Introduction
According to virtually all normative views of democracy, citizens are at the core of democratic processes (Manin, Reference Manin1995). Classical political theorists, like Bentham and Rousseau, argued that the public, its identity and its ability to act must take a central role in any theorization of politics and the overall democratic process. Empirical scholars, more recently, argued that public opinion can have substantial proximate effects on, or at least interplay with, the political process, and especially policy-making (Page and Shapiro, Reference Page and Shapiro1992; Stimson, MacKuen and Erikson, Reference Stimson, Mackuen and Erikson1995; Erikson, MacKuen and Stimson, Reference Erikson, MacKuen and Stimson2002; Soroka and Wlezien, Reference Soroka and Wlezien2010). While these two literatures often feud, they jointly promote a more precise understanding of citizens’ opinions and consider politics more than a sequence of elections. Without an effective and systematic understanding of what the public wants, it would be premature to adjudicate the intricate dynamics of democracy, representation and responsiveness.
The question then becomes whether – and if so, to what extent – the changes we observe in public opinion are systematic responses to citizens’ shifting environments. Regardless of important theoretical and empirical insights from Anglo-Saxon applications, there remains a gap in our understanding of how aggregated public opinion varies through time and particularly across contexts, as well as what constitutes the macro-societal correlates of comparative public opinions. This study proposes a framework from which the macro phenomena associated with public opinion (equally an aggregated construct) can be understood and analyzed. We set out to contribute to this literature by focusing on a specific domain of public opinion, namely immigration.
Data scarcity has made the comparative and aggregate analysis of public opinion towards immigration, or ‘immigration opinions’, challenging until now. With that in mind, this study relies on a combination of more than 500 unique item series and 2,500 survey marginals to estimate aggregate measures of immigration opinions across 13 European democracies ranging from 1980 to 2017. We then scrutinize opinion change by means of a multi-faceted framework. More precisely, we consider four key theoretical drivers of immigration opinions. First, we expect economic hardship to fuel anti-immigration opinions. Second, thermostatic responsiveness would suggest that restrictive immigration policies incite more favourable immigration opinions. Third, we argue that immigration itself moves immigration opinions. Fourth, we anticipate that political parties can steer immigration opinions. Altogether, our findings provide key insights into the proposed theoretical accounts, not the least because they highlight that – on average – social and economic factors are not the sole drivers of immigration opinions, but there is also an important role for political parties in this regard.
The potential applications and contributions of this study are fourfold. (i) The estimation of a comprehensive aggregate index provides an empirical opportunity to explain variation in immigration opinions through time and across countries, contributing to a growing literature analyzing immigration opinions from a macro-societal perspective (Jennings, Reference Jennings2009; Ford, Jennings and Somerville, Reference Ford, Jennings and Somerville2015; Claassen and McLaren, Reference Claassen and McLaren2021; Van Hauwaert, Reference Van Hauwaert2022). (ii) The finding and concept of common movement in immigration opinions may revise our understanding of political opportunities and supply-side dynamics. If immigration opinions across Europe share much of their variation, we may consider European integration and supranational policy design as more effective than nationally tailored political responses. (iii) The drivers of immigration opinions may exhibit important interactions with other indicators, such as spatial or partisan variables, offering a range of new theoretical and empirical questions that remain unexplored. (iv) The estimation of immigration opinions may offer a new explanatory variable in macro-political models. This study thereby offers a public opinion measure that extends existing analyses and can provide new insights into political and voting behaviour related to immigration.
Immigration opinions as an issue-specific aggregate measure
Public opinion is an aggregate of current or long-term individual policy preferences, attitudes, or beliefs regarding a specific set of issues and questions (Key Jr., Reference Key1961). When studying aggregate public opinion, scholars have taken two different conceptual and empirical approaches. The first is to regard public opinion as an aggregate of individuals and issues, a concept often called “policy mood” (Stimson, Reference Stimson1991). While such a singular interpretation of public opinion has important merits, the complexification of the political space and the increasing relevance of political issues over class-cleavages highlight the intrinsic multi-dimensionality and heterogeneity of politics. Therefore, a second approach emerged in recent years examining aggregate public opinion in issue-specific forms.Footnote 1 So far these accounts of issue-specific public opinion only represent a fraction of the current literature, most common in an American or broader Anglo-Saxon context, where scholars form opinion measures about race (Kellstedt, Reference Kellstedt2000), punitiveness (Enns, Reference Enns2014; Jennings et al., Reference Jennings2017), gay rights (Lax and Phillips, Reference Lax and Phillips2009) and gender equality (Koch and Thomsen, Reference Koch and Thomsen2017) – amongst others. In a comparative context, such measures are less frequent. Yet, we find notable examples of support for welfare (Brooks and Manza, Reference Brooks and Manza2006), redistribution (Lupu and Pontusson Reference Lupu and Pontusson2011; Romero-Vidal and Van Hauwaert Reference Romero-Vidal and Van Hauwaert2021), democracy (Claassen, Reference Claassen2020b, Reference Claassen2020a), Muslims (Cinalli and Van Hauwaert, Reference Cinalli and Van Hauwaert2021) and European integration (Bølstad, Reference Bølstad2015), amongst others.
