In 2014, Hainmueller and Hopkins published a landmark review of the literature on public attitudes toward immigration in the Annual Review of Political Science (Hainmueller & Hopkins, 2014, ARPS). Their review synthesized approximately 100 studies from more than two dozen countries, organizing the evidence around two broad theoretical traditions: political economy approaches emphasizing material self-interest and sociopsychological approaches emphasizing group-related attitudes and symbolic concerns. Their central finding was striking in its clarity: self-interest hypotheses had repeatedly failed to find empirical support, while sociotropic concerns—particularly cultural concerns about immigration's perceived effects on the nation as a whole—emerged as the most consistent predictors of restrictionist attitudes. The review offered seven conclusions that effectively defined the field's state of knowledge and its most pressing unanswered questions.
In the twelve years since, the study of immigration attitudes has been transformed. The original review has accumulated over 3,000 citations and spawned an enormous body of follow-up research. A conservative count identifies more than 1,800 papers citing the review, and the pace of publication has accelerated rather than slowed. The field has grown not merely in volume but in methodological sophistication, geographic scope, and substantive depth.
Three developments, in particular, make an updated review both timely and necessary.
First, the political landscape has changed dramatically. The 2015 European refugee crisis, the Brexit referendum of 2016, the election of Donald Trump, and the rise of populist parties across democracies placed immigration at the center of electoral politics in ways that were only beginning when Hainmueller and Hopkins wrote. These events did not just generate new data; they generated new questions about the stability of attitudes under extreme conditions, the role of elite cues in activating latent predispositions, and the relationship between immigration attitudes and vote choice.
Second, the field has undergone a methodological revolution. Conjoint experiments, which Hainmueller and Hopkins themselves helped pioneer (Hainmueller & Hopkins, 2014, AJPS), have become the dominant tool for studying immigrant preferences, with landmark studies extending the design across fifteen countries (Bansak, Hainmueller & Hangartner, 2016, Science). Natural experiments exploiting exogenous variation in immigrant exposure have provided causal evidence that was largely unavailable a decade ago (Hangartner, Dinas, Marbach, Matakos & Xefteris, 2019, APSR). Panel data have enabled researchers to track within-person attitude change over extended periods, yielding the striking finding that immigration attitudes are remarkably stable over time (Kustov, Laaker & Reller, 2021, JOP). And multiple meta-analyses have synthesized evidence across hundreds of studies, providing more reliable estimates of key relationships than any single study could offer (Kaufmann & Goodwin, 2018, SSR; Dražanová et al., 2023, JEMS; Weber et al., 2025, BJPS).
Third, the geographic scope of the literature has expanded substantially. While Hainmueller and Hopkins focused primarily on North America and Western Europe, recent research has examined immigration attitudes in the Middle East (Alrababa’h et al., 2020, CPS), South Asia (Gaikwad & Nellis, 2016, AJPS), East Asia, and other regions. This expansion has tested whether findings established in wealthy Western democracies generalize to different political and economic contexts.
This review updates the Hainmueller and Hopkins framework by surveying the post-2014 literature. I retain the original review's organizing distinction between political economy and sociopsychological approaches, while adding sections on topics that have emerged or matured since 2014: the role of contact and local conditions, partisanship and elite cues, attitude stability and change, and the political consequences of immigration attitudes. The review covers research published between 2014 and early 2026, drawing primarily on work in political science, sociology, economics, and psychology.
The remainder of the review is organized as follows. Section 2 takes stock of what has changed since 2014, briefly updating each of Hainmueller and Hopkins's seven conclusions and describing the transformed political and methodological landscape. Section 3 reviews new evidence on political economy approaches, including labor market competition, education, and fiscal concerns. Section 4 covers sociopsychological approaches, including sociotropic threats, ethnocentrism, emotions, national identity, and the role of media and information. Section 5 examines contact, context, and local conditions—an area where natural experiments have provided especially important advances. Section 6 addresses partisanship, ideology, and elite cues, filling what Hainmueller and Hopkins identified as one of the field's most significant gaps. Section 7 covers attitude dynamics—stability, change, and salience. Section 8 reviews the post-2015 refugee attitudes literature. Section 9 examines the conjoint revolution. Section 10 surveys innovations in measurement and research design. Section 11 explores the geographic expansion of the field. Section 12 offers updated conclusions and directions for future research.
