Immigration Threat,Partisanship,and Democratic Citizenship: Evidence from the US,UK,and Germany

Introduction

Politicians and media frequently invoke immigration as a national threat. In the US, President Trump characterized an “asylum caravan” as an “invasion” in the runup to the November 2018 midterm elections, even deploying US troops to the southern border to confront a threat that never materialized. Likewise, in the UK, one of the more visceral adverts during the 2016 EU Referendum (“Brexit”) campaign was Nigel Farage’s Leave campaign poster of Syrian refugees, walking through European countryside, overlaid with red font reading “Breaking Point: The EU has failed us all.” As immigration continues to be a core political issue in Europe (Ford & Jennings, 2020) and beyond, these types of threat narratives proliferate across the political spectrum (Dancygier & Margalit, 2019; Helbling, 2014), though mainstream left parties traditionally adopt more positive positions on immigration than centrist and right-wing parties (Carvalho & Ruedin, 2018).

There is a significant body of work that shows immigration narratives are effective in eliciting a wide variety of attitudes (Hainmueller & Hopkins, 2014) and emotions (Albertson & Gadarian, 2015; Brader et al., 2008a; Dinesen et al., 2016), as well as policy (Goldstein & Peters, 2014; Hellwig & Sinno, 2017) and voting preferences (Lubbers & Coenders, 2017). Generally speaking, citizens are responsive to immigration information (Van Hauwaert & English, 2019), and changing the valence or frame of that information (e.g., culture vs. economic threat; immigrant profiles) can alter attitudes and policy preferences, such as intake levels (Ford & Mellon, 2020) and views on integration (Sobolewska et al., 2017). And since immigration is generally perceived as a type of group threat (Riek et al., 2006), there is evidence that the nature of this outgroup threat stems specifically from cultural, not economic considerations (Dancygier & Donnelly, 2013; Hainmueller & Hiscox, 2010). And here we see broad consensus that immigration threat produces responses that favor sociocultural aspects of ingroup identity, like proficiency in a national language or feeling patriotic (Bonikowski, 2016; Schildkraut, 2005, 2014; Theiss-Morse, 2009; Wright, 2011).

Much of this work is thematically internal—for example, studying the effects of immigration on immigration attitudes—or looks at a subset of sociocultural characteristics about the ingroup, like language or religion. But the question of who “we” are, that is, the attributes and norms of the ingroup of citizens, extends beyond cultural markers. In a democracy, citizens rarely think about what it means to be a “good citizen” as linguistic proficiency or other arbitrary criteria. Rather, as Dalton (2008a, p. 78) defines, norms of citizenship comprise “a shared set of expectations about the citizen’s role in politics.” These include behavior and value commitments or beliefs about community, solidarity, and liberal democratic norms—like abiding by the rule of law and equality. Existing work tells us a lot about how ordinary citizens define norms of citizenship (Bolzendahl & Coffé, 2013; Dalton, 2008a), but we know little about how these definitions are challenged or affected by threat, specifically that of an outgroup. What does an immigration threat do to ordinary citizens? Does threat change their definition of civic obligation, that is, what their role as “citizen” should be in hard times?

This paper studies the effects of immigration threat on democratic citizenship norms. If presented with information about an immigration threat, do ordinary citizens embrace liberal democratic principles or do they veer toward intolerance and other illiberal norms? Do “good citizens” embrace diversity or hunker down and rally around attributes of national belonging? And do certain subsets of citizens respond differently? What role does partisanship play in filtering threat perception? Do only certain partisans view immigration as threatening?

To answer these questions, I embedded a preregistered, vignette experiment in original national surveys, fielded in the summer of 2019 in three countries: the United States, United Kingdom, and Germany. These cases are all advanced democracies, therefore we might expect external threats to have similar effects on a principally universal conception of norms of citizenship. However, they provide useful leverage for a most different design, to consider the role of partisanship across two- and multi-party systems. The cases also exhibit three different immigrant-receiving histories—the US a “classical” immigration country, the UK a former colonial, post-WWII receiver and Germany, a country that accepted over one million asylum seekers in 2015 but, despite decades of guest worker settlement, continued to describe itself as “not a country of immigration” up until the Residence Act of 2004. ¹ These historical narratives may not only affect how individuals perceive immigration threat but also how immigration is engrained as threatening (or not) to norms of citizenship including and beyond norms of belonging.

I find overall that information about an immigration threat leads individuals to change norms of citizenship. Moreover, different patterns emerge when we look directly at partisanship. The largest effects appear on the partisan left. In the US, we see decreased support for behavioral dimensions of good citizenship among Democrats, like the importance of voting and understanding politics. In the UK, we see increased illiberalism and national belonging among Labor supporters. By contrast, German respondents exhibit less partisanship, where norms of citizenship—like understanding politics—are activated across the political spectrum. The substance of these responses varies across case, but the pattern is significant. Immigration threat affects the partisan left, particularly in polarized political systems, while more consensus is exhibited in less polarized, less partisan contexts.

This paper makes two main contributions. First, I illustrate the asymmetrical effects of crisis by partisanship. External threats typically produce a rallying-around-the-flag effect (Albertson & Gadarian, 2015), or differential responses based on emotions, where those that feel anxiety are more likely to seek out information (Gadarian & Albertson, 2014). In the case of immigration threat here, we observe the emergence of an inward-looking, allegiant citizen: decreasing support for liberalism and increasing support for norms of national belonging. However, when we look by partisanship, we see this change is largely taking place on the partisan left. This indicates not only a ceiling effect for moving the partisan right, but also the vulnerability of norms of citizenship in hard times on the partisan left.