Here, we focus on public opinion related to immigration. In advanced democracies, immigration is one of the most salient issues since the 1980s. As a socio-political issue, it has equitable polarization potential, meaning “reasonable people can take either side” (Stimson, Reference Stimson1991, p. 8). Therefore, it is not only possible but also theoretically justified to conceptualize and operationalize an issue-specific public opinion towards immigration. We can understand this as a holistic and harmonized estimation that describes the evolution of a set of collective preferences or aggregated opinions towards the levels and impact of immigration for any given population.
In the recent literature, we find several studies that already use aggregate measures of public opinion about immigration. Initial efforts focused on Great Britain or the UK (Jennings, Reference Jennings2009; Ford, Jennings and Somerville, Reference Ford, Jennings and Somerville2015), while more recent studies provide proper comparative insights into immigration opinions. Van Hauwaert and English (Reference Van Hauwaert and English2019) find that regional-level immigration opinions in Belgium, France and the UK become more favourable towards immigrants when immigration levels increase. Claassen and McLaren (Reference Claassen and McLaren2021) argue there is some negative short-term public reaction to increasing immigration, but this is largely cancelled out when a country develops more extensive experience with immigration. Van Hauwaert (Reference Van Hauwaert2022) examines the linkage of immigration opinions with policy across European democracies, finding negative policy feedback when citizens formulate their immigration opinions. We rely on immigration opinion data from the latter study – and further update it – to examine the origins of these immigration opinions. Our goal in this study is to provide a broader overview of the macro-contextual drivers of public opinion about immigration. As such, a careful comparative analysis adds to our understanding of European public and their overall responsiveness to social environments.
The origins of aggregate immigration opinions
Most of the literature on the formation or change of immigration sentiments are individual-level studies (Finseraas, Pedersen and Bay, Reference Finseraas, Pedersen and Bay2016; Hatton, Reference Hatton2016; Pardos-Prado and Xena, Reference Pardos-Prado and Xena2019), and employ either case study (experimental) designs or cross-sectional strategies. This provides important insights into “who is opposed to immigration,” “what opinions people hold” and “who stands out compared to the mean voter or average citizen.” Additionally, the focus on individuals has allowed scholars to integrate insights from psychology about the role of political parties in shaping public opinion, unveiling mechanisms producing partisan biases, which in turn produce opinion polarization (Strickland, Taber and Lodge, Reference Strickland, Taber and Lodge2011).
Per definition, however, this approach leaves dynamics of aggregate societal change largely unexplored. This is unfortunate, to say the least. After all, what matters (most) for policy-makers is the “wisdom of the crowds,” that is the shifts and the responses observed amongst the public as a whole, rather than the (largely unconstrained and uninformed) individual opinions (Page and Shapiro, Reference Page and Shapiro1992).Footnote 2 How these aggregate opinions may be affected and how they are shaped in a crucial policy domain such as immigration is of the utmost importance for policy-makers. Therefore, this study theorizes systematic change that leads to aggregate movement between time points. As citizens change their immigration opinions, we subsequently examine what role the environment plays and how this shapes the macro polity (Erikson, MacKuen and Stimson, Reference Erikson, MacKuen and Stimson2002). In other words, our goal is not to explain individual attitudinal patterns, but the results of a multitude of individual changes that are large and uniform enough to move the averages. That is, our goal is to explain major societal shifts.
Drawing from the literature, we identify four key drivers that can theoretically underlie such opinions. First, in line with earlier macro-findings, we expect economic hardship to fuel anti-immigration opinions. Second, public responsiveness models suggest that restrictive immigration policies incite more favourable immigration opinions. Third, we argue that immigration itself moves immigration opinions. Fourth, we anticipate that political parties and their positions on the topic can steer immigration opinions. Together, this provides a combined understanding of the extent to which publics rationally respond to their environments when deliberating immigration opinions.
Economic conditions shaping the public immigration sphere
Aggregate analyzes of economic voting have for a long time recognized the connection between electoral outcomes and macro-economic benchmarks, most often empirically linking economic performance to political support (Mueller, Reference Mueller1970; Hibbs, Reference Hibbs1977). Since then, scholars have systematically considered the general conditions of the economy to affect different interpretations of political behaviour. Altogether, economic factors are probably amongst the most examined environmental determinants with regards to the conditions under which opinions form and preferences shape (Powell Jr and Whitten, Reference Powell and Whitten1993; Nannestad and Paldam, Reference Nannestad and Paldam1994).
