10 This study sets out to examine the effect of a contextual factor (political frames) on an individual-level outcome (anti-immigrant attitudes), as well as how this effect is moderated by a specific individual-level feature (intergroup contact). Due to their ability to handle this kind of hierarchical structures, multi-level models (Hox 2010; Steenbergen & Jones 2002) emerge as the most effective tool to pursue the aim of the study. The multi-level technique is further justified by the nested data structure, as multi-level models can control for the fact that individuals clustered within countries share a certain share of attitudinal variance. While a constant concern in studies using countries as the main contextual unit is attaining a sufficiently large number of level-two units, the study at hand covers 21 countries (20, in parts of the analysis) which correspond to other applications in comparative research (McLaren 2003; Schneider 2008).

3.11 In order to test the two hypotheses, I estimate a number of additive and multiplicative random intercept and random slopes models. The first model (Model 0) is calculated without explanatory variables to examine whether there is enough attitudinal variation between the countries to justify further analysis. Model 1 tests the first hypothesis. Besides the indicator of nationalistic frames, the model also includes the control variables on individual and contextual level. However, as the problems of few level-two units and the consequent limited inter-country variability are further accentuated if models include many contextual variables simultaneously – especially when cross-level interactions are involved – I have been careful not to include too many contextual variables in the same model. The contact indicators are added in Model 2 and 3 respectively.

3.12 The multiplicative models, 4 and 5, are used to test the second hypothesis. As preliminary analyses revealed Greece as an extreme outlier, both with regard to anti-immigrant attitudes and nationalistic frames, and as highly influential with regard to the outcome of the multiplicative models, Greece was at this stage excluded from the analysis.^[13] The analyses in Models 4 and 5 thereby cover 20 and not 21 countries.

Results

4.1 The first model (Model 0) reveals that 10,5 % (ICC= 0.1056) of the variation in peoples' attitudes towards immigrants can be attributed to the country-level. Hence, in addition to the arguments outlined in the theoretical section, this serves as an empirical indicator that country-specific features have bearing on anti-immigrant attitudes. To examine whether the political climate is one such factor, I start by testing the first hypothesis: The more political representatives use nationalistic frames, the higher the levels of anti-immigrant attitudes.

4.2 In Model 1, nationalistic frames emerge as positively and significantly related to the dependent variable.^[14] More nationalistic framing by political parties predicts higher levels of anti-immigrant attitudes, an effect that remains after including the control variables. Thus, the impact of nationalistic frames cannot be attributed to economic conditions or to the share of foreign-born in each country. Nor can it be explained by any uneven country-distribution of the individual-level controls, variations in unemployment-levels or particular political circumstances in the transfer countries during the period of study.^[15] Taken together, this suggests that the first hypothesis is correct in that more nationalistic framing is associated with higher levels of anti-immigrant attitudes. Further, that GDP per capita and the share of foreign born emerge as unrelated to anti-immigrant attitudes, when modelled together with nationalistic frames, provide support for Citrin and Sides' (2007) suspicion that the politicization of immigration have reduced the relevance of actual competitive conditions for peoples' attitudes.

Table 1: Nationalistic frames and anti-immigrant attitudes. Multi-level models

4.3 The individual level controls display the effects I expected, based on previous research (see for example Semyonov et al. 2006; Kunovich 2004). Age is positively related to the dependent variable, indicating that older people are more inclined than younger people to hold anti-immigrant attitudes.^[16] There is a slight difference between men and women, with women, on average, being somewhat less anti-immigrant as compared to men. The difference, however, is a very small one, 0.37 points on a scale 0-100. The expectation that people in more vulnerable socio-economic positions are more negative towards immigrants is supported by the effects of education and unemployment. That political interest is negatively related to anti-immigrant attitudes means that those with less political interest tend to have more anti-immigrant attitudes. The positive effect of left/right orientation suggests that people further to the right (on a scale 0-10) are more likely to hold anti-immigrant attitudes.