Second, I introduce a wider set of measures of democratic citizenship norms that bring together cultural dimensions of national belonging and liberal democratic beliefs, which allows for the exploration of the subtle and conditional effects of immigration threat. In doing so, it marries together two related but distinct literatures on citizenship: the immigration literature, which studies citizenship as a formal institutionalization of national belonging that denotes insiders and outsiders by their access to rights (e.g., Howard, 2009), and the democracy literature, in which political behavioralists examine democratic citizenship norms like participation, active duty, and other behaviors of good citizenship (e.g., Dalton, 2008a, 2008b). Their conceptual disconnection does not mirror empirical changes in the past two decades, where immigrants are increasingly required to demonstrate commitments to liberal norms (Goodman, 2014) and where increasing support by native-born citizens of illiberal practices (Graham & Svolik, 2020; Malka et al., 2020) suggests a taken-for-granted quality of these values to begin with.

This paper proceeds in five parts. First, I define norms of citizenship as attributes of a “good citizen,” which include norms of liberal democratic values, norms of behavior, and attributes of national belonging. Second, I consider how citizenship norms are affected by threat. Included in this section is a specification of an immigration threat frame that distinguishes it from demographic change or personal attitudes. I add to existing accounts by presenting a partisan theory of threat response, arguing citizens interpret norms of good citizenship in response to outgroup threat in line with partisanship. The third section introduces the embedded vignette experiment, the novel measures for citizenship norms, followed by a discussion of my analytical approach. Fourth, I present evidence of asymmetrical norm change. I also include an additional test that shows how norms are affected when respondents express objection to an immigration threat frame. The final section concludes by outlining directions for further comparative research on the attitudinal consequences of immigration threat. My findings troublingly suggest how a depiction of immigration as threatening could be an effective device to depress citizenship norms—and, potentially, civic activity—among a constituency traditionally supportive of immigration.

Citizenship Norms and the Effects of Immigration Threat

Immigrants represent the quintessential outgroup and citizens the quintessential ingroup, but how does a threat of the former affect internal norms of good citizenship in the latter? First some clarification, because citizenship is a multidimensional concept with attitudinal, behavioral, and normative dimensions. Citizenship is a legal status. It is the passport you hold and the political, civil, and social rights you bear as a function of that status (see Marshall, 1950, pp. 41–43). Citizenship is also a membership group, a set of values that foster a sense of belonging, creating identities of insiders and outsiders (see Brubaker, 1992). Finally, citizenship is a type of behavior, what individuals do in a civic capacity (see Dalton, 2008b). In other words, citizenship can be what you have, who you are, and what you do. Immigration studies primarily uses the term “citizenship” to describe the first two, organized around questions of who is eligible for citizenship, why, and how. Therein, the object of these citizenship studies are immigrants that seek, become, or are denied citizenship. In contrast, a separate literature on democracy, including rational choice and comparative political behavior, largely study citizenship by this third meaning—examining the behavior of citizens and what norms guides that behavior. To borrow again from Dalton (2008a, p. 78), these norms of citizenship “tell citizens what is expected of them, and what they expect of themselves.”

Since Riker and Ordeshook (1968) first identified the important, normative role of “citizen duty” to explain voting behavior, there is an accumulation of evidence pointing to norms as a motivation to vote (Blais, 2000; Blais & Achen, 2019). These norms—referred to interchangeably as citizen/civic duty, good citizenship, and obligation—shape political behavior. For example, Bolzendahl and Coffé (2013) present evidence where “well internalized” citizenship norms shape participation, such as voting and membership in associations. Understanding why citizens participate or not is critical for studying democracies, and, therefore, so are the antecedent determinants of those norms.

There is no clear definition of what the ideal constellation of norms of democratic citizenship are. For Dahl (1967), the ideal citizen is minimally knowledgeable and interested in politics. In fact, democratic stability requires citizens who engage in electoral politics. But this leaves out the possibility that good citizens may also be non-participatory. In fact, one of the enduring legacies of Gabriel Almond and Sidney Verba’s The Civic Culture is the observation that democracies require a “blend of activity and passivity,” where “there is political activity, but not so much as to destroy governmental authority; there is involvement and commitment, but they are moderated; there is political cleavage, but it is held in check.” ²

Moving from an institutionally minimal definition of democracy and, therefore, expectations of its citizenry, to a more maximal definition of democracy, a citizen is not merely an informed participant but a holder of liberal democratic values. These include but are not limited to equality, respect for rule of law, and dispositions of mutual tolerance. This suggests two categories of norms: expectations about how to behave (e.g., vote, be informed) and expectations about what to believe.

Yet beliefs or norms about good citizenship do not end at liberal democratic values. They also include items of national belonging, such as speaking the national language and patriotism. Scholars typically do not study these items together. For instance, the International Social Survey Program runs two separate survey batteries reflecting this disjuncture: one titled “Citizenship” and one titled “National identity.” But both exhibit a type of taken-for-grantedness, in which it is assumed the ingroup already adheres to these norms. Moreover, immigrant integration policy is increasingly comprised of requirements to demonstrate commitments to liberal values (Goodman, 2014). By combining liberal norms alongside belonging norms, we can both corroborate findings on the effects of immigration/outgroup threat on norms from larger literatures on tolerance, discrimination, and national identity, as well as extend our knowledge on the effects of immigration threat on norms about behavior and liberal beliefs.