Initial scholarship relied on consumer choice theory to explain how economic conditions explain political behaviour (Fiorina, Reference Fiorina1978; Tufte, Reference Tufte1978; Lewis-Beck, Reference Lewis-Beck1988, Reference Lewis-Beck1990). While citizens typically prefer more to less wealth, it stands to reason that the derived utility decreases. That is, a threshold may be met where the relative weight of the simple desire for “more” is lessened with respect to an individual’s decision-making processes. Durr (Reference Durr1993) argues – and finds – the same is true for the public as a whole. The public will be more willing to redistribute a country’s wealth and pursue alternative goals in times of (economic) security. Citizens become more favourable to government intervention because the corresponding (left-wing) policies typically include more spending and are, thus, more expensive. When the economy appears vulnerable and (economic) returns become scarcer, however, the public will demand less redistribution. Rather than intervene, citizens demand that governments pay greater attention to the economy itself. Concerns for – often weaker or oppressed – minorities can be suspended rather quickly when a typically secure public suddenly feels it is in danger of economic hardship.
Most empirical studies agree that economic conditions have a profound effect on what people want, and thus serve as relatively easy input signals, most notably because economic messages are so pervasive in the media (Soroka, Stecula and Wlezien, Reference Soroka, Stecula and Wlezien2015; Soroka and Wlezien, Reference Soroka and Wlezien2022). Most studies – using a wide variety of indicators – find that economic prosperity stimulates opinions favouring government intervention and redistribution, thereby suggesting a pro-cyclical relationship between macro-economic conditions and public opinion (Stevenson, Reference Stevenson2001). Others, however, find that high unemployment triggers greater support for government intervention, hence suggesting counter-cyclical effects (Erikson, MacKuen and Stimson, Reference Erikson, MacKuen and Stimson2002; Bartle, Dellepiane-Avellaneda and Stimson, Reference Bartle, Dellepiane-Avellaneda and Stimson2011; Bartle, Dellepiane-Avellaneda and McGann, Reference Bartle, Dellepiane-Avellaneda and McGann2019; Wlezien and Soroka, Reference Wlezien and Soroka2021).Footnote 3 The close connection between the macro-economy and public opinion generally holds for the current economic evaluations, but expectations about the future state of the economy also align with previous findings (MacKuen, Erikson and Stimson, Reference MacKuen, Erikson and Stimson1992; Durr, Reference Durr1993; De Boef and Kellstedt, Reference De Boef and Kellstedt2004).
We posit that we can also find such alignment of economic conditions with immigration opinions. Both Semyonov et al. (Reference Semyonov, Raijman and Gorodzeisky2006) and Hatton (Reference Hatton2016) suggest anti-immigrant sentiments increase when (short-term) economic performance stagnates. Arzheimer (Reference Arzheimer2009) further indicates mass unemployment relates to an increasing demand for anti-immigrant policies. These expectations stem from the belief that expanding economies reduce competition for economic or financial resources and make them less scarce, which – in turn – reduces hostilities towards immigrants. Alternatively, the increased risk exposure that stems from economic hardship will relate to more restrictive immigration opinions. Van Hauwaert and English (Reference Van Hauwaert and English2019) find indicative evidence that unemployment indeed stimulates anti-immigrant opinions in Belgium and France, but not necessarily in the UK. Similarly, Hatton (Reference Hatton2021) finds that the recent economic recession did not necessarily create an anti-immigrant backlash. The question then becomes whether we can find systematic evidence of economic conditions affecting immigration opinions across Western Europe.
The opinion-policy nexus and targeted public responsiveness
A classic scholarship focuses on the so-called “opinion-policy nexus”, and particularly on how citizens and their preferences affect the polity (Dahl, Reference Dahl1971; Manin, Reference Manin1995). This relationship, termed policy responsiveness, highlights that government output is often in line with what citizens want, particularly in highly politicized fields such as immigration (Freeman, Reference Freeman1995; Jennings, Reference Jennings2009; Ford, Jennings and Somerville, Reference Ford, Jennings and Somerville2015). While most empirical scholarship confirms policy responsiveness, it is nonetheless only part of the intricate reciprocal and equilibrating relationship between opinions and policy (Deutsch, Reference Deutsch1963; Easton, Reference Easton1965). That is, policy-makers only respond to what citizens want if citizens, in turn, notice and reward such behaviour and punish its absence. Differently put, while governments (or at least their outputs) respond to public signals, citizens are also reactive to what governments do (Wlezien, Reference Wlezien1995).
This notion of policy-reactive citizens, or public responsiveness, is less studied, but remains firmly embedded in classic political theory and the necessary democratic underpinnings of any political system (see the combined works of Bentham, Mill and Rousseau). It presupposes that citizens acquire and process information about traditional political outputs – like policies – and subsequently adjust their preferences in a meaningful and deliberative manner (Popkin, Reference Popkin1991; Zaller, Reference Zaller1992).Footnote 4 While the so-called complexification of politics could stand in the way of this, publics make extensive use of cues and cognitive heuristics for exactly this reason. It allows them to remain informed, update their beliefs, respond to policy signals and eventually make decisions with less-than-perfect information (Lupia and McCubbins, Reference Lupia and McCubbins1998; Lau and Redlawsk, Reference Lau and Redlawsk2006).
