Introduction
As the population composition in Europe increases in heterogeneity, immigrants’ trust in political institutions and feelings of national belonging in the receiving countries become essential for social cohesion (Miller and Ali, Reference Miller and Ali2014; Norris, Reference Norris, Zmerli and Van der Meer2017). In order to achieve political trust and national belonging, the encounters between immigrants and the host society play an important role. If immigrants experience belonging to a group that is treated unfairly, they may turn away from the political society and distrust the institutions that uphold discriminatory structures. In contrast, if immigrants experience that their rights are promoted and that they are valued in the society, they are more likely to feel included as members of the political society and trust the political institutions.
While these are plausible propositions, we still know surprisingly little about how interactions between immigrants and the host society influence their political trust and national belonging, since previous work largely centres on individual level explanations (e.g., Huddy and Khatib, Reference Huddy and Khatib2007) or prior exposure to democratic and stable institutions (e.g., Röder and Mühlau, Reference Röder and Mühlau2012; Wals and Rudolph, Reference Wals and Rudolph2019). In this study, I address two types of interactions by asking how perceptions of discrimination, on the one hand, and expressions of pro-immigrant support, on the other, influence non-Western immigrants. I bring new evidence by testing the causal effect of perceptions of discrimination and pro-immigrant support on political trust and national belonging, in a survey experiment in two national contexts. Footnote 1
Discrimination is one of the most negative forms of interaction between immigrants and public or institutional actors within their host society. Yet, causal evidence of the effect of discrimination on political trust and national belonging is scarce. Within the research field, observational correlation studies indicate a negative relationship between perceptions of discrimination, national belonging and political trust (e.g., Heath and Demireva, Reference Heath and Demireva2014; Maxwell, Reference Maxwell2009; Schildkraut, Reference Schildkraut2005). The lack of causal studies, however, maintains that the relationship, to some extent, remains a black box in the literature (see Jasinskaja-Lahti et al., Reference Jasinskaja-Lahti, Liebkind and Solheim2009 and Fouka, Reference Fouka2019 for exceptions). This is unsatisfactory as the causal relationship may be reversed, in that individuals who are less trusting, for instance, can perceive their surrounding environment as more unfair rather than the other way around. In addition, there may be a third, unobserved factor that explains both perceptions of the society, political trust and feelings of national belonging. In order to address these endogeneity issues and isolate potential causal effects of discrimination, additional experimental evidence is needed.
Whereas research on discrimination, political trust and national belonging is extensive but lacking in causal evidence, there are few studies on the effect of pro-immigrant support (see Just and Anderson, Reference Just and Anderson2014 and Bennour and Manatschal, Reference Bennour, Manatschal, Steiner and Wanner2019 for cross-sectional exceptions, and Van Hook et al., Reference Van Hook, Brown and Bean2006 for effects on naturalization). Expressions of pro-immigrant support are, in contrast to discrimination, a positive type of encounter between immigrants and public or institutional actors within the host society. Pro-immigrant support can foster immigrants’ political trust and feelings of national belonging by increasing the immigrant group’s value. The influence of pro-immigrant support has, however, largely been overlooked, since existing research mainly focuses on negative types of encounters (e.g., Heath and Demireva, Reference Heath and Demireva2014; Maxwell, Reference Maxwell2009; Pérez, Reference Pérez2015). We know significantly less about the positive interactions between immigrants and the host society and how they potentially shape immigrants’ political trust and feelings of national belonging.
In this study, I utilize targeted Facebook advertisements to recruit immigrant respondents with non-Western backgrounds. This is a novel research design, previously used to study political behaviour among other specific subgroups (Hirano et al., Reference Hirano, Lenz, Pinkovskiy and Snyder2015; Alrababa'h et al., Reference Alrababa'h, Marble, Mousa and Siegel2021). My study is, however, the first to use Facebook advertisements as a tool to recruit immigrant respondents. A large share of the respondents in my sample are newly arrived refugees, immigrating to Europe from Syria during the so-called refugee crisis. I hereby contribute with unique data on this main immigrant group arriving in Europe today. Within this group of individuals, interactions with the host society can have significant effects on future levels of political inclusion. Yet, it is a group that is heavily underrepresented in existing research since traditional surveys normally reach those who have lived longer in the host country.
Following standard experimental procedures to test the hypotheses, I attempt to prime perceptions of group discrimination by asking questions on unfair treatment. Expressions of pro-immigrant support are, in turn, primed with facts about public and institutional support for immigrants’ rights. My findings underline the importance of equal treatment to achieve social cohesion. However, the results also paint a rather complex picture of discrimination and its psychological impact. To some extent, they challenge the causal interpretations of negative correlations between perceived discrimination, political trust and national belonging in observational studies. The remainder of the paper will disentangle these conclusions, starting with an overview of the theoretical framework.
Theory and literature review
National belonging commonly refers to a subjective sense of attachment towards the national community (Huddy and Khatib, Reference Huddy and Khatib2007). In times where multiculturalism is increasing, theorists of liberal nationalism argue that national identities (or belonging) function as a ‘glue’ that holds culturally diverse societies together (Miller and Ali, Reference Miller and Ali2014). Societies where both natives and immigrants feel that they belong to the nation are thereby better equipped to handle potential problems following diversity, since feelings of belonging provides an overarching and uniting shared identity. In contrast, if immigrants do not feel included in the political society, polarization between groups is likely to increase. National belonging is, in other words, a significant predictor for social cohesion of the political society.