Hypothesizing the Effects of Immigration Threat

I define immigration threat here as a strategic, deliberate framing exercise about immigration by elites. Immigration is a fact of life for advanced industrialized democracies, but framing it as a threat—either the arrival of asylum seekers or competition of jobs and resources—is a strategy designed to stoke fear and mobilize public attitudes, and is commonly employed by political elites (Helbling, 2014; Hellwig & Kweon, 2016) and the media (Brader et al., 2008a; Blinder & Jeannet, 2018; Caviedes, 2015).

This usage is also distinct from general feelings toward immigrants. In other words, we can think of two types of threats: as an attitude or feeling—“I feel threatened” and as a political event (e.g., an asylum caravan). The latter can often produce the former but keeping them distinct enables conceptual precision. This matters because the opposite of a threat feeling is a positive attitude or sentiment toward immigrants, or an idea that immigrants contribute to—instead of take away from—society. Such feelings are frequently captured as a thermostatic measure (e.g., Jennings, 2009) or on an ordinal scale. By contrast, the opposite of a threat event would not be warm feelings but being unaffected by “crisis,” disagreement about event details, or that the event is threatening at all.

The success of negative framing lies in it effectively tapping into a persistent public concern: almost every annual Gallup Poll and Eurobarometer places immigration as a major problem, a topic on which citizens have little accurate details (Scheve & Slaughter, 2001; Sides & Citrin, 2007) and for which accurate information does little to alter opinion (Hopkins et al., 2019). Thus, framing immigration as a threat—invoking imagery of “swarms” of people, a vulnerable border, taking of local jobs and replacement, crime, draining public resources, ineffectual immigration policy, and a government out of control—can mobilize a political base or sway undecided voters toward restriction, consistent with agenda setting theory (Dunaway et al., 2010).

So how might immigration threat affect norms of democratic citizenship? The largest literature on this question examines immigration and national belonging. A consistent finding in this literature is that immigration is associated with exclusive attitudes about national inclusion (Dinesen et al., 2016; Heath & Tilley, 2005; Jeong, 2013; Schildkraut, 2011; Theiss-Morse, 2009; van der Zwet, 2016; Wright, 2011). Since immigration is an inherently “group centric issue” (Nelson & Kinder, 1996) with attitudes rooted in group identity (Citrin et al., 1997), it is conveyed and perceived as a considerable threat to the autonomy of national identity, affecting outcomes like authenticity (Triandafyllidou, 2003) and levels of political trust (McLaren, 2017).

Existing work establishes a fundamentally conflictual relationship between the outgroup—immigrants—and the receiving ingroup—citizens—though it identifies a variety of mechanisms by which this conflict emerges. This ranges from intergroup contact (Enos, 2014; Hopkins et al., 2014; Homola & Tavits, 2018; Newman et al., 2012), resource competition (Dancygier, 2010), to broader observations that draw on social identity theory (Goodman & Alarian, 2019). And to the extent that a threat is framed as external, we expect to see ingrouping and rallying-around-the-flag effects (Albertson & Gadarian, 2015). In short, there are many avenues for how citizens draw the norms of national boundaries (Pérez et al., 2019; Schildkraut, 2011; Theiss-Morse, 2009; Wright, 2011), but the general expectation is that immigration threat increases the importance of national belonging norms, such as speaking the national language or feeling patriotic (hypothesis 1; H1).

Second, we also expect immigration threat to affect normative commitments to liberal democracy. There are, of course, large literatures that show how perceptions of outgroup threat shape prejudice and intolerance (Allport, 1954; Blumer, 1958; Kinder & Sears, 1981; Lajevardi & Abrajano, 2018; Quillian, 1995; Sides et al., 2018; Tesler, 2012) as well as reduce support for fairness (Appleby & Federico, 2018; Wilson & Brewer, 2013). Overall, group threat leads individuals to “hunker down,” preferring their own ingroup over others (Rokeach, 1960; Sullivan et al., 1982). This leads to ingroup siloing and exclusion of outgroups, including immigrants but also extremist and unpopular view-holders (Stouffer, 1955). And where exposure to diversity can moderate tolerance (Marcus et al., 1995), beliefs on their own (that is, without contact or exposure) induce negative attitudes like xenophobia (Hopkins et al., 2019).

In terms of norm changes in response to specific threat, Sniderman et al. (2019) illustrate in a recent study how terrorism causes a depression in tolerance. This dip, however, recalibrates in time (what they label a “perturbation effect”). But terrorism is, by its nature, an unexpected, external event while an immigration threat can be manufactured, framed, and timed for political purposes, like the migrant caravan approaching the US-Mexico border. If everyday citizens view an immigration threat as something more systematic and long-term (like demographic change), these effects may be more enduring. In balance, I hypothesize immigration threat decreases support for liberal democratic norms, like tolerance, patience, and support for diversity (H2).

Finally, moving beyond beliefs, how might immigration threat affect norms of citizen behavior? Note that this question looks at the effect of threat on norms (in this case, norms of behavior) and not actual political behavior, like voting or mobilization. While norms of behavior are strong predicters of actual behavior (Bolzendahl & Coffé, 2013), I maintain a sharp focus on norms in order to build as complete a picture of “good citizenship” as possible.