Hainmueller and Hopkins (2014) concluded their review with seven propositions about the state of the field. A decade of subsequent research has largely confirmed their core assessments while adding significant nuance to several and substantially filling the gaps they identified.
Conclusion 1: Labor market competition as a "zombie theory." Hainmueller and Hopkins described the labor market competition hypothesis—the idea that natives oppose immigration primarily because immigrants compete with them for jobs—as a theory that "has repeatedly failed to find empirical support." Subsequent research has largely sustained this verdict. The most notable refinement comes from Pardos-Prado and Xena (2018, AJPS), who show that skill specificity—the degree to which workers' skills are tied to particular occupations—matters more than general skill level, suggesting that labor market vulnerability operates through narrower channels than the original factor-proportions model implied. Yet even this refinement confirms the broader point: general measures of labor market position remain weak predictors of immigration attitudes (see Section 3.1).
Conclusion 2: Education's robust association with pro-immigration views. The positive correlation between education and support for immigration remains one of the field's most durable findings. What has changed is the availability of causal evidence. Cavaillé and Marshall (2018, APSR) exploit six compulsory schooling reforms in five Western European countries (Denmark, France, Great Britain, the Netherlands, and Sweden) as natural experiments, finding that additional years of education causally reduce anti-immigration attitudes. Lee (2023, JOP) further demonstrates that it is the content of education—specifically, exposure to ideas about national identity and diversity—rather than years of schooling alone that matters (see Section 3.2).
Conclusion 3: Sociotropic theories need better differentiation. Hainmueller and Hopkins noted that sociotropic theories—emphasizing concerns about immigration's effects on the nation—were less clearly differentiated from one another than self-interest theories. The post-2014 literature has made progress here, with cross-national conjoint experiments enabling sharper tests. Valentino, Soroka, Iyengar et al. (2017, BJPS) conduct parallel experiments across multiple countries and find that cultural concerns consistently outweigh economic ones as drivers of immigration opposition (see Section 4.1).
Conclusion 4: The need for better causal identification. This was perhaps the original review's most consequential recommendation. The field has responded with a wave of natural experiments, panel studies, and experimental designs that have substantially addressed the endogeneity concerns that pervaded earlier cross-sectional research. Hangartner et al. (2019) exploit the quasi-random assignment of refugees to Greek islands to identify the causal effect of exposure. Kustov, Laaker, and Reller (2021) use long-running panel data to demonstrate that immigration attitudes are stable within individuals over decades, a finding that casts doubt on many proposed "causes" of attitude change. Vertier, Viskanic, and Gamalerio (2022, PSRM) leverage the dismantling of the Calais Jungle camp as a natural experiment (see Sections 4 and 5).
Conclusion 5: Prejudice and ethnocentrism need clearer experimental manipulation. Research has responded by decomposing ethnocentrism into more specific components. Helbling and Traunmüller (2018, BJPS) disentangle Islamophobia from generalized anti-immigrant sentiment, demonstrating that opposition to Muslim immigrants specifically reflects both generalized ethnocentrism and group-specific concerns. Konitzer, Iyengar, Valentino, Soroka, and Duch (2018, JEMS) similarly distinguish ethnocentrism from group-specific stereotypes (see Section 4.2).
Conclusion 6: Information environments require better measurement. The original review noted that elite rhetoric and media coverage played important theoretical roles but had been measured only coarsely. Post-2014 research has moved on two fronts. On one hand, Hopkins, Sides, and Citrin (2019, JOP) find that providing accurate information about immigrants has "muted consequences" for attitudes—factual corrections alone do not substantially shift views. On the other hand, research on elite discourse suggests that how political leaders frame immigration has considerable power to shape public opinion (Flores, 2018, Social Forces; Schmidt-Catran & Czymara, 2022, JEMS). The asymmetry between informational and rhetorical effects is one of the most important post-2014 findings (see Sections 4.5 and 6.2).
Conclusion 7: Immigration attitudes are disconnected from partisanship research. This was perhaps the gap most dramatically filled in the intervening years. Immigration has become a—if not the—defining issue of partisan conflict in many democracies. Reny, Collingwood, and Valenzuela (2019, POQ) demonstrate that immigration attitudes were the strongest predictor of vote switching toward Trump in 2016, more powerful than economic anxiety. The partisan sorting of immigration attitudes is now a major area of research (see Section 6.1).