Political trust is, in a similar way, a significant predictor for satisfaction with the democratic system. If people distrust the political institutions, this indicates dissatisfaction for the society at large (Marien, Reference Marien, Zmerli and Hooghe2011; Norris, Reference Norris, Zmerli and Van der Meer2017). Trust in political institutions is more general than trust in political actors. If people distrust political actors, they can simply vote them out in the next election. People should, however, be able to trust political institutions since these are generally considered as the basic pillars of society (Marien, Reference Marien, Zmerli and Hooghe2011; Norris, Reference Norris, Zmerli and Van der Meer2017). Distrust in political institutions thereby indicates dissatisfaction for the society at large. In addition, scholars argue that trust in political institutions influence other individual attitudes towards the political society, such as the willingness to accept and comply with political decisions (Levi and Stoker, Reference Levi and Stoker2000). Political trust is therefore essential in itself and also has important implications for other political attitudes.
Trust in political institutions and feelings of national belonging are the type of attitudes that generally develop during adolescence and can be quite stable, but they are not fixed. This is especially the case for immigrants, who are likely to update their preferences based on experiences of migration and exposure to the new host society (Maxwell, Reference Maxwell2010; Röder and Mühlau, Reference Röder and Mühlau2012, see also Dinesen, Reference Dinesen2012 for similar findings on social trust).
Within the scholarly field, there are two major types of predictors used to explain political behaviour and attitudes among immigrants. The first type relates to individual-level explanations, where empirical findings show that factors such as time spent in the host society (Huddy and Khatib, Reference Huddy and Khatib2007), citizenship (Leszczensky et al., Reference Leszczensky, Maxwell and Bleich2019) and level of education (Norris and Puranen, Reference Norris and Puranen2019) play significant roles. The second type focuses on contextual factors. These include institutional conditions such as legal access to citizenship and voting rights (Ferwerda et al., Reference Ferwerda, Finseraas and Bergh2020; Hainmueller et al., Reference Hainmueller, Hangartner and Pietrantuono2015; Koopmans, Reference Koopmans2004, Reference Koopmans2005), prior exposure to democracy (Wals and Rudolph, Reference Wals and Rudolph2019) and the surrounding environment of the host society (Just and Anderson, Reference Just and Anderson2014; Pérez, Reference Pérez2015). My study builds on the latter nascent literature on contextual explanations, focusing specifically on immigrants’ encounters with discrimination and pro-immigrant support in the host society environment.
Discrimination
Discrimination is one of the most negative forms of interaction between immigrants and the host society. Discrimination, or unfair treatment, can be perpetrated in a multitude of ways by individuals or institutions. It is a direct type of encounter with anti-immigrant attitudes that can have consequences on the way immigrants view the political society and their position within it. It is thereby a contextual factor that shapes interactions between immigrants and the host society (Hopkins et al., Reference Hopkins, Kaiser, Pérez, Hagá, Ramos and Zárate2018). This is especially the case for non-Western immigrants, who are generally more exposed to discrimination than other immigrant groups (Hainmueller and Hangartner, Reference Hainmueller and Hangartner2013; Hangartner et al., Reference Hangartner, Kopp and Siegenthaler2021).
The individual response to discrimination can be understood in the light of social identity theories and realistic interest theories, which, respectively, centre on a symbolic or material threat toward the individual and/or social group. The notion of a symbolic threat originates from social identity theory (Tajfel and Turner, Reference Tajfel and Turner1979) and its off-shoot, self-categorization theory (Turner et al., Reference Turner, Hogg, Oakes, Reicher and Wetherell1987). According to these perspectives, individuals strive for a positive social identity. This can be achieved by favourable comparisons between one’s own in-group and a relevant out-group. A threat towards the social identity of a group is thereby symbolic in the sense that it challenges the social status of the group in comparison to other groups.
The mechanism related to realistic interest theories, in turn, centres on protecting shared material interests (Bobo, Reference Bobo1983; Huddy, Reference Huddy, Huddy, Sears and Levy2013). Such migration-related interests include, for instance, access to public welfare or employment, where discrimination can provide distinct threats to the individual or group material interests. Discrimination is thereby a form of interaction between immigrants and the host society that fits the framework of social identity theories and realistic interest theories. It relates to the value of an individual’s own group, but also has direct economic implications, constituting both a symbolic and material threat. In addition, experiences of discrimination can induce a sense that society is unjust (Schaafsma, Reference Schaafsma2013; Stroebe et al., Reference Stroebe, Dovidio, Barreto, Ellemers and John2011). If immigrants perceive the government as unable to ensure equal status between majority and minority groups, this is likely to decrease political trust (Michelson, Reference Michelson2003).
The effect of discrimination can have different directions depending on individual-level explanations or the type of discrimination one faces. Drawing further on social identity theories, an individual can choose to leave their social group for another when the value of the group is threatened, but the possibility to do so and thus strive for a more highly valued group membership depends, to some extent, on external labelling; you may have difficulties leaving the group if others perceive you as being part of the group, for instance, due to skin colour or cultural practices (Huddy, Reference Huddy, Huddy, Sears and Levy2013; Tajfel and Turner, Reference Tajfel and Turner1979). Consequently, a symbolic threat could theoretically increase feelings of national belonging among some immigrant groups as a way of signalling assimilation with the native majority identity, but decrease feelings of national belonging to the host community among those immigrants where appearance or cultural background is more distinct from the native majority, since the possibility to leave the targeted group is limited.
