Abstract
A number of studies have established a strong link between anti-immigration and Eurosceptic attitudes. But does this relationship necessarily imply that more immigration would increase public Euroscepticism in member states of the European Union? I evaluate this question by analyzing immigration data and Eurobarometer survey data over the period 2009–2017. The analysis shows no evidence that individual levels of Euroscepticism increase with actual levels of immigration. This result suggests that a strong link between anti-immigration and Eurosceptic attitudes does not necessarily translate into a strong link between immigration levels and public Euroscepticism. Public Euroscepticism can still be low even if immigration levels are high.
Introduction
Does immigration boost public Euroscepticism in European Union (EU) member states? 1 Much of the anti-immigration rhetoric featured in Eurosceptic discourses would lead us to believe the answer is yes. 2 Indeed, EU public opinion scholars have established a strongly positive link between anti-immigration and Eurosceptic attitudes (e.g. Boomgaarden et al., 2011; De Vreese and Boomgaarden, 2005; Lubbers and Scheepers, 2007). Recent work has also attributed the electoral success of right-wing, Eurosceptic parties to their emphasis on anti-immigration (e.g. Downes and Loveless, 2018; Downes et al., 2021; Evans and Mellon, 2019; Treib, 2014). 3 However, it remains theoretically and empirically unclear whether higher actual levels of immigration would necessarily increase public Euroscepticism.
One theoretical problem with the assertion of a positive link between actual immigration levels and Euroscepticism is that it assumes public knowledge of actual immigration levels. This assumption, however, is unlikely to be satisfied given a well-documented misalignment between actual and perceived levels of immigrant population in EU member states (Figure 1; see also Alesina et al., 2018; Duffy and Frere-Smith, 2014; Ipsos MORI, 2013) . Another theoretical problem with the assertion is its failure to consider intergroup contact (Allport, 1954; Pettigrew and Tropp, 2006). Under this framework, positive contact between immigrants and local citizens may reduce anti-immigration sentiments, which can decrease public Euroscepticism in turn.

Misalignment between actual and perceived changes in foreign-born population size. Note: The y-axis shows the perceived change in percentage points of foreign-born population in 12 years minus the actual change in percentage points of foreign-born population in 12 years. Data on perceived changes are from Round 1 and Round 7 of European Social Survey—the only two waves that asked about the perceived size of foreign-born population. Data on actual changes are from the OECD.
The relationship between actual levels of immigration and public Euroscepticism is therefore a priori ambiguous. Surprisingly, virtually no research to date has tested it empirically. 4 This study thus fills the gap by directly examining the link between actual immigration levels and public Euroscepticism across all EU member states. Using multilevel modeling to analyze immigration data and Eurobarometer survey data over the period 2009–2017, I find no evidence that individual levels of Euroscepticism increase with actual levels of immigration, regardless of whether the immigrants are from EU or non-EU member states. The finding thus challenges the common belief that higher levels of immigration would make EU citizens more Eurosceptic. Although further EU enlargement implies even more immigration across EU member states, my results suggest that the future success of the European project is not inherently doomed, since public Euroscepticism can still be low even if immigration levels are high.
This study makes three distinct contributions. First, by moving beyond the existing literature that uses immigration attitudes—instead of actual levels of immigration—to study public opinion on European integration (e.g. De Vreese and Boomgaarden, 2005; Roeder, 2011), I find that a strong link between anti-immigration and Eurosceptic attitudes (as suggested by the current literature) does not necessarily translate into a strong link between immigration levels and public Euroscepticism. This challenges the common perception that more immigration will increase Euroscepticism. Second, I expand the literature on EU asylum and immigration policy. While scholars in this field have invested productive efforts in studying it from a legal or public administration perspective (e.g. El-Enany, 2008; Geddes, 2008; Guiraudon, 2003; Kaunert, 2009, 2010; Kaunert and Léonard 2012a, 2012b; Léonard and Kaunert, 2019, 2020), I join recent studies by approaching it from a public opinion standpoint (Jeannet et al., 2021; Stockemer et al., 2020; Van der Brug and Harteveld, 2021), suggesting that public opinion on European integration may not be as sensitive to changes in EU asylum and immigration policies as one may have previously thought. Third, I contribute to the emerging literature on the political consequences of misperceptions about immigration (e.g. Alesina et al., 2018; Gorodzeisky and Semyonov, 2020; Herda, 2013). While past research has found that misperceptions about immigration can affect public support for redistribution (Alesina et al., 2018), this study takes an alternative direction by suggesting that such misperceptions may also have important implications for public support for European integration. It thus further highlights the political importance of correcting public misperceptions about immigration, especially in the EU context.