4.4 The contact indicators are included in model 2 and 3 and relate to the dependent variable in the way suggested by intergroup contact theory. Natives who have either immigrant friends or immigrant colleagues, regardless of whether they have many or just some, emerge as significantly less prone to hold anti-immigrant attitudes compared to those who are not at all involved in such contacts (the reference category). A comparison of the effects of friends and colleagues supports the idea that closer relationships are more effective in reducing negative sentiments, although it is important to note that the models concern different populations: all native born in model 2 and all native born who are currently working in model 3.

4.5 To examine whether nationalistic frames have different effects depending on the degree of intergroup contact, I include cross-level interaction variables in random slope models, 4 and 5. According to Hypothesis 2 it is expected that Political representatives' use of nationalistic frames influences attitudes towards immigrants primarily among natives without personal contact with immigrants. As indicated by the models, as well as graphically illustrated in Figures 1 and 2, this expectation is largely confirmed by the analysis. There is a significant interaction between nationalistic frames and intergroup contact in the sense that people who have many immigrant friends or colleagues are less prone to be influenced by nationalistic frames compared to those with no immigrant friends or colleagues.^[17] Note that while there is a significant difference in the degree of anti-immigrant attitudes among those with some, compared to those with no, immigrant friends/colleagues, the effect of nationalistic frames is the same in those two groups. Rather, it appears to be primarily the presence of many friends/colleagues that counters the effect of political frames. In fact, in the groups 'many friends' and 'many colleagues', the effect of nationalistic frames is not significantly different from zero.^[18] This suggests that nationalistic framing is important only among those with no or limited intergroup contacts, and not among natives who have either many immigrant friends or many immigrant colleagues. Although the contact indicator focuses on the size of intergroup networks (none, some, many), it is also likely to tap into contact quality, as having many friends/colleagues increases the probability that at least one of these contacts is of the long-term close kind considered to be most effective in reducing negative out-group attitudes (Pettigrew 1998).

Table 2: Nationalistic frames by intergroup contact. Multi-level models

4.6 That contact has an effect also at the workplace is interesting given that people have far less influence as to who their colleagues are than when choosing friends. Still, it is important to note that friends and colleagues in many cases are likely to overlap, so that one's colleagues also are one's friends and vice versa. The 'colleagues vaccine' may in other words also accommodate a significant dose of 'friends vaccine'. Although the effect of colleagues remains if the variable 'friends' is added as an additional control, which suggests that colleagues are important also in their own right, this points to a potential problem related to the contact indicators. The self-reported measurements of intergroup contact are potentially problematic as people are likely to have different ways of defining 'friends'/'colleagues' as well as 'many'/'few'. While this limitation cannot be solved using the available data, future studies should seek to construct more fine-tuned measurements that enable a closer examination of the moderating role of different kinds of contacts.

Figure 1. Nationalistic frames by immigrant friends

Figure 2. Nationalistic frames by immigrant colleagues.

4.7 In sum, the findings from Models 1-5 provide support for Hypotheses 1 and 2. There is a significant positive effect of nationalistic frames on anti-immigrant attitudes. This effect is moderated by personal experiences of intergroup contact so that it is primarily natives with no or only a few immigrant friends/colleagues that are influenced by what the political parties convey. Having many contacts seems to inoculate against political influences, and this regardless of whether the contacts are friends or colleagues. Adding the country-level controls (GDP per capita, per cent foreign born, unemployment level, transfer countries) does not interfere with the findings in either of the models.

4.8 All analyses in this study are carried out using cross-sectional data. In the future, longitudinal and/or experimental studies would be useful in closer establishing the causality of both the contact-attitude relationship and the political framing-attitude relationship. The focus of this study has been the effects of political frames and intergroup contact, which implies that more knowledge is needed with regard to what is likely to be a complex interplay between attitudes and contact and attitudes and party rhetoric. With increased availability of longitudinal data (on countries and on individuals) comes greater opportunities to study attitudinal developments over time, for example through comparing attitudes before and after politically important events such as national elections or referendums. Still, the findings from the current study bring new insights related to the relationship between political frames, intergroup contact and anti-immigrant attitudes. The implications of these findings are further discussed in the concluding section.