Previous research has also found the type of discrimination to matter for the direction of effect, at least when it comes to other forms of political behaviour, such as political engagement (Jones-Correa, Reference Jones-Correa2001; Okamoto and Ebert, Reference Okamoto and Ebert2010; Pantoja et al., Reference Pantoja, Ramirez and Segura2001). By threatening the symbolic or material value of the immigrant group, individuals can react by either withdrawing from the society or engaging to raise the group status or secure material interests (Ellemers et al., Reference Ellemers, Spears and Doosje2002; Simonsen, Reference Simonsen2020). According to Oskooii (Reference Oskooii2016, Reference Oskooii2020), political mobilisation is most likely to occur when the threat is political, for instance through discriminatory laws and campaign messages. Societal threats, where people are discriminated in public or private settings by other individuals, are assumed to have a demobilising influence on political engagement.
In this study, I centre primarily on societal discrimination in the surrounding host societies. In line with the theoretical assumptions raised above, and empirical findings in previous observational correlation studies, I expect perceptions of discrimination to have a causal negative effect on immigrants’ sense of belonging to the national community and trust in political institutions. Thus, I anticipate that:
H1: Perceptions of discrimination decrease political trust and national belonging.
Pro-immigrant support
Taking the positive aspect of encounters into account, I propose that it is relevant to also examine the effect of pro-immigrant support among the native majority. Pro-immigrant support is often generally expressed in highlighting the cultural or material benefits of immigration more broadly by political actors or within the general public, without specifying support for a certain immigrant group. In contrast to the established research field on discrimination, considerably less theoretical and empirical attention has been paid to how pro-immigrant support may shape immigrants’ political integration, especially related to expressions of support within the general public (but see Ager and Strang, Reference Ager and Strang2008 for qualitative findings on neighbourhood support, and Koopmans, Reference Koopmans2005 for research on more formal types of support).
Pro-immigrant expressions can relate to material interest on both the individual and group levels, but the social identity theory may be particularly relevant in explaining the causal mechanisms. Building on social identity theories, individuals are expected to prefer being socially accepted and valued (Verkuyten and Martinovic, Reference Verkuyten and Martinovic2012). This is, as mentioned, a way to gain a positive social identity. In this sense, pro-immigrant support signals a form of favourable comparisons that increases the value of a social identity.
In line with the nuance raised above related to a symbolic threat, a signal of symbolic inclusion could theoretically have different outcomes for political trust and feelings of national belonging. An elevated status of immigrant groups may enable increased identification with the migrant group, which could potentially, but not necessarily, result in decreased identification with the national identity. In contrast, an elevated status of immigrant groups can induce a sense of belonging with society at large and thereby increase national identification and political trust. Hence, while decreased national identification is possible, it is more likely that a signal of symbolic inclusion increases political trust and national belonging, especially in contexts where immigrants are able to identify both with the national and the migrant identity. Related to material interests, the nature of the effect is more straightforward, since pro-immigrant support can provide material gains that benefit interests on both the individual and group levels. For instance, this can occur through affirmative action aiming to improve representation of immigrant groups on the labour market.
In understanding the causal mechanisms further, research related to discrimination and psychological well-being show that pervasive discrimination can be countered by re-establishing the world as just. This is based on the assumption that individuals have a fundamental need to believe that people generally get what they deserve. One way to reinforce the world as just is to communicate to those who are disadvantaged that they are valued within the society, rendering some sense of belief that a just world can be achieved even though there is discrimination in the present state (Stroebe et al., Reference Stroebe, Barreto and Ellemers2010; Stroebe et al., Reference Stroebe, Dovidio, Barreto, Ellemers and John2011). This form of communication can be reflected in pro-immigrant expressions or actions that raise the immigrant group’s value. By increasing the symbolic or material value of the immigrant group, I therefore propose that pro-immigrant support has the potential to increase immigrants’ political trust and feelings of national belonging, hypothesizing that:
H2: Expressions of pro-immigrant support increase political trust and national belonging.
Research design
I test the hypotheses by fielding an online survey experiment in Sweden and Germany. These are two of the European countries with the largest immigrant populations, constituting roughly 15–20 percent of the total population of each country. Footnote 2 In both countries, immigrants with non-Western backgrounds make up a large share of the immigrant population (Statistics Sweden, 2021; World Atlas, 2019). In addition, in terms of the opinion climate towards immigrants, Sweden and Germany are examples of countries where we can see discriminatory expressions towards immigrants as well as pro-immigrant support in the societal and political arenas. For instance, both countries have parties in the parliament that are outspokenly anti-immigrant (i.e., the Sweden Democrats and the Alternative for Germany). There is also, in general, support for immigration and immigrants’ rights. This can be illustrated by survey findings from Pew Research Center (2019), where more than 60 percent of respondents in Sweden and Germany say that immigrants make their country stronger.