Euroscepticism and immigration
Connecting different strands of literature, I argue that higher actual immigration levels do not necessarily lead to stronger public Euroscepticism. This is due to two reasons. First, actual and perceived levels of immigration are often weakly aligned due to public misperceptions, and it should be the latter—rather than the former—that influences individual Eurosceptic attitudes. As a result, while higher perceived levels of immigration could lead to stronger anti-immigration sentiments and therefore Eurosceptic attitudes, the effect of actual levels of immigration on public Euroscepticism may not be as strong. Second, immigration promotes intergroup contact, which can weaken Eurosceptic attitudes by reducing individuals’ anti-immigration sentiments. Such a negative effect (resulting from intergroup contact) may offset the positive—but already weak—effect of actual immigration levels on public Euroscepticism (resulting from public misperceptions of immigration levels). Figure 2 summarizes my general argument, proposing that actual immigration levels play an a priori ambiguous role in shaping public Euroscepticism. The rest of this section develops this argument further by discussing, connecting and synthesizing multiple strands of literature.

Ambiguous link between Euroscepticism and actual immigration levels. Note: The plus/minus sign indicates the positive/negative causal relationship either according to the literature or by direct implication. On one hand, due to misalignment between actual and perceived levels of immigration, the overall effect of immigration on public Euroscepticism through the perceived immigration channel is likely to be positive but weak. On the other, due to intergroup contact facilitated by immigration, immigration has a theoretically negative effect on public Euroscepticism through the intergroup contact channel. These two effects may cancel out each other, leading to a non-relationship between Euroscepticism and actual immigration levels.
Anti-immigration sentiments and Euroscepticism
It is well established that anti-immigration sentiments lead to public Euroscepticism. As freedom of movement for citizens across EU member states is a direct implication of European integration, individuals with stronger anti-immigration sentiments are more likely to reject the idea of European integration (De Vreese and Boomgaarden, 2005). Indeed, it has been found that anti-immigration sentiments are “a key variable for understanding reluctance about integration” (De Vreese and Boomgaarden, 2005: 59; see also Boomgaarden et al., 2011), and that perceived threats from immigrants strongly explain Eurosceptic attitudes (Lubbers and Scheepers, 2007). Recent literature also suggests that Eurosceptic voting can be explained by anti-immigration sentiments. For example, Hobolt and De Vries (2016) find that anti-immigration sentiments significantly increase EU citizens’ likelihood of voting for right-wing Eurosceptic parties; Barbulescu and Beaudonnet (2014) show that such sentiments substantially decrease Italians’ support for European integration; and Meleady et al. (2017), and De Vyver et al. (2018) demonstrate that such sentiments strongly predict British voters’ support for Brexit. In short, past research has established a positive link between anti-immigration sentiments and Euroscepticism.
Perceived immigration levels and anti-immigration sentiments
Past research has also established that higher perceived levels of immigration lead to stronger anti-immigration sentiments. This is because immigrants activate citizens’ perceived cultural (McLaren, 2002, 2006), identity (McLaren and Johnson, 2007), ethnic (Schneider, 2008), and economic (Blalock, 1967; Blumer, 1958; Bobo and Hutchings, 1996) threats, which intensify anti-immigration sentiments in turn. For instance, Sides and Citrin (2007) find that perceived cultural and identity threats posed by immigration incite anti-immigration sentiments among EU citizens; Hainmueller and Hiscox (2010) show individual perceptions of how immigration affects the domestic economy as a whole strongly explain anti-immigration sentiments; and Sniderman et al. (2004: 46) find perceived economic threats from immigrants “evoke hostility to immigrants.”