A growing number of empirical studies find public responsiveness, and more specifically, a negative relationship between public opinion and policy movement. In the USA, for example, scholars find that public opinion becomes more liberal (conservative) when governments reduce (increase) spending or pass less (more) redistributive policies (Page and Shapiro, Reference Page and Shapiro1992; Durr, Reference Durr1993; Wlezien, Reference Wlezien1995; Erikson, MacKuen and Stimson, Reference Erikson, MacKuen, Stimson, Niemi and Weisberg2001). Stevenson (Reference Stevenson2001) finds that citizens across 14 democracies move right when left-wing governments are in place or following increased social spending. Similarly, scholars find that higher income tax rates push the British electorate to the right (Bartle, Dellepiane-Avellaneda and Stimson, Reference Bartle, Dellepiane-Avellaneda and Stimson2011; Bartle, Dellepiane-Avellaneda and McGann, Reference Bartle, Dellepiane-Avellaneda and McGann2019).
While most research concentrates on redistributive or economic policies, Pacheco (Reference Pacheco2013) uses anti-smoking policies to argue that highly tangible policies affect societal perceptions and can, thus, have large feedback effects on different forms of public opinion. Expanding this rationale to Great Britain, Jennings (Reference Jennings2009, p. 866) finds support for targeted public responsiveness in the field of immigration, indicating that the “(…) public appears to notice and respond to government outputs related to administration of asylum”. To some extent, this counters earlier claims that immigration opinions are only moderately and slowly reflected in government action (Freeman, Reference Freeman1995), as well as their agenda-setting (Binzer Hobolt and Klemmensen, Reference Binzer Hobolt and Klemmensen2008; Morales, Pilet and Ruedin, Reference Morales, Pilet and Ruedin2015), thereby contributing to a genuine policy gap. This interpretation renders the question whether, considering the saliency and overall impact of different aspects of immigration, we can find evidence of public responsiveness in the immigration field across European democracies. After all, linking public opinion to elite decision-making is invariably considered an important component of democratic functioning.
The role of immigration
While studies typically highlight macro-economic conditions and public policy, we also examine the role of more social dynamics, such as immigration. It is, after all, reasonable to expect that the public’s feelings of a phenomenon react to (changes in) the actual phenomenon. Despite the continuous domination of political and societal agendas since the 1980s and a nearly universal (exponential) increase in net immigration across Europe, there remains quite some conflicting and anecdotal evidence about how immigration affects immigration opinions.Footnote 5 That is, we know relatively little about how the aggregate public reacts to increasing levels and dissimilar developments of immigration across advanced democracies. More specifically, there remains ambiguity about how citizens respond to immigration, the extent to which they respond, whether they respond to levels or change, and the direction of any such response.
On the most general level, the underlying mechanism may follow a simple rationale. The public is a rational actor, reasonably informed about its social environment (Page and Shapiro, Reference Page and Shapiro1992). Particularly when an issue is salient and widely reported (Franklin and Wlezien, Reference Franklin and Wlezien1997), citizens do not need much information to formulate informed and interpretable opinions about that issue (Zaller, Reference Zaller1992; Sniderman, Hagendoorn and Prior, Reference Sniderman, Hagendoorn and Prior2004; Chong and Druckman, Reference Chong and Druckman2007). Considering the salience and pervasiveness of immigration in the media, it unsurprisingly serves as an input signal (i.e. a cue or heuristic) to update opinions about that exact phenomenon (Igartua and Cheng, Reference Igartua and Cheng2009; Green-Pedersen and Otjes, Reference Green-Pedersen and Otjes2019).
Much like the interpretation of economic conditions, publics only require an equivocal notion of immigration. Even if they often misjudge the actual state of immigration, all segments of the public should – at least – notice ups and downs (cf. Stimson, Reference Stimson1991). That means when there is a change in immigration, regardless of whether it is due to policy changes or external circumstances like government formation, citizens respond in an orderly fashion by adjusting their preferences accordingly. For example, the British public indeed changes opinions towards out-groups in direct response to the actual presence and anticipation of these out-groups (Jennings, Reference Jennings2009).Footnote 6
The directionality of the public’s response remains a complex question and is the object of much debate. Some studies argue the presence of out-groups threatens the material and social interests or increases competition for scarce resources. A larger immigrant population, thus, adds to the perception of (cultural) conflict and increases social tension, thereby stimulating anti-immigration opinions (Sniderman, Hagendoorn and Prior, Reference Sniderman, Hagendoorn and Prior2004; Sides and Citrin, Reference Sides and Citrin2007; Finseraas, Pedersen and Bay, Reference Finseraas, Pedersen and Bay2016). An alternative perspective argues that more opportunities for interaction with immigrants can reduce conflict, thereby promoting tolerance and reducing prejudice. That is, increasing immigration renders citizens’ immigration opinions more favourable (Stolle, Soroka and Johnston, Reference Stolle, Soroka and Johnston2008; Pettigrew et al., Reference Pettigrew2011). A recent study claims truth in both, suggesting that, in the short- to medium-run, macro publics react adversely to high levels of immigration, while this effect cancels out in the longer run (Claassen and McLaren, Reference Claassen and McLaren2021).