To develop a hypothesis for norms of behavior, it is instructive to pivot to work in political psychology that focuses directly on threat. In times of high threat, news consumption increases as people take steps to mitigate and avoid threats to their physical health and safety (Althaus, 2002). Gadarian and Alberston (2014), for example, observe that when citizens experience anxiety from threat, they seek information. MacKuen et al. (2010) differentiate behavioral responses based on emotions; anxious people seek out information, while those who experience anger use information to confirm their priors. Beyond valuing information-gathering, anxious individuals also participate in other forms of political action when confronted with immigration frames, like contacting representatives (Brader et al., 2008b). Building from these insights, I hypothesize that immigration threat increases the importance of political activity (H3), like being informed and other aspects of participation, like associational life. I have no specific theoretical expectations for why immigration threat might increase support for the importance of voting that is not filtered through political identity, so I flag this as inductive. This, however, does raise the potential of the important moderating role for partisanship, where I now turn.

The Role of Partisanship

The choice to frame immigration as a threat is a fundamentally political one. Therefore, we need to consider the role of political identities in filtering and responding to information on immigration. This necessitates examining how individuals interpret threat—and thus define norms of good citizenship—according to their dominant political identity: partisanship.

Partisanship is primarily how citizens participate in politics. It is not just the political party one identifies with at election time, the type of policy platforms one supports, or the box one checks on voter registration. Partisanship is a social identity (Huddy et al., 2015) that is increasingly tied to other important identities (Mason, 2018; White et al., 2014) and even personality type (Hetherington & Weiler, 2018). Partisanship guides which social groups are allies (Huddy et al., 2015), which groups you should avoid (Finkel et al., 2020; Iyengar et al., 2012), what leaders you should listen to (Lenz, 2013) and vote for (Hetherington, 2001), and also, increasingly, who you are (Mason, 2018).

This argument is constructed upon social identity theory (Tajfel & Turner, 1979), which suggests that when a group’s salience increases, people begin to identify with it and use it to guide behavior (Tajfel et al., 1971). Group identification leads to group differentiation, and ultimately ingroup favoritism. Accordingly, a number of studies use SIT to show how national identity (Goodman & Alarian, 2019) and patriotism (de Figueiredo & Elkins, 2003), for example, shape outgroup attitudes. Thus, in a polarized political environment (Levendusky, 2009), where party identification becomes more salient, citizens rely on partisanship as a heuristic for attitudinal and behavioral cues. And where citizens perceive party polarization, they are more likely to become more partisan (Lupu, 2015).

Much of these insights about partisanship stem from the US case, where political polarization is a fundamental feature of American politics (McCarty et al., 2016). Yet, plenty of comparative evidence locates the US at one end—or sometimes in the middle—of a spectrum, where polarization (McCoy & Somer, 2019) and negative partisan affect (Boxell et al., 2020; Gidron et al., 2020) exhibits scale variation. This underlines the importance of examining partisanship in comparative perspective, to take into account relevant institutional contexts, like a two-party system, that amplifies polarization and ingroup identification, ³ or a multiparty system, that may diffuse partisan strength through consensus-based institutions, by minimizing differences between winners and losers (Anderson & Guillory, 1997). ⁴ A diverse body of work shows multiparty system with diverse political discourse can lead to social tolerance (Dunn et al., 2009) and high issue diversity can diffuse contestation and conflict and, therefore, increase coalition stability (Greene, 2017).

We know that “partisan pressure,” generated by ingroup social pressure, determines civic duty and impacts voter turnout (Fieldhouse et al., 2020). Partisanship also generates its own kind of political obligations (Bonotti, 2012) and, specifically, increases polarization over national symbols (Satherley et al., 2018). Thus, we expect partisanship not only dictates that citizens will respond to threat according to party identification but also that these responses will differ between parties given differences in pre-existing attitudes toward immigration (Knoll et al., 2011). In the US context, existing research provides some initial support for this expectation (Jerit & Barabas, 2012). Hopkins (2014) shows that reactions to bilingual education differ among American whites based on partisanship. Abrajano and Hajnal (2017) find that anger reduces trust for Democrats and increases trust in Republicans. Research in Europe corroborates the moderating effects of partisanship on immigration (e.g., Bechtel et al., 2015; McLaren, 2001).

These insights support looking directly at partisanship as a moderator for threat perception on norms of citizenship. It also supports a cross-national research design to gain leverage on the effect of immigration threat on democratic citizenship norms more generally. Thus, this final hypothesis argues the effect of immigration threat on norms of citizenship is moderated by partisanship (H4).

Data and Methods

To test these hypotheses, I embedded a pre-registered survey experiment into three, nationally representative surveys conducted over the summer of 2019 in the United States, United Kingdom, and Germany. ⁵ In the US, the survey firm YouGov fielded the survey between June 21 and July 9, resulting in a total sample of 1,599 participants. For Germany and the UK, participants were recruited through the survey firm Respondi. The Germany survey was fielded between August 6 and 14 and the UK survey between August 6 and August 16, collecting 1,470 and 1,350 respondents, respectively. ⁶ Sample sizes were determined by budgetary considerations, and to sufficiently power a test of four hypotheses across 14 dependent variables. The timing of the surveys in summer 2019 coincided with negative news headlines in each of these cases. In the US, the main story was child separation along the U.S.-Mexico border, while in Europe, the summer months bring the perennial “crisis”: Europe-bound migrants making the perilous Mediterranean crossing from Libya.