The political landscape in which immigration attitudes are studied has changed fundamentally since 2014. Three overlapping disruptions reshaped both the substance and the urgency of this research.
The 2015 European refugee crisis brought more than one million asylum seekers to Europe in a single year, concentrated heavily in a few frontline states. This created natural experiments that the field had long lacked: researchers could study the effects of sudden, large-scale refugee arrivals on host communities' attitudes. The crisis also generated a distinct subfield of refugee attitude research with its own theoretical and empirical dynamics, as the characteristics that shape preferences over refugees (persecution, vulnerability, integration prospects) differ in important ways from those that shape preferences over economic migrants (skills, labor market contributions).
The Brexit referendum and the 2016 U.S. presidential election demonstrated that immigration attitudes could reshape democratic outcomes. Both events prompted intensive research into the relationship between immigration concerns and voting behavior. Schwartz, Simon, Hudson, and van Heerde-Hudson (2020, BJPS) find that the Brexit vote itself softened anti-immigrant attitudes among both Leave and Remain supporters, suggesting that high-salience political events can have complex, non-obvious effects. Carreras, Carreras, and Bowler (2019, CPS) provide evidence that economic distress in communities predicted support for Brexit, though the mechanism ran partly through immigration concerns.
The rise of populist parties and leaders across democracies prompted debate about whether a "global populist wave" was underway. Art (2020, POP) challenges the notion of a unified global populism, arguing that populist movements in different countries respond to distinct national conditions. The relationship between populism and immigration attitudes proved more complex than initially assumed—populist success appears to be as much a consequence of the politicization of immigration as a cause of it (see Section 6.3). The refugee crisis reshaped party systems across Europe, as established parties struggled to respond and new competitors—notably the AfD in Germany, the Sweden Democrats, and others—capitalized on the salience of immigration (Mader & Schoen, 2018, WEP). These political disruptions were not merely background context for attitude research; they created new variation to study and new questions to ask about the relationship between attitudes, political behavior, and institutional outcomes.
The Brexit referendum has become the canonical case of this entanglement. Hobolt (2016, JEPP) shows that the Leave vote divided the winners and losers of globalization, with anti-immigration sentiment among the strongest individual-level predictors of voting to exit the European Union, a result that turned immigration attitudes from a dependent variable into one of the most studied electoral forces of the decade.
The methodological toolkit available to researchers studying immigration attitudes has expanded considerably since 2014.
Conjoint experiments have become the field's workhorse design for studying preferences over immigrant characteristics. Building on Hainmueller and Hopkins's (2014, AJPS) introduction of the method, Bansak, Hainmueller, and Hangartner (2016) fielded conjoint experiments in fifteen European countries, demonstrating a striking cross-national consensus: respondents across all fifteen countries preferred asylum seekers who were younger, had higher employability, and came from more vulnerable circumstances. This "hidden consensus" across diverse European publics was one of the decade's most influential findings.
Natural experiments have provided the causal evidence that Hainmueller and Hopkins called for. Hangartner et al. (2019) exploit the quasi-random routing of refugee boats to different Greek islands to show that exposure to refugees increases hostility and support for restrictionist policies. Vertier, Viskanic, and Gamalerio (2022) leverage the dismantling of the Calais Jungle refugee camp to study how the sudden removal of a refugee presence affected attitudes in nearby communities. Schaub, Gereke, and Baldassarri (2020, CPS) study refugees quasi-randomly assigned to communities in eastern Germany.
Panel data have enabled researchers to move beyond cross-sectional snapshots. Kustov, Laaker, and Reller (2021) analyze multiple waves of panel surveys spanning over two decades in the United States, finding that immigration attitudes are remarkably stable within individuals—comparable in stability to partisanship itself. This finding has important implications for research on attitude change: if attitudes are stable over decades, then many purported "causes" of attitude change may instead reflect compositional shifts or measurement artifacts.