As discussed by Koopmans (Reference Koopmans2004), Sweden and Germany have historically had different citizenship and integration regimes (i.e., different political opportunity structures for immigrants), and Sweden is generally regarded as more inclusive than Germany (MIPEX, 2020). In terms of immigration, however, both Sweden and Germany distinguished themselves from other European countries during the refugee crisis in 2015, receiving the largest number of refugees relative to population size (Hagelund, Reference Hagelund2020). The countries are thereby fairly similar in terms of the share of immigrants and attitudes towards immigration. Moreover, they stand out as two of the European countries where the issue of social cohesion is particularly salient considering the large refugee reception in recent years. By testing the hypotheses in both countries, I increase the external validity of the experiment. In addition, the fact that we can find aspects of both discriminatory and pro-immigrant expressions in the countries included, increases the validity of the experimental manipulations.
Proceedings
Survey participants were recruited through Facebook advertisements. This approach has been used previously in studies on political behaviour among other specific subgroups (Hirano et al., Reference Hirano, Lenz, Pinkovskiy and Snyder2015; Alrababa'h et al., Reference Alrababa'h, Marble, Mousa and Siegel2021), and survey experiments distributed through Facebook have produced similar results as in national representative samples (Samuels and Zucco Jr, Reference Samuels and Zucco2014). My study is, however, the first to use Facebook advertisements as a tool to recruit immigrant respondents, targeting the advertisements to users with the Arabic language setting. Section A in the online appendix covers a more detailed description of the survey distributions.
To begin the survey, respondents completed a number of pre-treatment questions gauging their gender, age, political interest, region of origin, years lived in host society, and citizenship status. For the analyses, I limit the sample to include only non-Western born. The studies yielded 952 responses in Sweden and 947 responses in Germany for this specific category of respondents. In terms of survey completion, more than 90% of the respondents who were treated completed the survey. Tables A1 and A2 in the appendix reports some basic descriptive statistics from each study.
As a way to contrast my sample with that of immigrants in traditional surveys, Table A3 in the appendix includes comparable statistics with non-Western born respondents from the European Social Survey (2020). The European Social Survey (ESS) is widely known for its high standards in survey design. As the table shows, the samples from the ESS and Facebook surveys resemble each other on many key traits. This indicates that the respondents from my surveys are not atypical, apart from the integration levels; the share of respondents with citizenship in the new country is lower in the Facebook sample than the ESS. My sampling strategy thereby enabled attracting a group that is not represented in traditional surveys, namely those who are newly arrived and less integrated. It is also relevant to note, in addition to the statistics shown, that a large share of the sample immigrated from Syria during the refugee crisis in 2014–2016. The implications of this, and how the composition of respondents may affect the results, is discussed further in the conclusion. For a full list of origins included in the sample, see section B in the appendix.
Treatments
Following the introductory questions, respondents were randomly assigned to one of the following three conditions: discrimination, pro-immigrant support, or control. When the causal effect of discrimination has been tested on psychological outcomes (although other than political trust and national belonging), this has been in laboratory settings (e.g., Kuo et al., Reference Kuo, Malhotra and Mo2017 and Taylor et al., Reference Taylor, Wright and Ruggiero1991). In this study, I increase external validity by testing the causal effect outside the lab. This means I cannot manipulate actual discrimination. Instead, I remind the respondents of perceptions of discrimination by using priming questions. This is in line with prior experimental studies on other issues, where findings show that reminders of past experiences works as a powerful treatment (e.g., Sprecher, Reference Sprecher2018; Williamson et al., Reference Williamson, Adida, Lo, Platas, Prather and Werfel2021).
Respondents were primed by answering questions on unfair treatment towards people with Arabic backgrounds, with which they can agree or disagree. Footnote 3 For instance, whether people with Arabic backgrounds are unfairly denied jobs or are victims of Islamophobia. The priming thereby relies on group discrimination, rather than discrimination towards the respondents themselves, capturing a more contextual component. Perceptions of discrimination are, in addition, more frequently expressed at the group level than at the individual level, known as the personal-group discrepancy (Schildkraut, Reference Schildkraut2005). Respondents are therefore more likely to agree with questions on group level discrimination than individual level discrimination, thereby priming perceptions of discrimination within a broader group of individuals. The treatments vary slightly between the two countries since the questions on discrimination include those statements that respondents in each country agreed to the most in pilot studies. It is important to note, in relation to this, that the treatment focuses on perceptions, which may or may not correspond to objective reality.
For the main analyses, all respondents assigned to the discrimination treatment are considered as treated, regardless of their response to the discrimination questions. In other words, I make an initial assumption that most respondents have perceived discrimination towards Arabic immigrants, and that the questions will prime these perceptions. This is a reasonable assumption, since prior work indicates that non-Western immigrants are frequently exposed to discrimination (e.g., Hainmueller and Hangartner, Reference Hainmueller and Hangartner2013; Hangartner et al., Reference Hangartner, Kopp and Siegenthaler2021, see also Lajevardi and Oskooii, Reference Lajevardi and Oskooii2018 on anti-Muslim attitudes). Whether people in fact have perceived discrimination against those of Arabic origin or not, and the individual’s reaction to these events, are, however, outside of my control. For an opportunity to compare between people with or without perceptions of discrimination, respondents assigned to the other conditions were also shown the discrimination questions at the very end of the survey (without possibility to back-track). This simple approach enables an additional comparison between subgroups, including those who agree with the discrimination statements and are asked before the outcome questions (treated) and those who agree but are asked after the outcome questions (untreated). I hereby follow a similar procedure as in previous correlational studies, but handle the endogeneity problem by exogenously inducing perceptions of discrimination.