The logic of case selection allows for two types of comparison. First, these cases represent most similar systems, in that they are all immigrant-receiving. advanced democracies. This enables us to see general effects of immigration threat and democratic citizenship norms, and whether norms exhibit any national variation. Second, we see most different systems when narrowing in on partisanship and institutional context. The US is a winner-take-all, two-party system, structuring politics as a series of zero-sum contests. The UK pairs a similar electoral system with a multiparty system, theoretically reducing the strength of partisanship in shaping outgroup attitudes. And Germany, with a multiparty and proportional representation system, structures politics as a series of positive sum contests, maintaining a system requiring coalitions, and consensus. Differing immigration histories may also shape the context in which immigration is perceived to shape—or undermine—national citizenship norms.

Since each country surveyed has experienced immigration crises, I am able to employ a standardized vignette across cases, maximizing control across the experiment. One concern is that public perceptions of the target—in this case immigrant group—may vary across time and place. Research shows members of the public hold different conceptions about who an immigrant is (Blinder, 2015), and different types of migrants are associated with different threats (Hellwig & Sinno, 2017). For example, Americans may be thinking more about undocumented migrants, while Europeans may be thinking about Syrian refugees (or Muslims generally). However, the terminological differences matter little compared to the frame (Merolla et al., 2013) and opinion in some European countries does not depend on race or ethnicity of immigrants (Scheve & Slaughter, 2001). Therefore, I use “immigration” as an empty signifier, allowing for cross-national difference in interpretation but keeping the threat frame the same in all three cases.

Participants were recruited for online participation through a stratified random sampling strategy, with quotas for age, gender, and geographic area. ⁷ Once recruited, respondents were randomly assigned into one of two groups: the immigration treatment or control. ⁸ For each of the three countries, the external threat vignette read as follows:

Western democracies have received unprecedented numbers of immigrants in recent years, including [COUNTRY]. Many leaders in [COUNTRY] argue immigration is out of control. They believe that [COUNTRY] is experiencing an immigration crisis, as this increasing immigration leads to strains on public services and housing shortages, increasing unemployment and, in some instances, violence. Eventually these immigrants may even become citizens and vote in national elections.

Many experts fear immigration is a threat to our country. As the chief researcher at the National Policy Institute (a leading, non/bipartisan think tank) states, “We cannot control immigration; it creates a strain for everyone and it is changing who we are. We cannot have cohesion because we lack shared values. Immigration is a crisis for our country.”

The outcomes of interest are norms of citizenship. To capture norms of behavior and value commitments, I require an expansive set of items. The ISSP Citizenship module (run in 2004 and 2014) provides a useful starting point, as it asks a set of questions which tap into select behavior and beliefs about good citizenship. For each item, the question reads: “There are different opinions as to what it takes to be a good citizen. As far as you are concerned personally on a scale of 1 to 7, where 1 is not at all important and 7 is very important, how important is it. . .” and then includes several items, from voting to helping people. I use seven items from this battery that span behavioral and liberal norms: (1) obey laws and regulations, (2) votes in elections, (3) keeps watch on the government, (4) active in social or political associations, (5) understand the reasoning of people with other opinions, and (6) help people worse off than yourself.

To add national belonging, I look again to ISSP questions, this time pulling from the National Identity panel (run in 1995, 2003, and 2013). Respondents in these surveys were asked: “Some people say that the following things are important for being truly [NATIONALITY]. Others say they are not important. How important do you think each of the following is. . .” and then includes eight different items, including birth in the country, speaking the national language, and respecting political institutions and laws. For national belonging, I select two achievable attributes: (7) feels [nationality] and (8) speaks [national language].

Further, I add original items on liberal value commitments: (9) maintain friendship or ties with people with different opinions, (10) accept people of diverse backgrounds, (11) have patience, recognizing sometimes your side wins and sometimes loses, (12) understand how government and politics work. Last, I add two questions about attitudes toward government directly, reflecting the roles that system attitudes and civic activism play in constructing good citizenship: (13) support the actions of government, (14) protest or dissent when you disagree with actions of government.

Together, this yields 14 indicators for norms of citizenship. Varimax-rotated factor analysis instructs these items load onto these three dimensions for the US and UK, although some items loaded on these dimensions in surprising ways (e.g., watching the government is considered a norm about behavior, similar to voting, and not a liberal commitment; obey the law loads with norms of allegiance and patriotism, not as an expectation of behavior). ⁹ Germany only produced two, distinct dimensions (behavioral norms and liberal value commitments). For this reason, I use the factor analysis dimensions as a heuristic for organizing and presenting the results—items are presented in rows that reflect these three dimensions—but not in analysis (e.g., as an index of items). Thus, the dependent variable in each model is one of the 14 questions norms of citizenship (“How important is it for a good citizen to. . . .?”). ¹⁰ Higher values correspond to more support for that norm. To test H4—whether partisanship affects how individuals think about democratic citizenship under threat—I include a standard measure of party identification per country.

In what follows, I present statistical analyses to test the effect of immigration crisis on civic norms. The baseline (OLS) models regresses a binary indicator for the immigration treatment T on outcome Y. To improve efficiency, I also estimate extended statistical models that adjust for the following pre-treatment variables: age, race (or, in the case of Germany, immigrant-origin), gender, region, and level of education. ¹¹ Adjustments follow the recommendations in Lin (2013) with every model taking the form of a regression of Y on T, X, and T * X, where X is a set of mean-centered covariates ( X i − X ¯ ) . I estimate robust standard errors throughout. Each figure presents coefficient plots for the covariate adjusted effects of receiving the immigration treatment on norms for each of the 14 items. ¹² These figures plot the treatment effects estimated from the OLS regressions in the extended models. The dot represents the regression coefficient, and the horizontal line maps its 95% confidence interval. Each dot thus compares respondents who received the treatment to the control group. Recalling that the dependent variable varies along a scale of 1-5, the x-axis measures the difference in average responses between the treatment group and the control group.