Meta-analyses have become feasible as the literature has grown large enough to support systematic quantitative synthesis. Kaufmann and Goodwin (2018) conduct a meta-analysis of studies on ethnic diversity and social cohesion. Dražanová et al. (2023) provide an individual-level meta-analysis of immigration attitudes across European countries. Weber et al. (2025) offer a comprehensive meta-analysis of attitudes toward migrants and displaced persons. These syntheses provide more reliable effect size estimates than any single study and help identify the boundary conditions under which different findings hold.
The Hainmueller and Hopkins review focused primarily on North America and Western Europe. Post-2014 research has expanded to new regions, testing whether the established findings generalize. Gaikwad and Nellis (2016, AJPS) study attitudes toward internal migrants in Mumbai, India—a context in which the "immigrants" are fellow citizens moving between states—and find that perceived economic and cultural threat operate similarly to cross-national immigration contexts. Alrababa’h et al. (2020) study Jordanian attitudes toward Syrian refugees, demonstrating that contact and economic competition dynamics are present in middle-income, non-Western settings. Ariely (2019, JEMS) examines how collective memory shapes immigration attitudes in Israel. These studies collectively suggest that the core dynamics—the dominance of sociotropic over egocentric concerns, the importance of cultural threat perceptions, the weakness of simple labor market competition models—are not unique to wealthy Western democracies, though specific contextual factors moderate their relative strength.
Political economy approaches explain immigration attitudes through the lens of material self-interest: natives oppose immigration when they perceive it as threatening their economic position, whether through labor market competition, fiscal burden, or broader macroeconomic effects. Hainmueller and Hopkins (2014) concluded that this tradition had produced theories with admirable precision and testable predictions but that the empirical evidence had largely failed to support them. Post-2014 research has mostly confirmed this assessment, while adding refinements that suggest economic considerations operate in more conditional and indirect ways than the original models proposed.
The labor market competition hypothesis—derived from the factor-proportions model in trade theory—predicts that natives whose skills substitute for those of immigrants will oppose immigration, while those whose skills complement immigrants' will support it. Hainmueller and Hopkins described this as "something of a zombie theory" given its repeated empirical failures. The post-2014 evidence has not resurrected it in its original form, but it has identified more specific channels through which labor market concerns may operate.
The most important refinement comes from Pardos-Prado and Xena (2018), who argue that the standard approach of using education or occupation as proxies for labor market vulnerability is too coarse. They show that skill specificity—the degree to which a worker's human capital is tied to a particular occupation or industry, and therefore difficult to transfer—predicts immigration attitudes better than general skill level. Workers with highly specific skills that are difficult to redeploy are more opposed to immigration, presumably because they face higher adjustment costs if displaced. This finding rescues a version of the labor market competition logic, but one that applies to a narrower segment of the workforce than the original theory implied.
Other evidence continues to weigh against the general labor market competition model. Jeannet (2018, Research & Politics) exploits the transition to retirement as a natural experiment: if labor market competition drives immigration opposition, then workers should become less opposed to immigration upon leaving the labor force. Jeannet finds no such effect, providing a clean null result for the competition hypothesis. Valentino, Soroka, Iyengar et al. (2017) conduct parallel survey experiments across multiple countries and consistently find that cultural considerations outweigh economic ones in shaping immigration preferences—a finding that holds across countries at very different levels of economic development and immigrant composition.
Hopkins, Margalit, and Solodoch (2023, BJPS) provide further evidence against simple economic self-interest accounts. They examine the effects of personal economic shocks—job loss, income declines—on immigration attitudes and find that these shocks have limited and inconsistent effects. Economic conditions at the personal level, in short, are not reliable drivers of immigration opposition.
Laaker (2023, BJPS) introduces an important temporal dimension by showing that economic conditions during individuals' formative years—late adolescence and early adulthood—shape their immigration attitudes in adulthood. This "impressionable years" finding suggests that the relationship between economics and immigration attitudes operates through socialization rather than contemporaneous self-interest, consistent with the broader finding that immigration attitudes are remarkably stable over the adult life course (Kustov, Laaker & Reller, 2021).
Newman and Malhotra (2019, JOP) complicate the picture from a different angle, demonstrating that economic reasoning about immigration is often racialized. In their account, economic arguments against immigration function partly as vehicles for racial attitudes—the "racial hue of economic reasoning." This suggests that even when respondents articulate economic objections to immigration, the underlying driver may be group-based attitudes rather than material self-interest.