To test the effect of pro-immigrant support, I prime pro-immigrant support by displaying facts about positive attitudes towards immigration and institutional support for immigrants. This treatment does not depend on past experiences but rather provides new information that signals material gains and symbolic value. Footnote 4 The facts are based on real public opinion towards immigrants and institutional support gathered from various sources, including the Pew Research Center (2019) and the ESS. The opinion on these facts was tested in a pilot study, where a vast majority of the respondents favoured the statements included. For the experiment, respondents were asked to reflect on the information by answering whether they were aware of each fact in order to increase the impact. The facts were adapted slightly to reflect the situation within the specific country, taking into consideration the fact that the legal climate differs between the two countries included. The phrasing for the treatments is presented and discussed further under section C in the appendix.
Outcomes
Respondents were asked questions post-treatment to gauge their level of political trust and feelings of national belonging. For political trust, I used standard questions from the ESS commonly applied in previous work (e.g., Marien, Reference Marien, Zmerli and Hooghe2011 and Maxwell, Reference Maxwell2013.) The respondents were asked how much they trust the host country’s parliament, the legal system and the police, where 1 = complete distrust and 5 = complete trust. The respondent’s level of national belonging was captured with three questions: (i) How connected do you feel with [host country]? (1 = I do not feel a connection at all, 5 = I feel a very close connection), (ii) Do you feel at home in [host country]? (1 = No, not at all, 5 = Yes, completely) and (iii) Are you proud to live in [host country]? (1 = No, not proud at all, 5 = Yes, very proud). Footnote 5 For the analyses, the questions were combined into two indices of national belonging and political trust, with values ranging from 1 to 13. Footnote 6
Econometric design
In order to increase statistical precision, I aggregate the two experiments from Sweden and Germany into a single data set. This enables me to make use of the full available information, analysing the data jointly while controlling for country. Footnote 7 I test the hypotheses using OLS regressions, estimated with the following regression model for the main analyses:
$${y_{ic}} = {\alpha _0} + \sum {\delta _j}(Treatmen{t_{ic}} = j) + {X'_{ic}}\beta + {\theta _c} + {\varepsilon _{ic}}$$
where
${y_{ic}}$
represents the level of political trust or national belonging for respondent i in countryc. The three different treatments are captured by the coefficient vector
${\delta _j}$
,
${X'_{ic}}$
is a matrix of covariates (pre-treatment questions),
${\theta _c}$
is a country fixed effects, and
${\varepsilon _{ic}}$
is the error term. Respondents assigned to the control group are the reference category in terms of average levels of political trust and national belonging. Comparing this group to the treatment groups, I expected respondents assigned to the discrimination treatment to report lower levels of political trust and national belonging (H1). In contrast, I expected respondents assigned to the pro-immigrant treatment to express higher levels of trust and belonging (H2) than the control group.
Results
As a first step to test the hypothesis of discrimination (H1), I examine the survey results cross-sectionally to see whether the negative correlations between perceptions of discrimination, political trust and national belonging established in prior work is present also in my sample. Since all respondents answer the questions on discrimination during some stage of the survey, I am able to examine the correlation using the total sample of respondents. This means that perceptions of discrimination are not exogenously induced here for the majority of respondents, as I analyse the correlation in all treatment groups.
For measuring discrimination as the independent variable, I add the three statements on discrimination into an index where 0 = no discrimination and 12 = high discrimination. Results from these OLS models are displayed under section E in the appendix, showing consistent negative and significant correlations between the discrimination index, political trust and national belonging. These results suggest that immigrants who perceive the Arabic group being treated unfairly are less trusting and feel less belonging to the national community, in line with what we can expect from previous correlation studies.
To examine further whether there is a causal effect of discrimination in addition to correlation, the next step is to continue with the experimental analyses. For these analyses, I compare the group of respondents where perceptions of discrimination are randomly induced with respondents in the control group. I expect that respondents primed with the discrimination questions before the outcome questions feel significantly less trust and belonging than the control group. As mentioned, I test the causal effect of perceptions of discrimination in two ways. First, by estimating the effect of the discrimination prime in the treatment group and comparing them with the control group. Second, by comparing the treatment effect in subgroups with respondents who agree or disagree to the discrimination statements in treatment and control.
Results from the first analyses are displayed in Table 1. Footnote 8 As the table shows, the results go in an opposite way than I expected. Respondents assigned to the discrimination treatment report higher levels of political trust and national belonging in comparison to the control group. These positive effects of the discrimination prime indicate that thinking about discrimination increases immigrants’ feeling of belonging to the national community and trust in political institutions. This is in contrast to the established findings in previous research, and the correlational results presented above.
Effect of discrimination on political trust and national belonging
Note: *P < 0.05; **P < 0.01; ***P < 0.001. Robust standard errors in parentheses.
What explains this discrepancy from previous findings? Consider the character of my treatment. I assumed, in line with previous research, that most respondents have had negative encounters with discrimination, and that these would be the encounters that shape their perceptions of the host society. But the treatment can also work in the opposite way and prime positive experiences among those who do not perceive discrimination against those of Arabic descent. If the treatment works as a reminder of non-discrimination, the results follow the expectation that priming positive perceptions would increase political trust and national belonging by the same logic as the hypothesis for the pro-immigrant treatment, increasing material and symbolic values.