Concluding discussion

5.1 This study set out to test two hypotheses about the relationship between anti-immigrant attitudes, political circumstances and intergroup contact. More specifically, it examines whether nationalistic framing by political representatives has any bearing on anti-immigrant attitudes among native populations in Europe, and whether personal experiences of intergroup contact somehow moderate this relationship.

5.2 In accordance with the theoretical expectations, nationalistic frames emerge as significantly and positively related to anti-immigrant attitudes; the more political parties use frames that politicise the national way of life, the more negative are the attitudes towards newcomers in society. In terms of group threat theory, this indicates that political parties can contribute to a societal climate where feelings of threat and antipathies towards immigrants more easily emerge. By using certain kinds of frames they can influence how immigrant presence is interpreted and evaluated by the native population. Meanwhile, political elites are only one of many possible providers of frames with regard to immigrant presence, and the findings suggest that their influence on attitudes towards immigrants depend on natives' access to other - alternative - frames.

5.3 The analysis suggests that having many immigrant friends or colleagues significantly reduces the tendency to be influenced by nationalistic framing. This can be explained by the fact that natives involved in intergroup contact have greater access to alternative frames than those who lack these kinds of social ties. Contact generally implies better knowledge of the background and current situation of the group in question. In this sense, contact not only lowers peoples' inclination to hold anti-immigrant attitudes, as specified by intergroup contact theory, but also makes nationalistic political frames appear less attractive. Increasing awareness of what natives and immigrants have in common undermines political rhetoric that frames immigrants as alien and a threat to national culture. According to contact theory, intergroup contact also makes people less provincial (Pettigrew 1998) in the sense that they become less inclined to strongly identify with their own group and less convinced that their own group's ways of doing things are necessarily the best – with the consequence that they become less prejudicial. The findings in this study suggest that this effect may be important also with regard to contact as a moderator of frame effects. As discussed in the theoretical section, the heterogeneity of the immigrant populations in Europe raises questions of generalization as people might perceive discrepancies between the immigrants who are the focus of political frames, and the immigrants who are friends and colleagues. While many friends/colleagues of course can imply heterogeneous social networks (in terms of origin, ethnicity, language etc.) that bridge such discrepancies, it is not too far-fetched to assume that a waning identification with the nation also facilitates this kind of generalization.^[19] If people become less provincial, they are more likely to question frames that pit nationals and non-nationals, natives and immigrants against each other.

5.4 Taken together, the findings from the study confirm the contact hypothesis and add to this that contact can inoculate individuals against political influence. Given that the measurement applied to capture nationalistic frames is a proxy, and not a direct measurement of the framing in each country during the period of study, these findings emerge as all the more interesting. Although manifesto data can provide clues to what the parties deem important and where their focus lies, there is of course no perfect correlation between manifesto content and the frames the parties later communicate during their time in parliament. Other actors, such as media and other political parties, also have a say in what gets prioritized on the general political agenda. If issues of national belonging are given low priority, there will be less opportunity for the party to convey their message. The finding that nationalistic frames, measured in this indirect and general manner, nevertheless have a significant influence with regard to anti-immigrant attitudes, signals that politics indeed matter. Other interesting issues to pursue in future studies are the role of other political actors, media, and institutional arrangements as well as how intergroup contact and other individual features influence the effect of such political contextual factors. Also, can the frames provided by contact counteract any other contextual influences on anti-immigrant attitudes? McLaren (2003) and Schneider (2008) show that this is the case with regard to the size of the immigrant population, but what about the effect of colleagues for the implications of the level of unemployment? To gain a better understanding of anti-immigrant attitudes in this way is important; particularly as such attitudes are shown to impair the prospects of upholding social cohesion and welfare support in increasingly heterogeneous societies (Gilens 1995; Hjerm & Schnabel 2012).

Appendix 1

Notes

¹Goffman was the first to introduce frames as ways of making sense of the situation. Since then, frames and framing have developed into central themes for example in social movement studies (Benford & Snow 2000) and communication studies (Borah 2011).

²A 'framing effect' denotes the process by which political frames influence people's frames of thought (Druckman 2001a).