Results

I find strong differences in how the partisan left versus right interpret citizenship norms in the face of immigration crisis, both within and across cases. Effects are primarily absorbed on the partisan left. This asymmetrical effect illustrates not only how movable the partisan left is in response to immigration framing but how little the right has to move, suggesting close agreement with threat framing. Further, where we see similarities between the US and UK, we observe a different pattern of response among German respondents, where partisanship is weaker and institutional design decreases the likelihood that policy positions become associated with social identity. Finally, we see the substance of (primarily) left partisan responses differ across cases, where European respondents move toward illiberalism and stronger national identity norms in a way their American counterparts do not.

Beginning with general, aggregate results, Figure 1 shows results for the United States (diamond point; top row of each plot), the UK (dot point; middle row), and Germany (triangle; bottom row). We see that immigration threat affected respondents in aggregate, but in different ways. Beginning with the US, we see two significant effects: Americans are less likely to value voting and less likely to value helping people (both significant at the 95% level). The decline in solidarity is consistent with a general response of intolerance (H2) while a decreased importance in voting is evidence inconsistent with H3. Similarly, a decreased importance on associational life and supporting diversity corroborate this initial impression (though significant at the 90% level). There is no effect of immigration threat on items of national belonging among Americans in aggregate. As for the remaining 10 items, it is useful to note here that the absence of significance does not mean these norms of citizenship are unimportant, just that they are not significantly affected by information on immigration threat.

OPEN IN VIEWER

Figure 1.

Immigration threat treatment.

Immigration threat in the UK affected a larger number of items, with respondents more likely to embrace illiberalism and norms of national belonging. On liberal values, British respondents are less likely to value understanding others, support diversity, and solidarity/helping others (in support of H2). Regarding norms of national belonging, the effects on all four items—obey the laws, support government, feel British, speak English—are significant and positively signed (consistent with H1). This finding corroborates the vast literature on intolerance and national belonging in Britain (e.g., Heath & Tilley, 2005; McLaren, 2017), and the role of “us versus them” ethnocentrism in shaping political choices (Sobolewska & Ford, 2020). Last, we observe a decreasing importance of behavior norms in the UK, where respondents express decreasing importance of voting and protests in response to immigration threat—an important finding given that mass protests against Brexit would erupt in London 3 months after the survey was fielded.

Germany also shows strong aggregate effects of immigration threat on citizenship norms, but in some diverging patterns from the US and the UK. Immigration threat did not depress support for solidarity, nor did it decrease support for behavioral norm. In fact, Germany is the only case where aggregate respondents were more likely to increase the value of a behavioral norm (“understanding politics”) in response to immigration threat (H3). However, like the UK, we see evidence of illiberal effects of immigration threat, with declining support for diversity but also patience (H2). Finally, like the UK, we see strong effects of immigration threat on items of belonging, including obeying the law, feeling German, and speaking German (H1).

To summarize the aggregate findings, all three cases show threat decreases norms of tolerance. Only Germany exhibits an increase in support for behavioral norms (H3), and only the European cases show aggregate evidence that immigration threat increases national belonging norms (H1). On the one hand, the modest effects of the immigration threat treatment overall—and of national belonging in particular—among US respondents may seem surprising, given the bulk of research on immigration threat is conducted in the US context. However, the comparative perspective sheds light on degrees of potential xenophobia in which the US—a traditional immigrant receiving country—pales in comparison to attitudes in Western Europe.

An alternative interpretation of these results is that the prime did not effectively simulate a threat to the respondents. However, the aggregate picture does not account for the extent to which immigration has become a partisan issue and, therefore, how individuals use partisanship as a heuristic for how to “make sense” of an immigration threat. A closer analysis supports this intuition, that these modest results mask important treatment effect heterogeneity by party identification. I examine these case by case.

United States

Moving from the aggregate level to look at partisanship, we see a more nuanced picture. Figure 2 presents the effects of immigration threat in the United States only, separating respondents according to their partisan identification. Here we see a central finding. First, the main results from the aggregate results—decreased support for the importance of voting and solidarity (significant at the 99% and 95% level, respectively)—are driven by the partisan left. In other words, in the face of an immigration threat, Democrats think voting and helping others are less important to being a good citizen. We can add to this a third norm—decreased support for understanding politics (also at the 95% level). Other items approach significance for the left (decreased importance in joining associations) and right (decreased solidarity, watches government) but none past the 90% threshold. Importantly, no items of national belonging are significant, either in aggregate or by party ID. This leads us to accept hypotheses 2 (on liberal values) and 4 (on differential effects by partisanship) and reject hypotheses 1 (on increased salience of national identity) and 3 (on increased support for political behavior norms).

OPEN IN VIEWER

Figure 2.

Immigration treatment by partisanship, US.

These overall findings show that only Democrats are significantly affected by immigration threat, ¹³ but in a surprising way. Immigration threat decreases support for behavioral norms—where a good citizen is an active citizen. One possibility is that immigration threat may not trigger the type of behavior-inducing anxiety that the political psychology literature predicts but instead triggers a sense of helplessness (“it’s a big problem, what can I do?”) and, therefore, diminished sense of efficacy.