Importantly, existing literature shows that, while anti-immigration sentiments are strongly shaped by perceived immigration levels, they are less affected by actual immigration levels. Semyonov et al. (2004) find no evidence that perceived threats from immigrants are incited by a larger foreign population, but demonstrate that the perceived size of foreign population is strongly associated with perceived threats. Similarly, Gorodzeisky and Semyonov (2020) show that both perceived immigration levels and misperceptions of immigration levels—rather than actual immigration levels—play an important role in shaping EU citizens’ attitudes towards immigration. Hence, it is apparent that there is a positive relationship between perceived immigration levels and anti-immigration sentiments, but a weaker relationship between actual immigration levels and anti-immigration sentiments.
Actual and perceived immigration levels
Indeed, existing evidence shows that actual and perceived levels of immigration are often weakly aligned. A population-based survey by Ipsos MORI (2013), for example, shows that British citizens typically overestimate the percentage of foreign-born population in the UK. Among those who provided an estimate, 68% believed that the level was at least 16%, and 35% even thought that at least a quarter of the UK population were immigrants, when the actual level was in fact 13–15%. Only less than one-fifth underestimated the level of foreign-born population by suggesting that the level was 10% or below. Similar evidence of misalignment between actual and perceived immigration levels is also found in other EU member states, including France, Germany, Italy and Sweden (Alesina et al., 2018; Duffy and Frere-Smith, 2014).
My analysis of the European Social Survey data also reveals a misalignment between actual and perceived changes in foreign-born population size in European countries. Figure 1 suggests that European citizens are unable to accurately identify the actual changes of foreign-born population in their country. A closer examination of the data shows that while the actual percentage changes of Denmark’s and Hungary’s foreign-born populations were very similar (2.4% vs 1.6% in 12 years), citizens had a 7.4-percentage point difference in their perceived percentage changes of foreign-born population in their country (3.8% vs –3.6%). On the other hand, while the actual percentage changes of Austria’s and Ireland’s foreign-born populations had a 5.3-percentage point difference (2.7% vs 8.0% in 12 years), the perceived percentage changes were nearly identical among the public (5.6% vs 5.8%). Thus, misperceptions of immigration levels appear to be highly prevalent in EU member states.
Intergroup contact and anti-immigration sentiments
Importantly, the social psychology literature suggests that intergroup contact may reduce anti-immigration sentiments. According to Allport’s (1954) intergroup contact theory, intergroup contact can reduce prejudice between majority and minority group members. Pettigrew and Tropp’s (2006: 751) seminal meta-analytic study finds that intergroup contact “typically reduces” prejudice and that intergroup contact theory is highly extensible and generalizable to other groups. Thus, it is not inconceivable that the theory also applies in the EU immigration context (Schneider, 2008). Indeed, Binder et al. (2009), Meleady et al. (2017) and Voci and Hewstone (2003) find evidence from different EU member states that anti-immigration sentiments are reduced when local citizens have positive contact with immigrants (see also Graf et al., 2014). More specifically, McLaren (2003: 929) has shown that “contact does have an effect on attitudes toward immigrants in the European context,” and Wagner et al. (2006) have found that immigration from ethnic minority countries reinforces intergroup contact in Germany, thereby reducing Germans’ prejudice against immigrants. In short, higher levels of immigration may reduce anti-immigration sentiments through the channel of intergroup contact per se.
Actual immigration levels and anti-immigration sentiments: an ambiguous link
Most importantly, the existing literature fails to find a clear relationship between actual immigration levels and anti-immigration sentiments. Pottie-Sherman and Wilkes’s (2017) meta-analysis finds that the link between actual immigration levels and anti-immigrant sentiments is weak, whereas the perceived levels consistently explain anti-immigrant prejudice among individuals. Similarly, Stockemer (2016: 1010) suggests that anti-immigration sentiments are “completely unrelated” to the actual number of immigrants in the individuals’ neighborhoods. In addition, Hjerm (2009: 47) demonstrates that “the proportion of the foreign-born population has no effect on anti-immigrant attitudes,” and Schneider (2008) suggests that perceived ethnic threats can actually be reduced by intergroup contact.