Together, these observations clearly suggest that issue-specific opinions directly respond to societal implementations of that same issue and more narrowly defined environmental changes. While citizens monitor and react to societal trends and changes, particularly to salient issues like immigration, the evidence of how citizens respond to immigration remains underdeveloped. That is, do their immigration opinions become more or less restrictive?
Government composition and their positions on immigration
We know from previous research that – in a wide variety of policy domains – policy outcomes also affect public opinion, oftentimes in a thermostatic way (Wlezien, Reference Wlezien1995). While this link has been examined across a variety of issue dimensions, it is typically done as a singular and direct relationship. That is, scholarly accounts often assume – albeit implicitly – that the signposting effect of public policy on public opinion is an unmediated one. According to this rationale, citizens would observe policy changes and subsequently use this, and this alone, as the political signal upon which they base their political demand. While there is a plethora of evidence confirming this mechanism, we argue it is not the only route through which the political system signals its citizens. Politicians and their stated positions also serve this purpose.
Even more, this is a key role that political actors fulfil in contemporary democracies and one that goes to the heart of representation. This is particularly the case for more salient and politically visible issues (Franklin and Wlezien, Reference Franklin and Wlezien1997). Ever since the 1980s, immigration has been a very salient and politicized issue in West European politics. The ensuing political debate has contributed to setting the political agenda, not only in political campaigns and national elections but also in terms of governmental positions and policies. A burgeoning literature approximates immigration positions and salience by relying on different resources, such as media analysis (Kriesi et al., Reference Kriesi2012; Grande, Schwarzbözl and Fatke, Reference Grande, Schwarzbözl and Fatke2019), expert surveys and judgements (Benoit and Laver, Reference Benoit and Laver2006; Bakker et al., Reference Bakker2015; Polk et al., Reference Polk2017), and – most commonly – the analysis of party manifestos (Alonso and da Fonseca, Reference Alonso and da Fonseca2012; Lehmann and Zobel, Reference Lehmann and Zobel2018; Dancygier and Margalit, Reference Dancygier and Margalit2020).
Governmental positions and actual policy outputs both serve the role of political signals for citizens. On the one hand, citizens form their immigration opinions drawing from immigration policies as a very direct and unmediated signpost. On the other hand, citizens also shape their opinions based on what their elected representatives do and say about immigration, thereby being affected in a more intermediated manner. There are several features of the political environment that we can expect to operate as signals for immigration opinion formation. One is the party system’s overall position on the issue of immigration, which reflects both the parties’ policy wishes and their relative electoral strength. So, to give an example, a party system whose “centre of gravity” with respect to immigration leans towards a more restrictive position implies that either all or most of the parties offer policies that are generally more restrictive, or that the parties offering more restrictive policies are electorally successful. In line with this, we could equally anticipate that the success of parties with (strong) anti-immigrant positions would serve this purpose. A more detailed assessment of the political environment, however, would necessarily separate government parties, whose positions on immigration are more likely to be reflected in concrete policy choices, and opposition parties, whose role in the public debate over immigration is often central – and, in many cases, include parties such as those on the far-right, whose positions on immigration can be rather extreme. So as we shall see below, the empirical analyses will take both these aspects into account.
The measurement of immigration opinions across Europe
We update a unique dataset of aggregate immigration opinion measures that allows us to include an unprecedented number of data points (Van Hauwaert, Reference Van Hauwaert2022). It provides a broad selection of weighted immigration-related survey marginals with at least two observations throughout the time frame under analysis.Footnote 7 While the actual number of item series varies between countries – from 24 in Austria to 82 in Germany – the combined immigration opinion measures rely on nearly 2,500 survey marginals (an average of about 200 per country), producing a time series of the public’s support (low values) or opposition (high values) to various aspects of immigration in 13 countries: Austria, Belgium, Denmark, France, Germany, Great Britain, Ireland, Italy, the Netherlands, Norway, Portugal, Sweden and Switzerland.Footnote 8
For each country, we combine the information contained in all variables used into a single measure of immigration opinions, by employing a “dyadic ratios algorithm” (see Stimson (Reference Stimson1991, Reference Stimson2018) for a more extensive and in-depth discussion of the formal estimation procedure). The method relies on calculating the ratio between the share of positions opposed to immigration observed using the same variable in different time points. This procedure is repeated for all variables observed, and it is then averaged over variables and standardized in each country. As Stimson (Reference Stimson2018) shows, this measure produces estimates that are comparable to those produced by more sophisticated techniques (McGann, Reference McGann2014; Caughey, O’Grady and Warshaw, Reference Caughey, O’Grady and Warshaw2019; Claassen, Reference Claassen2020a).