³While this study focuses on political frames, there are also other ways that political factors may influence anti-immigrant attitudes. For example, if issues of immigrant presence are highly salient on the national political arena, this entails a kind of priming (Weaver 2007) that is likely to increase immigrant visibility and thereby trigger feelings of threat among the native population. Also, as people who hold negative attitudes will be more likely to speak them out loud, as well as to advocate or act on their beliefs, if they resonate with views and interpretations conveyed in rhetorical frames, political actors may enhance anti-immigrant attitudes through legitimizing such views (cf. Dahlström & Sundell 2012). Further research is needed to closer establish the relative importance of these ways of influence in regard to anti-immigrant attitudes.

⁴Due to the specific status of immigrants in Israel, Israel has been left out from the study. For a further discussion, see Gorodzeisky (2011).

⁵0.731 in LU to 0.883 in UK

⁶This fourth alternative was given in most, but not all, countries (not in CZ, IT, PT). As this may have generated an overrepresentation of No, none at all answers in these countries, which potentially distorts the general findings, additional models were run without CZ, IT and PT. The alternative models, however, did reveal the same general pattern as the models with all countries included.

⁷A closer description of the coding process and illustrative examples are found in Klingemann et al. (2006).

⁸In the CMP data set, there is also a category which captures 'negative references to the national way of life'. Meanwhile, this category is primarily about regional autonomy and self-determination (visible, for example, in the kind of parties that score high on this category), which makes it less relevant to the study of anti-immigrant attitudes

⁹An alternative would be to use the share of seats in parliament, but such a measurement is more sensitive to the fact that different countries have different electoral systems. In this sense, it risks to exclude parties that, as in the case of Front National in France, have significant impact on the general political climate although the electoral rules limit their possibility to reach national representation. If one disregards France, however, weighting strategy has little bearing on the framing scores for the years 1990-2002, and the results from the models presented in Table 1 and 2 are the same regardless of size indicator.

¹⁰Descriptive data found in the appendix.

¹¹To attain a measurement that is on the same time horizon as the framing variable, the measurements are calculated as means for each country for the years 1990-2002. Both these measurements and measurements of economic conditions in 2002 are tested in the analysis.

¹²For a majority of the countries data is available for 2001. The exceptions are Germany and Slovenia, where data instead is taken from 2000 (DE) and 2002 (SI).

¹³As evident in Table A1, Greece displays the highest degree of anti-immigrant attitudes and has more than three times the framing level (86.61) compared to the country in second place (Austria: 27.34). Although a closer examination fall outside the scope of this article, it is interesting to note how supportive norms are deemed necessary for contact to reduce prejudice (Pettigrew 1998). The exceptionally hostile political climate may thereby explain why contact is not more effective in reducing anti-immigrant attitudes and the effect of nationalistic frames in Greece. More analyses, however, are necessary to gain a better understanding of, and draw any conclusions concerning, the case of Greece.

¹⁴This is also the case when the outlier Greece is excluded from the model.

¹⁵ For the purpose of presentation, only one model with country-level controls is displayed in Table 1. Results from the additional models are available from the author upon request. Besides the averaged measurements (1990-2002), GDP per capita and unemployment rates in 2002 were also tested as indicators of the economic conditions, with the same results.

¹⁶In the full sample, the effect is no longer significant when controlling for friends (Model 1).

¹⁷The moderating effect of colleagues decreases slightly, but is still significant when controlling for friends.

¹⁸Separate analyses show that while the effect of nationalistic frames is .31 (.08) in the group 'no friends', it is .15 (.09) in the group 'many friends'. For colleagues, the effect is .31 (.08) in the group 'no colleagues' and.10 (.08) in the group 'many colleagues'.

¹⁹This is further supported by how nationalistic frames are measured in this study. Political parties' manifesto articulation to politicize the national way of life focuses specifically on defining and establishing what is so good with one's own nation – frames that are likely to be less appealing to people with a more 'deprovincialized' identity.

²⁰1000 current international $ PPP, data from the World Bank.

²¹Data from OECD for 2001, except for Germany (OECD 2000) and Slovenia (OECD 2002).

²²Unemployment, total (% of total labor force), data from the World Bank.

References