As a result, the common belief that immigration boosts public Euroscepticism is theoretically contestable for two reasons. First, due to public misperceptions about immigration levels, perceived immigration levels may be insensitive to changes in actual immigration levels. In turn, the positive effect of actual immigration levels on public Euroscepticism through the perceived immigration channel is unlikely to be strong (see Figure 2). Second, owing to the increased intergroup contact facilitated by higher levels of immigration, anti-immigration sentiments and, consequently, Eurosceptic attitudes among local citizens may decrease. Such a negative effect of actual immigration levels on public Euroscepticism through the intergroup contact channel could offset—or even trump—any positive effect that is derived from the perceived immigration channel (see Figure 2). Given such tension, I argue that higher actual immigration levels do not increase public Euroscepticism. The rest of this article, then, is to empirically test this argument.
Data and methods
I test whether actual immigration levels are positively associated with public Euroscepticism by leveraging both country-level immigration data and individual-level survey data. This section describes the data and explains my estimation strategy.
Data sources
I construct a unique dataset that consists of both individual- and country-level variables from the EU-28 (Austria, Belgium, Bulgaria, Croatia, Cyprus, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, the Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden and the UK). Eurobarometer surveys provide the individual-level data for this study. They are commonly used for research on public Euroscepticism and support for European integration (e.g. Christin, 2005; Clark and Hellwig, 2012; Foster and Frieden, 2021; Hooghe and Marks, 2007; Lubbers and Scheepers, 2005, 2010; Van Klingeren et al., 2013). With a multi-stage, random probability sampling method, Eurobarometer surveys are nationally representative, and are regularly conducted by the European Commission to collect public opinion on EU-related topics such as the EU budget, media use in the EU and the EU as an institution in general. All current and official candidate EU member states are sampled in each wave. For each member state, at least 1000 individuals aged 15 years or above are interviewed in each wave, except for Cyprus, Luxembourg and Malta which have around 500 respondents per wave. Data are collected through face-to-face interviews in respondents’ homes, based on uniform and detailed instructions in respondents’ native language. The Eurobarometer surveys used in this study start from 71.3 (2009) and end with 88.1 (2017). I begin my analysis with year 2009 because there are many missing data in the immigration dataset for year 2008. The end year is 2017 because some variables used in my analysis are not available afterwards.
Eurostat provides the immigration data for my analysis (migr_imm5prv). This dataset is used because it contains immigration data by country of previous residence, which allows me to separate my analysis by EU and non-EU immigrants. This distinction is important because EU and non-EU immigration may have different effects on public Euroscepticism. On one hand, we may expect EU immigration to have a larger impact on Euroscepticism because one primary policy of the EU is precisely the freedom of movement for citizens across EU member states—a fundamental principle enshrined in the Treaty on the Functioning of the European Union. On the other, we may instead expect non-EU immigration to be more influential on Euroscepticism because of racist or prejudicial attitudes towards non-EU immigrants. 5 If non-EU immigration in a white neighborhood is more salient, and intergroup contact is less positive between local citizens and non-EU immigrants due to racism, it is theoretically possible that EU and non-EU immigration have different effects on public Euroscepticism. This is especially plausible given recent research which shows that some EU citizens would support EU immigration but, at the same time, oppose non-EU immigration (Blinder and Markaki, 2019).
All economic and social variables at the country level are taken from Eurostat, except for GDP growth and population data which are obtained from the World Bank. Moreover, political variables are taken from Cruz et al. (2018) and the World Bank.