Recent literature highlights that immigration opinions have a unique pattern through time and across countries. Figure 1 provides comparative country-level evidence confirming this for the broader selection of 13 European democracies. Since the absolute values of each measure are dependent on a country’s individual series, we standardize the measures using country-means. We also include a polynomial fitted line for each country, which provides insights into each country’s overall evolution of immigration opinions.
Figure 1. Standardized restrictive immigration opinions, by country. Note: Higher values indicate more restrictive immigration opinions.
Figure 1 shows there is no uniform evolution of immigration opinions across West European democracies. This is illustrated by the variability in slopes of the fitted lines. Citizens in Belgium, Denmark and Germany are almost systematically becoming less restrictive towards immigration (negative slopes). In other countries, such as Ireland, Italy and Great Britain, we observe a more general anti-immigrant trend, or Verrechtsing (positive slopes). It is rare that immigration opinions do not follow a descending or ascending trend over time, but this is, for example, the case in Austria and Norway, where short-term shifts do not seem to add up to broader changes through the years.
A second observation is more nuanced, namely that immigration opinions are cyclical in nature. That means periods of restrictive immigration opinions follow more permissive trends and vice versa. Descriptive statistics further highlight the limited range of average immigration opinions across countries (about 10 points on a 100-point scale), as well as an average standard deviation of only 6.72 across countries.Footnote 9 This indicates the cycles are modest, at best, rather than erratic. Altogether, this signals extensive similarities between immigration opinions across countries, both in terms of levels and patterns of variance.
Independent variables
We include two separate indicators of the short-term economic conditions, namely GDP per capita (in thousands of USD, constant prices, 2010 PPPs) and the unemployment rate (as a percentage of the labour force).Footnote 10 We do so because research shows that people might respond differently to these indicators (Stevenson, Reference Stevenson2001; Erikson, MacKuen and Stimson, Reference Erikson, MacKuen and Stimson2002; Bartle, Dellepiane- Avellaneda and Stimson, Reference Bartle, Dellepiane-Avellaneda and Stimson2011).
Relating policy directly to immigration is more complicated. More restrictive or less restrictive immigration policy (e.g. as indicated by spending) does not automatically indicate an ideological policy position. Additionally, European democracies are restricted in their policy-making abilities regarding immigration because of their participation in the European Single Market. Altogether, this makes finding a meaningful operationalization of immigration policy challenging.
A recent attempt to quantify immigration policy is the Immigration Policies in Comparison (IMPIC) project, which ran an expert survey in 33 countries, including those that are part of our analysis, focusing on the period between 1980 and 2010 (Helbling et al., Reference Helbling2017). The IMPIC data include a general Immigration Policy Index that combines in a single indicator several policy fields such as family reunifications, labour migrants, asylum seekers, and co-ethnics (higher values mean more restrictive immigration policies). Although the IMPIC data covers a shorter period than our public opinion data, we include the index to test the relation between immigration policies and public opinion until 2010. We, therefore, run two sets of models: The first using the full-time series going from 1980 to 2017 and excluding the Immigration Policy Index, and the second model going from 1980 to 2010 including the Immigration Policy Index.
We include, in both models, two indicators of inflow of immigrants. The first is the inflow of asylum seekers. This indicator is used by Jennings (Reference Jennings2009) as a proxy for immigration policy, even though we note this is not correlated with the IMPIC Immigration Policy Index (r = –0.05). Nevertheless, this variable provides an indication of immigration policy interpreted as an immediate political outcome (i.e. inflows), rather than the much more stable immigration regime (i.e. regulations regarding citizenship). Secondly, we include the annual inflow of foreigners as an indicator of the objective and actual immigration population.Footnote 11 We do not distinguish between nationalities or citizenship, as we want to know the effect of overall immigration, rather than public reactions to specific sub-groups. The use of foreign-born inflow as a proxy measure for the immigration rate is fairly common in immigration studies (Semyonov et al., Reference Semyonov2004; Semyonov, Raijman and Gorodzeisky, Reference Semyonov, Raijman and Gorodzeisky2006; Hjerm, Reference Hjerm2007, Reference Hjerm2009; Finseraas, Pedersen and Bay, Reference Finseraas, Pedersen and Bay2016).Footnote 12 We also include a dummy for the migration crisis which has value 1 in 2015, 2016 and 2017, and 0 in all the other years. We include this indicator only in the first model, which excludes the Immigration Policy Index variable, as the latter is missing in all the years in which the crisis occurred.