A separate strand of the economic literature has examined how globalization-related shocks and technological change shape immigration attitudes. Colantone and Stanig (2018, APSR) show that regions in the United Kingdom more exposed to import competition from China were more likely to vote for Brexit, with immigration concerns serving as a mediating channel. In a companion study, Colantone and Stanig (2018, AJPS) extend this finding across fifteen Western European countries, demonstrating that exposure to trade globalization shocks increases support for nationalist and radical-right parties. Importantly, these effects operate through perceived cultural and identity threats rather than direct labor market competition, reinforcing the pattern in which economic disruptions fuel anti-immigration sentiment via sociopsychological channels. Wu (2022, PSRM) documents a related mechanism—misattributed blame—showing that natives may hold immigrants responsible for job losses that are actually caused by other economic forces such as trade liberalization. Gamez-Djokic and Waytz (2020, Psychological Science) find a similar pattern with automation: concerns about technological job displacement increase opposition to immigration, suggesting that immigration serves as a more psychologically available target for diffuse economic anxieties.
The weight of the evidence thus confirms Hainmueller and Hopkins's original assessment while refining it. Labor market competition in its general form remains poorly supported as a driver of mass attitudes. Where economic considerations do operate, they tend to work through narrow channels (skill specificity), formative socialization (impressionable years), globalization and automation shocks that are misattributed to immigration, or in ways entangled with cultural and racial attitudes (racialized economic reasoning).
The positive association between education and pro-immigration attitudes was one of the most consistent findings in the pre-2014 literature, but its interpretation remained contested. Education correlates with many things—higher income, different social networks, cosmopolitan values, knowledge about immigration—making it difficult to isolate the operative mechanism.
The most important post-2014 advance is the identification of a causal effect. Cavaillé and Marshall (2018) exploit compulsory schooling reforms in France and the United Kingdom, which increased the minimum years of schooling for cohorts born after specific cutoff dates. Using a regression discontinuity design, they find that additional years of education causally reduce anti-immigration attitudes. This result moves the discussion beyond correlation and establishes that education itself—not merely the characteristics of people who happen to be more educated—contributes to pro-immigration views.
But what aspect of education matters? Lee (2023) addresses this question by examining the content of education rather than just its quantity. Lee finds that educational experiences that expose individuals to ideas about inclusive national identity and cultural diversity are particularly important in shaping pro-immigration attitudes. The implication is that the "education effect" is not simply about cognitive sophistication or labor market position but about the specific ideas and values transmitted through educational institutions.
These findings have implications for the debate between political economy and sociopsychological explanations. If the education effect operated primarily through labor market channels—educated workers facing less competition from immigrants—then it should be strongest among workers in occupations where immigrants are concentrated. Instead, the evidence points toward education's role in shaping values, cultural attitudes, and conceptions of national identity, placing the education effect squarely within the sociopsychological tradition.
The fiscal burden hypothesis—the idea that natives oppose immigration because immigrants consume public services and require redistributive spending—received mixed support in the pre-2014 literature. Post-2014 research has continued to produce varied findings, but with important developments regarding the relationship between welfare states and immigration attitudes.
Hjorth (2015, EUP) examines welfare chauvinism—the desire to restrict social benefits to co-nationals—and finds that it is structured by national stereotypes about specific immigrant groups' work ethic and cultural compatibility. Opposition to immigrant welfare access is not a blanket response to fiscal concerns but is targeted at groups perceived as undeserving or culturally distant.
Levy (2021, JOP) challenges what she terms the "immigrationization" thesis—the claim that immigration has fundamentally altered welfare state politics by undermining solidarity. She finds that the link between immigration and welfare state retrenchment is weaker than commonly assumed, suggesting that fiscal concerns about immigration are often overstated as drivers of both attitudes and policy.
Magni (2022, AJPS) provides experimental evidence that information about immigrants' economic contributions—specifically, their net positive fiscal impact—can increase support for welfare provision to immigrants. This finding suggests that fiscal concerns, to the extent they exist, are partly based on misperceptions about immigrants' actual economic contributions and can be moderated by accurate information.
Allen et al. (2023, IMR) extend this line of research, showing that economic evidence about immigration's fiscal and labor market effects can change attitudes when presented credibly. The effect is modest but robust, suggesting that the informational environment surrounding fiscal debates matters for attitude formation.