Dependent variables
As past research suggests that there are multiple dimensions of Euroscepticism (Boomgaarden et al., 2011), I operationalize public Euroscepticism in two different ways. The first operationalization captures Instrumental Euroscepticism—the most common conceptual and empirical dimension of Euroscepticism in past research. Two questions from Eurobarometer surveys are used to create this variable, which then ranges from 0 (least Eurosceptic) to 4 (most Eurosceptic) by addition. The first question is: “Generally speaking, do you think that (our country’s) membership of the EU is a good thing (0); a bad thing (2); or neither a good thing nor a bad thing (1)?” Responses with “don’t know” are dropped from the analysis. Alternatively, I treat non-responses as neutral responses, with all substantive findings remaining unchanged. The second question is: “Taking everything into account, would you say that (our country) has on balance benefited or not from being a member of the EU? Benefited (0); not benefited (2).” Responses with “don’t know” are coded as 1 here. 6 This operationalization is largely based on Lubbers and Scheepers (2005: 228; 2010: 794), McLaren (2002: 556), and Van Klingeren et al. (2013: 695).
While the first operationalization captures the instrumental (or utilitarian) dimension of Euroscepticism, the second one aims to capture the affective dimension of Euroscepticism, which corresponds to “negative affection” as conceptualized by Boomgaarden et al. (2011). I use the following question from Eurobarometer surveys to measure Affective Euroscepticism: “In general, does the EU conjure up for you a very positive (0), fairly positive (1), neutral (2), fairly negative (3) or very negative image (4)?” Responses with “don’t know” are dropped from the analysis. Thus, Affective Euroscepticism also ranges from 0 (least Eurosceptic) to 4 (most Eurosceptic), facilitating the comparison of the two sets of results that are based on two different dependent variables.
Key explanatory variables
Actual levels of immigration from EU and non-EU member states are the key explanatory variables in this study. They are operationalized as follows:
Control variables
I control for a number of individual- and country-level predictors of public Euroscepticism suggested in the literature. For individual-level control variables, I include gender, age, education, difficulty paying bills, community size, employment status and occupation. I additionally control for left-right self-placement in the analysis where Instrumental Euroscepticism is the dependent variable (Van Elsas et al., 2016). I also control for self-reported life satisfaction, trust in the national government and intensity of national identity in my analysis where Affective Euroscepticism is the dependent variable. 8 These individual-level variables are commonly controlled for in empirical research on Euroscepticism, and are often important in explaining individual Eurosceptic attitudes (see, e.g. Baute et al., 2018; Lubbers and Scheepers, 2010). Thus, including them will improve the precision of my estimates, which is especially important for establishing a null finding.
For country-level control variables, I include economic variables including GDP growth rate (based on constant local currency), inflation rate and unemployment rate; social variables including tertiary educational attainment and social protection expenditure per capita (logged); and political variables including government effectiveness, political ideology of the incumbent and a dummy variable indicating whether a national election took place in the corresponding country-year. These variables are appropriately lagged to match the individual-level variables. Controlling for these country-level variables is important because excluding them would likely result in omitted variable bias. For instance, the economic variables are likely to be correlated with not only public Euroscepticism (Foster and Frieden, 2021) but also immigration levels, because people may tend to emigrate to countries with better economic prospects. Similarly, as people may tend to emigrate to countries with better education and welfare systems, it is important to control for tertiary educational attainment and social protection expenditure which are likely to be associated with public Euroscepticism (Baute et al., 2018). As public Euroscepticism is often influenced by the flow of political information (De Vries et al., 2011) and different governance could lead to different immigrant intakes and political communications, it is also theoretically important to control for the political variables.
The Online appendix provides additional information on the variables and datasets.
Estimation strategy
I use multilevel modeling to perform the analysis. Specifically, I estimate two-level hierarchical linear models (HLMs), with the individual level treated as the first level and the country level treated as the second. This approach has a clear advantage over an alternative method that pools all the data, adjusts for country clusters and then estimates the population-averaged effects. This is because such a method would insufficiently take into account the hierarchical and longitudinal structure of my dataset (Gelman and Hill, 2006; Hox, 2002). On the other hand, HLMs also appear more appropriate than fixed-effects models because the number of country clusters (28) is relatively small in my dataset. With only few clusters, fixed-effects models tend to produce inferior estimates compared to random-effects models (Clark and Linzer, 2015: 407; see also Plümper and Troeger, 2019), which are characterized by HLMs. Although precisely identifying the causal effect of immigration levels on public Euroscepticism is not the main aim of this study, it is still useful to note that the estimates reported in this article would be causal if, conditional on the control variables, different country-years were comparable in terms of all unobserved factors that jointly determine immigration levels and Euroscepticism. The Online appendix provides technical details for my estimation strategy.