Further, the lack of significance among Republicans does not mean these items are not important, just that they are unaffected by immigration threat. This is likely the result of a ceiling effect; that is, their beliefs already run close to the vignette, meaning that there is not much to move. But it is worth flagging the substantive finding from this experiment, that immigration threat does not significantly activate items of national belonging, either in aggregate or by partisanship. Instead, other items of “good citizenship,” like norms of behavior and liberal values, are instead affected. This suggests immigration threat has wide-reaching implications but also that some values may be insulated from manufactured shock. I flag these insights as we move forward to look at the other cases.

United Kingdom

Turning to the UK, we see a broadly similar pattern to the US, in that the partisan left is overwhelmingly affected by immigration threat compared to the partisan right. Figure 3 shows, like US Democrats, immigration threat depresses the value of behavioral norms (voting and protesting) as well as solidarity (“helps people”) for Labor supporters. But in a marked contrast to the US, we see a much stronger response that moves Labor supports on 10 out of 14 items. Specifically, in addition to the above-mentioned items, Labor respondents express less support for understanding others, more impatience, less supportive of diversity and significantly more nationalistic (along all four items of national belonging). In short, support decreases for liberal norms and increases for norms of national belonging (consistent with H2 and H1).

OPEN IN VIEWER

Figure 3.

Immigration treatment by partisanship, UK.

In a second similarity to the US, we see a more muted response from the partisan right in comparison to the left, where Conservative supporters express less support in having friends of different political opinions and accepting diversity (both significant at the 95% level). A similar pattern can be seen when the political cleavage changes from partisanship to Brexit position, where Remainers (i.e., those who voted to remain in the EU) look like Labor (the partisan left) on national belonging items. Likewise, Leavers (i.e., those who voted to leave the EU) look like Conservatives (the partisan right) when it comes to norms of citizenship, in that the only significant effect of immigration threat is on decreasing support for diversity. This, like US Republicans, may suggest a ceiling effect for moving norms among those on the partisan right. ¹⁴ On the other hand, individuals that support the United Kingdom Independence Party (UKIP), Farage’s far right party, show increased support for speaking English and decreased support for diversity, indicating that the ceiling for Conservatives may not yet have been reached.

Germany

Looking at Germany requires a more nuanced approach than a binary lens of partisan left and right. In addition to multiple parties on the left and right, parties also vary in size. Partisanship is weaker where electoral institutions produce centripetal effects for major parties, while smaller parties are pushed to extremes (Calvo & Hellwig, 2011). I look at four major parties: the coalition-organizing Christian Democrats (along with their Bavarian sister party, the Christian Social Union), the Social Democrats (SPD), the far-right Alternative for Germany (AfD), and the Greens. This approach may reduce the size of effects by splitting respondents into more categories, but it provides a more accurate view of the effect of threat on norms of citizenship.

In Figure 4, we see immigration threat experienced differentially across the political spectrum. The aggregate finding showed Germans as the only set of national respondents where immigration threat had a positive effect on a behavioral norm—understanding politics. At the partisan level, this finding is visible on both the left (the Greens) and the right (the CDU/CSU). Neither respondents on the left nor the right become specifically more illiberal in face of immigration threat though, at the aggregate level, both patience and support diversity norms were negative and statistically significant (at the 99% level). It may be that strong aggregate significance gets dissipated across multiple parties, or these norms are not affected by partisanship. By partisanship, we only see SPD voters are more likely to value understanding others. Finally, like in the UK, we see the strongest effects on the two items explicitly about national belonging. In aggregate, “feeling German” and “speaking German” are significant, and at the partisan level, the latter is significant for every partisan identity except the AfD.

OPEN IN VIEWER

Figure 4.

Immigration treatment by partisanship, Germany.

This evidence leaves two, non-mutually exclusive possibilities. First, examining by partisanship reduces the sample size by group to the point that we do not have sufficient power to observe treatment effects. Where we see aggregate evidence of a norms change but not at the partisan level (e.g., patience, support diversity), this may be one explanation. Second, partisanship may not be a strong heuristic for either interpreting threat or defining norms of citizenship. Because of institutional design, partisanship in Germany is comparatively low, which is one reason for why voter polarization can benefit extremist parties (Ezrow et al., 2014) and some of the only significance we see at the party level is among supports of smaller and newer political parties.

What if Respondents Reject the Threat Frame?

Just because a frame is used does not mean its successful. A narrative can be ineffective, as I report in the many areas of statistical insignificance above, or narratives can be rejected if they do not, for example, line up with an individual’s partisan position or other priors. This section considers if a respondent disagrees with the immigration threat treatment, and whether we observe differences between objectors and non-objectors. To test for treatment objection, respondents were given 30 seconds to “discuss the topic” they just read in an open-ended response following the treatment and control. On a functional level, this served to reinforce the vignette by encouraging them to concentrate for an extended period of time on immigration as a democratic threat. But these open-ended responses also provide me with an opportunity to assess how respondents reacted to the treatment. Open-ended responses were coded as objection if they disagreed with the facts of the frame or the premise that immigration was a threat. ¹⁵ To test the conditional hypotheses that these effects only obtain among respondents who do not object to the treatment, I subset the data, running separate analyses for the subset of respondents who objected to the treatment and for the subset who did not. I only interpret results for objectors, as non-objectors is a residual and heterogeneous category that ranges from those that agreed and those that were persuaded to those that did not know they could express objection.