In order to examine if this indeed explains the surprising findings, and how respondents react to the prime depending on their perceptions of discrimination, I test the effect of the discrimination prime by dividing the treatment group into two subgroups: those who have perceptions of discrimination and those who do not. Respondents are coded as having perceptions of discrimination if they answer that they ‘strongly agree’ or ‘agree’ to the discrimination statements, and not having perceptions if they answer ‘strongly disagree’ or ‘disagree’. Footnote 9 For the sake of simplicity, I henceforth refer to these groups as being primed with discrimination if they agree, and non-discrimination if they disagree. For more descriptive information about these subgroups, see section H in the appendix.
Results from the subgroup analyses are shown in Table 2, depicting the effects of discrimination on national belonging and political trust in Models 1 and 3, and non-discrimination in Models 2 and 4. We see a clear difference in effects here between the two groups depending on perceptions of discrimination: those in the treatment condition who perceive that the Arabic group is not discriminated against have significantly higher political trust than respondents with the same perceptions in the control condition. Similarly, the effect on national belonging is positive albeit being just below standard levels of statistical significance (P = 0.07). This shows that the surprising result we saw earlier was indeed an effect of respondents being primed with non-discrimination.
Effect of discrimination and non-discrimination on political trust and national belonging
Note: *P < 0.05; **P < 0.01; ***P < 0.001. Robust standard errors in parentheses.
The effect size is quite large among the group with perceptions of non-discrimination, equalling roughly .5 of a standard deviation increase in political trust and .3 in national belonging. Interpreted as an inverse of the hypothesized effect, the findings suggest that reminders of equal treatment fosters political trust and national belonging. My experimental design did not aim to capture this, but the results are in line with the theoretical arguments related to pro-immigrant support. In a symbolic sense, equal treatment raises the social status of the group. From a materialistic perspective, it provides potential material benefits.
Against my expectations, I find no effect of the discrimination prime among those respondents who perceive discrimination against those of Arabic background. In other words, in comparison to the control group, I find a positive effect of the discrimination prime among individuals who perceive that people with an Arabic background are treated equally in the society, but no effect among those who perceive their discrimination. Based on previous correlational studies, this is where I expected the most pronounced effects of the treatment. It may be, however, that the main analysis hides differences within the group depending on time spent in the host country. Previous work indicates that encounters with the host society can influence political trust and national belonging in different ways depending on where the individual is in the migration process. I will return to this potential explanation shortly.
Turning to the effect of the pro-immigrant support treatment, I expected respondents assigned to this condition to display higher levels of political trust and national belonging than respondents in the control group (H2). As shown in Table I1 in the appendix, however, there is no main effect of pro-immigrant support on respondents’ national belonging or political trust when comparing treatment and control. Showing facts about attitudinal and institutional support for immigrants’ rights thereby did not have the expected effect on trust and belonging in the treatment group.
Conditional effects of time spent in the host country
In addition to the hypothesized main effects, there are reasons to believe that the effects of discrimination and pro-immigrant support are heterogeneous depending on the individual’s amount of time spent in the host country. Immigrants who have lived in the host society longer may react either more strongly or more weakly to the encounters than those who are newly arrived. There are, in other words, two contrasting ways of thinking of the heterogeneous effects.
Drawing on the assimilation theory, immigrants who have lived in the host society for many years are expected to have similar terms of identity and socio-political attitudes as natives (Alba and Nee, Reference Alba and Nee1997; Michelson, Reference Michelson2003). Immigrants who have lived longer in the country may therefore react less to discrimination, since they identify more with the native majority. The effect of discrimination is instead likely to be most pronounced among immigrants who are recently arrived. In contrast to this perspective, the paradox of social integration (Heath and Demireva, Reference Heath and Demireva2014; Michelson, Reference Michelson2003; Verkuyten, Reference Verkuyten2016; Platt, Reference Platt2014, see also Lajevardi et al., Reference Lajevardi, Oskooii, Walker and Westfall2020) proposes that second-generation immigrants are more sensitive to discriminatory structures. Since this group of immigrants have had more time in the country, they compare their situation with natives and thereby react stronger to unfair treatment than the newly arrived. From this perspective, we can expect a stronger effect of discrimination among second generation immigrants or immigrants who have spent more time in the country.
On the pro-immigrant side, the conditional effect is more unclear. According to Bennour and Manatschal (Reference Bennour, Manatschal, Steiner and Wanner2019), the assimilation theory indicates that pro-immigrant support will have a stronger influence among the more integrated. Acting as a catalyst, they suggest that pro-immigrant support can amplify the positive effect of time spent in the host society. To explore for such potential conditional effects, I test whether the effect of discrimination and pro-immigrant support vary depending on years spent in the host society. Footnote 10
Starting with the conditional effect of discrimination, I continue only with the subgroup of respondents in treatment and control who perceive discrimination towards the Arabic group. This is in order to see whether there is an effect of discrimination that is conditional on the years in the host society among those with perceptions of discrimination. Figures 1a and 2b illustrate the marginal effects of discrimination by the respondents’ time in the host country. Footnote 11 As shown, there are conditional effects of discrimination on national belonging and political trust depending on years in the host country. For immigrants who are recently arrived, we see the anticipated negative effect of discrimination on both outcomes. This finding is in line with what we can expect from the assimilation theory, in that those who have spent less time in the country are more negatively affected by perceptions of discrimination.