Results
Does immigration boost public Euroscepticism? I first present the results associated with Instrumental Euroscepticism. If the common perception that higher levels of immigration increase public Euroscepticism is true, then the estimated coefficients for EU immigration (
Results of hierarchical linear models for Instrumental Euroscepticism.
Note: All country-level variables, except for dummies, are standardized. Individual variables for age and left-right placement are also standardized. Estimates for individual variables for occupations are not shown to conserve space. See the Online appendix for definitions and sources of the variables. Robust standard errors are clustered at the country level. Wald tests are used to compute the
Now I turn to the results associated with Affective Euroscepticism. If immigration boosts public Euroscepticism on the affective dimension, then the estimated coefficients for EU immigration and non-EU immigration should be positive. Table 2, however, shows that all estimated coefficients for non-EU immigration are again negative. On the other hand, while the estimated coefficients for EU immigration are mostly positive, they are imprecisely estimated. More importantly, the effect size is substantively small: my calculations based on Model 8 and the 2017 immigration data suggest that Germany and the UK have to respectively increase their EU immigrant intakes by approximately 1.3 times and 1.7 times in order to increase their citizens’ Affective Euroscepticism by 0.15 points on a 5-point scale. As a reference, Model 8’s estimates suggest that the effect of a 0.3-standard deviation increase in social protection expenditure per capita (logged) in my dataset is already strong enough to impact Affective Euroscepticism by the same magnitude.
Results of hierarchical linear models for Affective Euroscepticism.
Note: All country-level variables, except for dummies, are standardized. Individual variables for age are also standardized. Estimates for individual variables for occupations are not shown to conserve space. See the Online appendix for definitions and sources of the variables. Robust standard errors are clustered at the country level. Wald tests are used to compute the
In sum, the results suggest no evidence that public Euroscepticism increases with actual levels of immigration. This is in line with my earlier theoretical discussions that intergroup contact may act as a countervailing force, which offsets the positive but weak effect of actual immigration levels on public Euroscepticism resulting from the perceived immigration channel. I probe the mechanism further by studying the heterogeneous effects by community type.
Heterogeneous effects by community type
Because immigrants tend to settle in cities but not in villages, I subset my analysis by the type of community where the individuals live. Specifically, I analyze whether the effects of immigration on Euroscepticism are different for those who live in large towns compared to those who live in smaller communities. On one hand, we may expect an even stronger negative link between immigration levels and Euroscepticism among citizens living in large towns because they are much more likely to have (positive) contact with immigrants. On the other, we may expect a less negative—or even a positive—link among these people because the salience of immigration may be higher in large cities, which leads to not only a better alignment between actual and perceived levels of immigration but also stronger perceived threats from immigrants.
My analysis supports the former view. For individuals living in large towns, their Instrumental Euroscepticism is negatively and significantly correlated with the levels of EU immigration (
Robustness checks
One potential concern is that public Euroscepticism may be sensitive to the speed of demographic transformation due to immigration, instead of the level of immigration in each separate year. To address this concern, I develop an alternative measure of immigration to check the robustness of my results: percentage change of the actual immigration level. Specifically, I create two new variables, which measure the one-year percentage changes of the numbers of EU and non-EU immigrants in a country in a given year:
The new results suggest an even more precise null effect of immigration on public Euroscepticism (see the Online appendix for full results). In almost all specifications, the estimated coefficients for both newly created immigration variables are statistically indistinguishable from zero. On both instrumental and affective dimensions, the zero effects of changes in immigration levels on public Euroscepticism appear to be precisely estimated.