The fiscal burden literature thus presents a nuanced picture. Pure fiscal self-interest—"immigrants cost me money through higher taxes"—is not a strong driver of mass attitudes. But welfare chauvinism, understood as a belief that immigrants are undeserving of public benefits, is widespread and structured by group-specific perceptions. Information about immigrants' actual economic contributions can moderate these concerns, though the effects are typically modest.
Sociopsychological approaches emphasize that immigration attitudes are shaped less by material self-interest than by concerns about how immigration affects valued social groups—the nation, the ethnic or cultural in-group, or the community. Hainmueller and Hopkins (2014) concluded that these approaches had consistently outperformed political economy models but remained internally heterogeneous and insufficiently differentiated. The post-2014 literature has sharpened some of these distinctions while revealing new psychological mechanisms.
The dominance of sociotropic over egocentric concerns remains the field's most robust finding. Citizens evaluate immigration based on its perceived effects on the country as a whole, not on their personal economic situation. The post-2014 literature has strengthened this conclusion through cross-national experimental evidence and sharper tests.
Bansak, Hainmueller, and Hangartner (2016) provide the most ambitious test to date. Their conjoint experiment, fielded in fifteen European countries during the refugee crisis, shows that respondents in all fifteen countries share similar preferences over asylum seekers: they favor applicants who are more employable, have experienced more severe persecution, and are more compatible with the host country's religion and language. The consistency of these preferences across countries with very different economies, immigrant populations, and political contexts is striking. Critically, humanitarian concerns—the severity of persecution—independently shape preferences even after controlling for economic and cultural attributes, suggesting that sociotropic evaluation incorporates both national interest and humanitarian considerations.
Valentino, Soroka, Iyengar et al. (2017) conduct parallel experiments in five countries and find that cultural concerns consistently outweigh economic concerns as predictors of immigration attitudes. The finding holds across countries at very different levels of development and with very different immigration histories, suggesting that the cultural dimension of sociotropic evaluation is not an artifact of any particular national context.
Hopkins, Margalit, and Solodoch (2023) further demonstrate the limits of egocentric explanations (see Section 3.1). The contrast between the weakness of egocentric and the strength of sociotropic predictors has only grown sharper since 2014.
Hainmueller and Hopkins (2014) noted that prejudice and ethnocentrism consistently predicted immigration attitudes but that the field disagreed about whether these attitudes were generalized (directed at out-groups broadly) or group-specific (directed at particular immigrant populations). Post-2014 research has made progress on this question by decomposing ethnocentrism into more specific components and by examining how group-specific evaluations interact with generalized predispositions.
Helbling and Traunmüller (2018) tackle the question of Islamophobia directly, decomposing opposition to Muslim immigrants into generalized anti-immigrant sentiment and religion-specific hostility. They find that both components contribute, but that opposition to Muslim immigrants has a distinctive component not reducible to general ethnocentrism. This matters because it suggests that interventions targeting generalized prejudice may not fully address opposition to specific groups.
Konitzer, Iyengar, Valentino, Soroka, and Duch (2018) examine the relative roles of ethnocentrism (generalized out-group hostility) and group-specific stereotypes in shaping immigration attitudes across countries. They find that both matter but that their relative importance varies across national contexts, suggesting that the generalized-versus-specific debate may not have a single answer.
Flores and Schachter (2018, ASR) introduce the concept of the "social construction of illegality," showing that perceptions of who is an "illegal immigrant" are shaped by racial and ethnic stereotypes. Latinx individuals are more likely to be perceived as undocumented regardless of actual legal status, and this racialized construction of illegality shapes both attitudes and policy preferences. Schachter (2016, ASR) complements this by examining how perceptions of similarity and difference between native and immigrant groups structure evaluative judgments.
Hellwig and Sinno (2016, JEMS) demonstrate that different immigrant groups elicit different types of perceived threat—economic, cultural, or security-related—and that these differentiated threats predict distinct policy preferences. This finding reinforces the argument that immigration attitudes are not monolithic but structured by the specific groups under consideration.