Finally, we rely on data from the Manifesto Project database (MP) to gauge the political environment. We calculate the position of each party on issues such as nationalism and multiculturalism by means of two components: (1) the share of positive statements about national way of life (category 601.2) plus (2) the share of negative statements about multiculturalism (category 608), divided by the share of negative statements about national way of life (category 602.2) plus the share of positive statements about multiculturalism (category 607). We then take the log of this proportion, following the formula suggested by Lowe and colleagues (Reference Lowe2011). Finally, we take the weighted average of these parties’ positions based on their seat-share to formulate an aggregate index of party system position and linearly interpolate the missing data for non-election years.
Two points are worth discussing. First, taking the categories about national way of life and multiculturalism in the MP data is suboptimal, considering that such categories do not capture parties’ statements on immigration, but on broader topics that arguably include immigration. In fact, other measures of party positions that focus on immigration more directly are available, such as the re-calculation based on manifesto data done by Dancygier and Margalit (DM) (Reference Dancygier and Margalit2020), or expert survey data from the Chapel Hill Expert Survey (CHES) (Jolly et al., Reference Jolly2022). However, both these data sources cover a smaller number of countries and have less time points than our data. The CHES data does not include Norway and Switzerland, and the series starts in 2006, leaving out most of our observations. This is an important limitation, which makes the CHES data difficult to use for our purpose. The DM data covers a longer period; however, they also do not include two countries in our sample, namely Ireland and Portugal, and the time series stops in 2013, leaving out the years of the migration crisis. Moreover, the DM data provides positions on fewer parties than the original MP data. Hence, in our main analyses, we rely on the MP data.Footnote 13
A second point relates to the aggregation of party positions to compute a country-level indicator. The mechanism through which we expect parties to influence public opinion is the politicization of the immigration issue. This implies bringing awareness of a problem in the public discourse (salience), and campaigning to promote a specific solution to that problem (position). A feature of the MP data is that, building on the “saliency theory” of voting, salience and position are correlated (Budge, Reference Budge2015). The proportion between the share of quasi-sentences that are in favour of, say, multiculturalism and those that are against it reflects both how much a given party talks about multiculturalism, and how consistently it does in a positive or negative way. Hence, to calculate the average position of the party system implies observing how salient the issue is among the parties in each context, but also whether such parties are, on aggregate, more positive or negative towards it. We expect this combination of salience and direction to correlate with the average public opinion position.Footnote 14
Empirical results: What moves immigration opinions?
We design a time-series cross-sectional model to estimate the impact of these variables on immigration opinions, where change in immigration opinions at time t is estimated as a function of change in the predictors in the previous year t-1. We model the first differences because a set of augmented Dickey-Fuller tests of the different variables, most notably immigration opinion measures, do not reject the null hypothesis of the presence of a unit root throughout the period under analysis, and this in all countries.Footnote 15 Table 1 presents results from Prais-Winsten regressions, adjusted with correlated panel-corrected standard errors to account for time-invariant spatial dependence structures. We further control for panel-specific autocorrelation (AR1).Footnote 16 All predictors are standardized within country, and we include a set of dummies for the decades.
Table 1 reports several interesting observations. In line with Durr (Reference Durr1993), we notice that the state of the current economy indeed matters. More specifically, when unemployment increases, immigration opinions become more restrictive – indicating that people seek stability and favour the preservation of traditions under those conditions. It suggests that citizens become more sceptical of certain out-groups, or – at least – towards the overall phenomenon of a country absorbing more immigrants when job opportunities become scarce. While we observe this through an increased desire for laissez-faire economics and free market capitalism (Soroka and Wlezien, Reference Soroka and Wlezien2010; Bartle, Dellepiane-Avellaneda and McGann, Reference Bartle, Dellepiane-Avellaneda and McGann2019), it also shows through the promotion of traditional values and national culture (Finseraas, Pedersen and Bay, Reference Finseraas, Pedersen and Bay2016).
Table 1. Public responsiveness for restrictive immigration opinions
Note: Standard errors reported in parentheses; * P < 0.05, ** P < 0.10. We include model prediction plots in Table A.1 of the supplementary materials.
Consistent with widespread findings that policy shapes public opinion, the inflow of asylum seekers has a positive effect across the different models (Jennings, Reference Jennings2009). When the number of asylum seekers increases, immigration opinions become more restrictive. This confirms out-group scepticism, at least in response to border liberalization. Furthermore, we find no evidence that the Immigration Policy Index relates to immigration opinions. This is likely due to the fact that the IMPIC data capture policy regulations that hardly change over time, hence the variance of the index is rather small when compared with the variance of our aggregate policy indicator.