Marginal effects of discrimination by year in country (with 95% confidence intervals and frequency distribution).
Marginal effects of proimmigrant support by years in country (with 95% confidence intervals and frequency distribution).
For immigrants who have lived longer in the country, the result goes against expectations, indicating a positive effect of discrimination on trust and belonging. The shift begins around five years, where the effect of discrimination shifts from negative to positive. This result points to a more complex picture of discrimination and its political effects than anticipated from most findings in existing research and is not in line with the assimilation theory nor the paradox of social integration. If trust and belonging followed the trajectory of the assimilation theory, the negative effect of discrimination would simply decrease as time in the host country increased. If the results were in line with the paradox of social integration, the negative effect would be most pronounced among those who have lived in the country longer. This sample of respondents does not demonstrate either of these outcomes. The implications and potential interpretations of these findings are discussed further in the concluding section.
Moving forward, I continue by testing for potential heterogeneous effects of pro-immigrant support, examining whether there is an effect of pro-immigrant support that is conditional on time spent in the host country. Figures 2a and 2b displays the marginal effect of pro-immigrant support by years in country. Footnote 12 As before, we see that there is a conditional treatment effect on national belonging (but not political trust) that was hidden in the main analyses. The results here suggest that immigrants who have lived longer in the host society are more positively influenced by the pro-immigrant prime than those who are newly arrived in the country. Among the newly arrived, the positive effect on national belonging is absent. These results are in line with those of Bennour and Manatschal (Reference Bennour, Manatschal, Steiner and Wanner2019), where pro-immigrant integration policies were shown to influence national belonging only among more integrated immigrants.
Additional tests were conducted in order to test the robustness of the experimental result, including analyses of a placebo outcome, alternative versions of the outcome variable and a manipulation check. The first two analyses support the findings reported above. The manipulation check failed its main purpose but provided other useful information that indicates the respondents answered the survey coherently. For further discussion and tables of the robustness tests, see section K in the appendix.
Discussion and conclusions
Drawing on social identity theory and realistic interest theories, I have examined how interactions with the host society shape non-Western immigrants’ sense of national belonging and trust in political institutions. I tested the effect of perceptions of discrimination, on the one hand, and expressions of pro-immigrant support, on the other, with a survey experiment in two national contexts. By utilizing Facebook advertisements targeted to people with Arabic backgrounds, I was able to contribute with unique data on a group that is notoriously difficult to reach.
The experiment yielded important but surprising results. Overall, the findings indicate that the relationship between discrimination, political trust and national belonging is more complex than commonly perceived. In line with standard expectations from previous studies, I find negative correlations between discrimination, political trust and national belonging when examining the data cross-sectionally. However, when testing the causal effect experimentally, I only find the hypothesized negative effect of perceptions of discrimination among newly arrived immigrants. This finding corroborates expectations based on the assimilation theory (Alba and Nee, Reference Alba and Nee1997; Michelson, Reference Michelson2003), where discrimination is anticipated to mainly influence those who have spent less time in the host country.
Among respondents that did not agree with the statements about unfair treatment, however, the treatment has an unexpected positive effect on political trust and national belonging by priming perceptions of non-discrimination. While the experimental design did not aim to test this, the result is reasonable and in line with the hypothesis related to the pro-immigrant treatment. When non-Western immigrants are treated as equal, it indicates that they belong to a group that is valued in the society and that they can trust the political institutions. I underestimated such perceptions of non-discrimination by assuming, in line with previous research, that most respondents have had encounters with discrimination, and that these would be the encounters that shape their perceptions of the host society. While the group with perceptions of discrimination is indeed larger than those without, it is the latter group that are primarily affected by the prime.
In a third subgroup including immigrants who have lived longer in the country, the results indicate a positive effect on political trust and national belonging even when the prime works as the expected reminder of perceptions of discrimination. This positive effect is somewhat puzzling, but could relate to the type of discrimination studied here, which centres on discrimination towards the Arabic group rather than discrimination towards the individuals themselves. Findings by Bourguignon et al. (Reference Bourguignon, Seron, Yzerbyt and Herman2006) indicate that group-level discrimination can increase individual well-being by allowing people to believe that they are not alone in their plight. Group discrimination thereby provides a feeling of togetherness. Applied to the setting of this study, those who have spent more time in the country are more likely to have established connections with other immigrants from the Arabic group in the host society than immigrants who are newly arrived. Among immigrants who have lived longer in the country, group discrimination may therefore induce a sense of togetherness and positive emotions. Within the newly arrived immigrants, however, such positive feelings are likely to be absent.
The conditional effect could also relate to the distinction in discrimination raised by Oskooii (Reference Oskooii2016, Reference Oskooii2020) in terms of political or societal discrimination. Immigrants who have lived in the country longer may perceive the discrimination prime differently than those who are more newly arrived. The prime primarily captures societal discrimination, expected to decrease political engagement. Those who have lived longer in the country might, however, also take the political situation into account which may be perceived as more fair, thus not necessarily reflecting the perceptions of discrimination at the societal level.