The post-2014 literature thus suggests that ethnocentrism operates at multiple levels simultaneously: a generalized predisposition toward in-group favoritism provides the broad foundation, but group-specific stereotypes and perceptions of particular immigrant populations shape the intensity and character of opposition in ways that generalized measures miss.
The role of emotions in shaping immigration attitudes has received growing attention since 2014, moving beyond the anxiety-focused research that dominated the pre-2014 period.
Renström, Bäck, and Carroll (2023, Political Psychology) examine the relationship between discrete emotions and affective polarization around immigration. They find that anger and fear produce distinct attitudinal profiles: anger is associated with more rigid opposition and affective polarization, while fear is associated with more nuanced and potentially persuadable attitudes. This distinction matters for understanding when and how immigration attitudes become entrenched versus malleable.
Leong, Chen, Willer, and Zaki (2020, PNAS) use neuroimaging to study neural responses to immigrants and immigration-related stimuli, finding that empathic neural responses to immigrants' suffering predict pro-immigration policy preferences. This work connects immigration attitude research to the broader neuroscience of empathy and intergroup relations.
Aarøe, Petersen, and Arceneaux (2017, APSR) provide foundational evidence linking the behavioral immune system—an evolved psychological mechanism that heightens sensitivity to potential sources of contamination—to immigration attitudes. They show that individuals with stronger physiological disgust responses are significantly more opposed to immigration, a relationship that holds even when controlling for ideology and ethnocentrism. This evolutionary psychology account suggests that some anti-immigration sentiment has deep biological roots in pathogen-avoidance mechanisms, a finding with implications for how malleable such attitudes can be. Building on this, Clifford, Erisen, Wendell, and Cantú (2022, PLS) confirm that disgust sensitivity predicts anti-immigration attitudes across multiple countries, consistent with the pathogen-avoidance account.
Giani (2020, CPS) provides an important corrective to the assumption that anti-immigration sentiment is always rooted in prejudice. Studying the impact of jihadist terrorist attacks across six countries, Giani finds that such attacks significantly increase security fear but do not increase ethnic prejudice—a pattern he describes as "fear without prejudice." This distinction has implications for both theory (not all opposition is prejudice-based) and interventions (reducing prejudice may not address fear-based opposition).
Conceptions of national identity—who belongs to the nation, what defines membership, and what obligations members owe one another—have long been recognized as important correlates of immigration attitudes. Post-2014 research has deepened this line of inquiry.
Lee (2023) demonstrates that the content of education matters for how individuals conceptualize national identity. Educational experiences that promote inclusive, civic conceptions of national membership—as opposed to ethnic or ascriptive ones—are associated with more pro-immigration attitudes. This finding connects the education literature (Section 3.2) to national identity research, suggesting a common mechanism.
Williamson, Adida, Lo, Platas, Prather, and Werfel (2020, APSR) examine how family immigration histories shape contemporary attitudes. They find that individuals whose families have their own immigration history are more supportive of immigration, even after controlling for other demographic characteristics. This suggests that personal or familial narratives of migration serve as a powerful frame through which individuals interpret contemporary immigration.
Ariely (2019) studies how collective memory shapes immigration attitudes in Israel, a country with a founding narrative deeply intertwined with migration and displacement. The interaction between collective memory and contemporary immigration politics demonstrates that national identity is not static but is continuously reconstructed in response to political circumstances.
Kustov (2020, JEMS) extends the study of migration attitudes beyond immigration by examining attitudes toward emigration—the departure of co-nationals. He finds that opposition to emigration parallels immigration opposition, with both driven by similar place-based sentiments and concerns about community integrity. This suggests that the attitudes typically labeled "anti-immigration" may reflect a broader attachment to demographic stability and community preservation rather than out-group hostility per se.
Kustov (2020, CPS) develops this insight further through a theory of "parochial altruism"—the idea that voters are genuinely altruistic but direct their concern primarily toward compatriots. Under this framework, opposition to immigration is not driven by selfishness or hostility but by a bounded sense of solidarity: voters who care about their fellow citizens' well-being may oppose immigration that they perceive as contrary to the national interest, even as they exhibit altruistic behavior in other domains. The parochial altruism framework reframes immigration opposition as a product of in-group concern rather than out-group antipathy, with important implications for how researchers conceptualize and measure the motivations behind restrictionist attitudes.