As we would expect, the immigration rate is positive. This suggests that an increasing inflow of foreigners may contribute to more restrictive immigration opinions. This is in line with recent findings arguing for short- and medium-run adverse reactions of the public to higher levels of immigration (Claassen and McLaren, Reference Claassen and McLaren2021).Footnote 17 It is important to note, however, that our findings only hold for the inflow variable, not for the corresponding stock of foreigners (Meuleman, Davidov and Billiet, Reference Meuleman, Davidov and Billiet2009).Footnote 18
Disaggregating the political environment
The role of the political environment as a driver for immigration opinions is negligent across modelling efforts in Table 1. Yet, we could argue that the signal remains obscured and hides cross-national patterns of variance within Europe. With that in mind, we additionally examine whether we can distinguish between the immigration environment as created by government and opposition parties separately. Table 2 presents the corresponding results, like in Table 1, estimated on the full sample and the immigration crisis dummy in model (3), and estimated on a smaller sample but with the Immigration Policy Index in model (4).Footnote 19 Both models include a set of dummies for the decades.
Table 2. Public responsiveness for restrictive immigration opinions, scrutiny of party positions
Note: Standard errors reported in parentheses; * P < 0.05, ** P < 0.10. We include model prediction plots in Table A.2 of the supplementary materials.
Findings are fairly stable throughout models (3) and (4), but equally when compared with results from Table 1. All observations related to the economic and policy variables, as well as the inflow of asylum seekers and foreigners, hold when we disaggregate the political environment. Furthermore, both government and opposition dynamics play a role in explaining immigration opinions. Not only does this confirm that governmental participation shapes how parties can change public opinion, but – even more importantly – it highlights that citizens respond differently to the political positions of government and opposition parties.
More precisely, the political coefficients in Table 2 suggest that citizens’ immigration opinions become more restrictive under two complementary conditions. First, when the immigration environment created by opposition parties becomes more hostile. This is perhaps not all that surprising, considering that most of Western Europe’s far-right parties remain opposition forces, most notably in our sampled countries. Second, citizens respond to their governing elites when it comes to immigration opinions, but they do not follow their lead. On the one hand, we can feel encouraged by this if we think of it as citizens responding thermostatically to governing parties’ immigration signals. This reflects, in a more indirect manner, what previous studies have already shown (Jennings, Reference Jennings2009; Van Hauwaert, Reference Van Hauwaert2022). On the other hand, if we maintain a more suspicious approach, this could mean citizens are purposefully antipodal or antagonistic to their ruling elites on this topic, which – in turn – might be indicative of distrust, scepticism and cynicism towards those elites (Foster and Frieden, Reference Foster and Frieden2017; Bøggild, Reference Bøggild2020). Altogether, it is safe to say that citizens do take their cues from political parties.
Conclusions
As of the second decade of the 21st century, immigration remains one of the most salient and divisive political issues in many representative democracies. It is no surprise, then, that the issue receives considerable attention from public opinion scholars. After all, in democracies, what citizens think and feel is politically relevant and integrating it into politics (and, more precisely, policy) goes to the core of democracy. The goal of the present study is to contribute to this literature by looking at what might explain aggregate immigration opinions and examine this in 13 European countries over a period ranging from 1980 to 2017.
Initial consideration of within- and between-country trends shows a somewhat idiosyncratic pattern within countries, but also a general trend towards more positive opinions towards immigration in the aggregate 13-country pool observed. A more in-depth analysis subsequently sets out to explain the change in immigration opinions on a macro-societal level. We show that four key drivers matter in this regard: Economic conditions (such as unemployment), immigration-related policy (such as a country’s regulated inflow of refugees), the immigration rates and party positions on immigration. Particularly this latter of observation is noteworthy, as we find that people follow political signals from the opposition when they formulate their immigration opinions but not from the government. Rather, citizens proceed countercyclically when responding to ruling elites’ immigration positions.
This study contributes to the ever-growing literature on immigration opinions by presenting a fourfold framework grounded in theory that assesses changes in these aggregate immigration opinions empirically through methodological rigour. A broad-spectrum focus on economic, policy, social and political explanations and the use of a time-series cross-sectional design present a comprehensive understanding of “what moves immigration opinions.” Such an understanding of public opinion formation, related to one of the most salient cultural issues ever, gives us unique insights into the practical functioning of modern-day European democracy. It helps us understand where immigration opinions come from, how they are shaped and what the appropriate avenues are for policy-makers to change them. Naturally, the list of factors affecting macro-level immigration opinions can be much longer than those considered here, but we leave the task of finding additional factors to improve our understanding of what moves immigration opinions to future research endeavours.
Here, it is important to remember that our contributions are different from and complementary to an extensive literature studying individual differences in immigration attitudes. These latter studies tend to focus on micro-level phenomena that have great value when explaining the short-term formation of individual attitudes. However, our macro-perspective allows us to unveil regularities in broad patterns that unfold over longer periods of time and across countries. In short, while individual-level studies are adept to examine differences between individuals (or from the mean), a more aggregate perspective allows us to examine movement and change through time.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1755773925000037.
Acknowledgements
We are grateful for the invaluable comments and feedback we received from the various anonymous reviewers, probably at the expense of other academic tasks that are more easily and explicitly validated in academia.
Open access