As a positive encounter between immigrants and the host society, I tested the effect of pro-immigrant support by showing the respondents facts about public and institutional support for immigrants’ rights. While I found no main effect here, further analyses point to positive effects of pro-immigrant support on national belonging that are conditional on the amount of time in the host society; I find a positive effect of pro-immigrant support that is only present among immigrants who have lived in the country longer, with the feeling of national belonging increasing as time in the host country increases. This result is in line with the correlational findings by Bennour and Manatschal (Reference Bennour, Manatschal, Steiner and Wanner2019), where pro-immigrant integration policies were shown to influence national belonging only among more integrated immigrants. As the scholars suggest, this can be interpreted as support for the assimilation theory, where pro-immigrant support amplifies the positive effect of time spent in the host country.
Substantially, the results from this study can be viewed with some optimism, since the perceptions of non-discrimination have positive consequences on political trust and feelings of national belonging. Moreover, my findings indicate that non-discrimination matters more than explicit expressions of pro-immigrant support. A possible distinction is that non-discrimination represents a status quo where immigrants and natives are treated equally. This is potentially more important for the sense of inclusion and trust than expressions that specifically attempt to raise the value of the immigrant group. Or, it may be that it is more difficult to prime pro-immigrant support on the societal level than perceptions of non-discrimination on the personal group level. An important future avenue of research is to look further into the impact of these different types of positive interactions.
The effects shown in this study can be expected to be more pronounced among those who also identify more strongly with the Arabic group. Drawing further on social identity theory, the individual response to discrimination is likely to vary depending on the level of commitment to the identity of the targeted group. Those who are more strongly committed to the identity of the Arabic group are more likely to react negatively to the reminder of group discrimination than those with low commitment (see Pérez, Reference Pérez2015 for similar finding on Latino residents). Additional studies are needed to test such conditional effects further. In doing so, an important step forward is to randomly manipulate the levels of group identification, since there are a number of potential unobserved confounders that can relate to both levels of group identification, political trust and national belonging.
The limitations regarding estimating conditional effects should also be acknowledged in relation to an immigrant’s time spent in the country, since this aspect could be associated with other immigrant characteristics. I control for some of the standard variables in the analyses, but the experimental design does not handle the fact that there may be unobserved factors related to the length of time. While the approach to interact an experimental treatment with variables that have not been randomly distributed is used also in other studies (see e.g., Pérez, Reference Pérez2015), the conditional effect of time spent in the country should be explored further in future work in order to ensure it is not driven by unobserved confounders.
As a final point, it is important to note how the composition of respondents in the sample may influence our understanding of the results, as well as the societal implications of the findings. First, the sample presumably includes a large share of recent refugees, since many of the respondents arrived to Sweden and Germany from Syria following the civil war. Experiences related to flight can shape attitudes towards the host society in various ways. For instance, a recent study by Hall and Werner (Reference Hall and Werner2022) shows that experiences of trauma shape refugees’ trust in institutions differently depending on the type of experience, indicating both negative and positive effects on institutional trust within the receiving country. There are also effects related to the asylum process in the receiving country, where Esaiasson et al. (Reference Esaiasson, Lajevardi and Sohlberg2022) show that the migration decision affects asylum seekers’ perceptions of the host country, with positive effects of acceptance and negative effects of rejection. For newly arrived refugees, such recent experiences are likely to influence whether the host society is perceived as discriminatory and the effects of such perceptions.
Among those who do perceive the society as discriminatory, the effects on political trust and national belonging may also be stronger among refugees than voluntary migrants since there are less positive interactions with the host society to counteract them. Findings by Phillimore (Reference Phillimore2011) show that refugees who were exposed to societal forms of discrimination in the UK lacked confidence to seek contact with local people. In an already vulnerable position, discrimination thereby risks excluding refugees further from the host society.
Second, the sample is likely to (mainly) include Muslim immigrants. Within this group, discrimination has been shown to have different effects on political behaviour depending on the type of discrimination (ethnic or religious) (Ysseldyk et al., Reference Ysseldyk, Talebi, Matheson, Bloemraad and Anisman2014) and the type of religious practices (‘strict’ or ‘moderate’) (Baysu and Swyngedouw, Reference Baysu and Swyngedouw2020). These are important nuances that I do not capture within this study and that should be explored further.
Lastly, my findings come from a sample of respondents where some arrived in Europe as recently as last year. Previous research suggests that the period after arrival represents an ‘integration window’. During this window of time, initial experiences in the host country can have long-term effects on integration (Ferwerda et al., Reference Ferwerda, Finseraas and Bergh2020; Hainmueller et al., Reference Hainmueller, Hangartner and Lawrence2016). Equal treatment towards immigrants during the early migration process is thereby key, since this is where their first encounters with the host society are likely to take place.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1755773923000139.
Acknowledgements
For helpful comments and suggestions, I thank the three anonymous reviewers, Peter Esaiasson, Frida Boräng, Linna Martén, Kristina Bakkær Simonsen, Lene Aarøe, Rahsaan Maxwell, Kyle Marquardt, the Immigration Policy Lab Zürich, and participants at the Gothenburg Research Group on Elections, Public Opinion and Political Behavior internal conference and the Polsek seminar of Uppsala University.
Open access