I additionally analyze the effect of net immigration on public Euroscepticism (see the Online appendix). The results consistently suggest no evidence that higher levels of net immigration are associated with greater public Euroscepticism. On both instrumental and affective dimensions, net immigration is negatively—rather than positively—associated with public Euroscepticism. This suggests further evidence that actual levels of immigration and public Euroscepticism are not positively correlated.
Since public Euroscepticism can be fluid, year effects are not necessarily linear. Thus, I also check the robustness of my results by conducting the multilevel analysis by year (see the Online appendix). Figure 3 shows the year-by-year analysis. In Figure 3(a), there is no evidence that actual immigration levels from EU and non-EU member states boost Instrumental Euroscepticism in any year between 2009 and 2017. In line with previous results, the new estimates seem to suggest that, if there is any relationship between EU immigration and Instrumental Euroscepticism, it will be slightly negative: more immigration from other EU member states is associated with weaker Instrumental Euroscepticism among the public. This again suggests that the negative effect through the intergroup contact channel is strong enough to offset the positive effect through the perceived immigration channel (as outlined in Figure 2). Similarly, Figure 3(b) suggests no evidence that actual immigration levels from EU and non-EU member states boost Affective Euroscepticism, with nearly all coefficients precisely estimated at zero. 10 Together with the results reported in Tables 1 and 2, these results consistently show no evidence for the common belief that immigration boosts public Euroscepticism in EU member states.

Coefficients for immigration variables by year—multilevel linear estimates. Panel (a): Instrumental Euroscepticism. Panel (b): Affective Euroscepticism. Note: All covariances are assumed to be independent in all estimates. Error bars indicate 95% confidence intervals.
Since the dependent variables are ordinal in nature, I also perform multilevel ordered logistic regressions by year. The results are reported in Figure 4. There is, again, no evidence that a higher level of immigration from EU or non-EU member states is associated with stronger public Euroscepticism, either instrumentally or affectively.

Coefficients for immigration variables by year—multilevel ordered logistic estimates. Panel (a): Instrumental Euroscepticism. Panel (b): Affective Euroscepticism. Note: All covariances are assumed to be independent in all estimates. Error bars indicate 95% confidence intervals.
As a final robustness check, I subset my data to only include the EU-15 (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden and the UK). This is to take into account the possibility that public opinion towards the EU is systematically different in older EU member states (Anderson and Kaltenthaler, 1996). Once again, the results suggest no evidence that higher actual immigration levels increase public Euroscepticism. For Instrumental Euroscepticism, the estimated coefficients for EU and non-EU immigration are –0.31 (
Conclusion
The main aim of this contribution is to fill an important gap in the Euroscepticism literature: does immigration boost public Euroscepticism in EU member states? Contrary to common belief, the results suggest no evidence that individual levels of Euroscepticism increase with actual levels of immigration. Across EU member states, public Euroscepticism remains generally unaffected by actual immigration levels on both instrumental and affective dimensions. Drawing upon different strands of literature, I attribute this finding to public misperceptions of actual immigration levels, as well as the increased intergroup contact facilitated by higher levels of immigration.
My findings have important implications for the future development of the EU. Due to rising anti-immigration sentiments in Europe and the electoral success of right-wing Eurosceptic parties in recent years, there seems to be a common perception that the future of the European project is at stake: if people dislike immigration and are therefore Eurosceptic (as suggested by past research), public support for further European integration will be inevitably low. However, here I argue that it is probably anti-immigration sentiments per se—instead of actual levels of immigration—that explain public Euroscepticism. Although further EU enlargement implies even more immigration across EU member states, the future success of the European project is not inherently doomed, since public Euroscepticism can still be low even if the immigration levels are high (as suggested by my findings). Therefore, EU officials and pro-EU politicians can still be optimistic about the EU’s future, as long as anti-immigration sentiments and democratic deficits (Norris, 1997, 2011; Rohrschneider, 2002; Van Spanje and De Vreese, 2011), for example, are properly dealt with.