This test reveals important findings among objectors. Beginning with the US, Figure 5 portrays norms of citizenship among those who objected to the treatment and those who did not.

OPEN IN VIEWER

Figure 5.

Immigration treatment by objection, US.

First, we can confirm there were differential treatment effects, where 31% of the total treatment group expressed objection. Among Republicans, 11% rejected the treatment, while 39% of Democrats and others did. Objectors indicate decreasing importance for “watches the government,” increasing importance for “supports diversity” and decreased support for both “feeling American” and “speaking English” as attributes of a good citizen. This objection analysis shows a sharper image of the kind of citizen unpersuaded by immigration threat frames.

In the UK, objection was similarly large (at 27%) but produces substantively different results to the US, where objectors continued to embrace all four items of national belonging (Supplemental Appendix F). And in Germany, objection among Germans was the lowest across the three cases (5% of total, or 36 respondents). This invites us to be cautious. It may be that German respondents do not view open-ended instructions to “discuss” as an invitation to disagree. It may also reflect a type of immigration “fatigue,” where the EU did not declare the Syrian Refugee Crisis as “over” until May 2019. However, among those that objected, we see broad support for norms like always voting, understanding politics, joining associations, as well as understanding others and having friends with different political opinions (Supplemental Appendix G).

Discussion

Norms of citizenship are not static. Citizens are responsive to threat and interpret norms and expectations of “good citizenship” accordingly. While the definition of being a “good citizen” has always been historically contingent and context dependent (Goodman, 2020), contemporary norms changes in response to immigration threat are instructive. This manuscript shows immigration threat leads citizens to devalue norms of behavior, become more intolerant, and hunker down on national belonging. We see important differences cross-nationally. While US changes were largely confined to behavioral norms, both European cases exhibited strong rallying effects around national belonging and a depression in liberal value commitments. Taking a wide-angle view of citizenship norms to include behavior, liberal values and national belonging allows us to see past the immediate and expected effects of immigration threat to observe how widespread its reverberations can reach.

Moreover, that citizenship norm changes predominately occur on the partisan left reveals the uneven impact of an immigration threat. As a depiction of an immigration threat may reaffirm and align with preexisting beliefs of individuals on the right, it is also persuasive in altering norms of citizenship for those on the left. As such, a threat frame could be an effective device to depress the norms that would otherwise induce engagement and participation (Bolzendahl & Coffé, 2013), especially among a constituency traditionally more supportive of immigration. That threat is filtered through partisanship is worrying on its own; civic duty and attitudes toward obligation are distinct from partisan identification. The identity and status of “citizen” precedes and supersedes partisan factions. That in a polarized environment, it affects only one group or side of partisans, raises deep concerns that a democratic citizenry could jointly define—much less confront—shared problems.

There are important differences across cases worth further consideration. Above all, we see three different immigration experiences shape three different understandings of threat and its impact on citizenship norms. It may be that the long, engrained presence of the immigrant narrative in US citizenship (Aptekar, 2012) inoculates it from more ephemeral threat frames. Levendusky (2018) even finds a shared sense of American national identity may temper affective polarization. By comparison, the past two decades in European citizenship politics have been preoccupied with defining citizenship in a way that can incorporate immigrants through integration, but does not feature their immigrant status as part of a shared national identity (Goodman, 2014). Immigration politics in the United States is, of course, highly effective and often invoked as a racialized wedge issue (Abrajano and Hajnal, 2017), but it’s effects on citizenship norms appear to be limited. By contrast, as Europe continues to grapple with immigration, identity, and the politics of ethnocentrism and the far right, citizenship is evolving alongside.

Future work may build on this work in several directions. First, researchers may be interested in identifying specific and potentially different mechanisms that drive civic norms acquisition and diffusion. Insights from social psychology may be useful in this regard. Second, we could consider how norms translate to behavior. What are the consequences if liberal value holders are behaviorally inactive, or illiberal value holders active? Third, researchers may want to examine more comprehensively how a wider range of threats affect the broad contours of democratic citizenship. Immigration is only one of many issues. Terrorism (Berinsky, 2009; Merolla et al., 2007) and foreign electoral interference (Tomz & Weeks, 2020) strongly influence public opinion and are filtered through partisanship. But how do these affect norms of citizenship? The answers to these questions matter enormously. Understanding how individuals interpret civic obligation in hard times—from economic recessions to the global, COVID-19 pandemic—help diagnose how governments rebuild economies, institutional trust, and other elements of vibrant advanced democracy. That these strategies may need to overcome partisan barriers only make rebuilding more difficult. It may also be the case that other social identities effectively moderate the effect of threat on democratic citizenship. This is a fourth avenue for future research. In particular, we know that race and ethnicity play a critical role in shaping national identity (Carter & Pérez, 2016), but gender, education, religion and other demographics may also play an important role in shaping citizenship norms, and shaping citizen response to threat.

To conclude, these findings illustrate how threat frames can damage important values and attitudes of democratic citizens. In an age of democratic recession (Diamond, 2015), where advanced democracies prove vulnerable to populist and authoritarian power-grabs, the malleability of a citizen’s sense of responsibility and obligation may prove critical to either offset or accelerate further erosion. Thus, as citizens are susceptible to restrictive message framing by media and elites, an effective immigration threat can outlive the “crisis” itself in the changes it makes to quotidian citizen norms. We do not know how long these changes last, but that may be irrelevant. The strategic timing of a crisis and an election can capitalize on even ephemeral effects and leverage them against democracy.

Supplemental Material