This study advances the literature on Euroscepticism in three distinct ways. First, while past research has focused on how anti-immigration sentiments lead to Eurosceptic attitudes among the public, this study is among the first to empirically test the relationship between actual levels of immigration and public Euroscepticism. Therefore, it enriches our understanding of how immigration is (or is not) related to public Euroscepticism. Instead of studying how immigration attitudes affect Euroscepticism, studying the relationship between actual immigration levels and Euroscepticism is arguably more useful, since the former is essentially a “regressing attitudes on attitudes” research problem which tends to produce unclear causal directions and have limited relevance to public policy (Hainmueller and Hopkins, 2014: 243).
Second, this study expands the literature on EU asylum and immigration policy by approaching it from a different academic perspective. While past scholars have studied it extensively from the legal and public administration perspectives, I follow recent work (Jeannet et al., 2021; Stockemer et al., 2020; Van der Brug and Harteveld, 2021) by studying it from a public opinion standpoint. Doing so, I offer a new perspective by suggesting that Eurosceptic attitudes among the public may not be as sensitive to changes in EU asylum and immigration policies as one may have previously thought. Third, this study enriches the emerging literature on the political consequences of misperceptions about immigration. Past research has, for example, shown that such misperceptions have important implications for public support for redistribution (Alesina et al., 2018). This study takes a new direction by suggesting that misperceptions about immigration may also have implications for public support for European integration. This further highlights the political importance of correcting public misperceptions about immigration, especially in the EU context.
Given that imposing stricter immigration restrictions may not be a right way to address public Euroscepticism, future research should study how such sentiments can be appropriately mitigated by public policy efforts. There are recent efforts by the European Commission to publicize materials which debunk myths about migration to Europe and the work of the EU in immigration issues. To the extent that preventing EU citizens from overestimating the level of immigration in their country could reduce public Euroscepticism, such policy efforts seem to be very promising, and future research should test their effectiveness empirically. Moreover, future research may find it useful to study how immigration policies may be affected by public Euroscepticism. A common understanding is that immigration policies affect immigration levels, which then affect public Euroscepticism. This study challenges the latter link by arguing for a non-relationship between immigration levels and public Euroscepticism. Yet an important question remains: to what extent does public Euroscepticism affect the policy-making process of immigration policies? Addressing this question will contribute to both our understanding of Euroscepticism and immigration policy. Certainly, it will be an important avenue for future research.
Supplemental Material
sj-pdf-1-eup-10.1177_14651165211030428 - Supplemental material for Does immigration boost public Euroscepticism in European Union member states?
Supplemental material, sj-pdf-1-eup-10.1177_14651165211030428 for Does immigration boost public Euroscepticism in European Union member states? by Eddy SF Yeung in European Union Politics
Supplemental Material
sj-txt-1-eup-10.1177_14651165211030428 - Supplemental material for Does immigration boost public Euroscepticism in European Union member states?
Supplemental material, sj-txt-1-eup-10.1177_14651165211030428 for Does immigration boost public Euroscepticism in European Union member states? by Eddy SF Yeung in European Union Politics
Supplemental Material
sj-zip-1-eup-10.1177_14651165211030428 - Supplemental material for Does immigration boost public Euroscepticism in European Union member states?
Supplemental material, sj-zip-1-eup-10.1177_14651165211030428 for Does immigration boost public Euroscepticism in European Union member states? by Eddy SF Yeung in European Union Politics
Footnotes
Acknowledgements
I am grateful to Toke Aidt, Natália Bueno, Peter Carroll, Siv Cheruvu, Yuequan Guo, Nahomi Ichino, Lazare Kovo, Guoer Liu, Siniša Mirić, Eitan Paul, Hubert Tworzecki, Hsu Yumin Wang, Liam Yuen, and Htet Thiha Zaw for their feedback. I also thank the editor and four anonymous reviewers at EUP for their thoughtful suggestions.
Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author received no financial support for the research, authorship and/or publication of this article.
Supplemental material
Supplemental material for this article is available online.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