Simonsen and Bonikowski (2022, CPS) examine the moralization of immigration—the process by which immigration attitudes become infused with moral conviction rather than remaining matters of pragmatic evaluation. They find that moralizing immigration is associated with greater attitude extremity, resistance to persuasion, and willingness to take political action. When immigration attitudes become moral convictions, they are particularly resistant to change, which has implications for the stability findings discussed in Section 7.
The identity literature has also turned from out-groups to in-groups. Jardina (2019, Cambridge University Press) shows that white identity—attachment to one's own racial group rather than animus toward others—independently predicts immigration restrictionism among a substantial share of white Americans, establishing that in-group solidarity and out-group hostility are empirically distinct pathways to the same policy preferences.
Hainmueller and Hopkins (2014) noted that information environments and elite rhetoric played central theoretical roles but had been measured only coarsely. The post-2014 period has seen substantial progress on this front, though the findings are more complex than simple "information deficit" models would predict.
The most sobering finding comes from Hopkins, Sides, and Citrin (2018), who conduct a series of experiments providing accurate information about immigrants' characteristics—their numbers, fiscal contributions, crime rates, and so on. The effects on attitudes are, in their word, "muted." Correcting misperceptions does not substantially change immigration preferences, a result consistent with the broader finding in political psychology that factual corrections have limited effects on policy attitudes when those attitudes are rooted in values and group identities rather than in factual beliefs.
This does not mean that the information environment is irrelevant. Czymara and Dochow-Sondershaus (2018, ESR) show that media coverage of immigration shapes public concerns about immigration, with more negative coverage associated with heightened concern. The effect operates not by providing new facts but by making immigration salient and framing it in ways that activate existing predispositions.
Huang (2023, JEMS) documents the extent of misperceptions about immigration flows, finding that citizens in many countries substantially overestimate the number of immigrants. While correcting these misperceptions has limited effects on policy preferences (consistent with Hopkins et al., 2018), the prevalence of misperceptions suggests that the information environment systematically distorts the factual basis on which attitudes might rest.
Allen et al. (2023) provide a partial counterpoint, finding that presenting economic evidence about immigration—carefully framed as credible research findings rather than partisan claims—can shift attitudes. The key may be the perceived source and framing of information rather than the information itself. When economic evidence is presented in a way that respondents find credible and relevant, modest attitude change is possible.
The information and media literature thus suggests a revised understanding of how information environments matter. Simple factual corrections have limited power because attitudes are rooted in values and identities rather than factual beliefs. But media coverage and elite framing shape the salience and emotional valence of immigration as an issue, which in turn activates the value-based and identity-based predispositions that drive attitudes. The distinction between information (facts about immigration) and framing (how immigration is discussed and constructed as an issue) is central to understanding the information environment's role.
How does the actual presence of immigrants in a community affect natives' attitudes? This question lies at the intersection of the contact hypothesis—the idea that interpersonal contact with out-group members reduces prejudice—and the threat hypothesis—the idea that larger out-group populations increase perceived threat. Hainmueller and Hopkins (2014) noted that results on contextual effects were mixed, with local demographics having no fixed influence and broader political context playing a key moderating role. The post-2014 literature has made significant advances, particularly through natural experiments that address the endogeneity problems that plagued earlier observational work.
The most important methodological advance in the study of contextual effects has been the use of natural experiments that exploit exogenous variation in immigrant exposure.
Hangartner, Dinas, Marbach, Matakos, and Xefteris (2019) leverage the quasi-random assignment of refugee arrivals to different Greek islands. Because refugees arriving by boat in the Aegean Sea are routed to different islands based on proximity to their departure point and maritime conditions—factors unrelated to islanders' prior attitudes—the study identifies the causal effect of refugee exposure on host community attitudes. The results are sobering: exposure to refugees increases hostility, support for restrictionist policies, and votes for far-right parties. This finding runs against the simple contact hypothesis and suggests that large-scale, involuntary, and unstructured contact can exacerbate rather than reduce intergroup tensions.
Vertier, Viskanic, and Gamalerio (2022) provide a complementary natural experiment by studying the dismantling of the Calais Jungle refugee camp in October 2016. The camp's inhabitants were redistributed to reception centers across France, creating sudden and exogenous variation in refugee presence across municipalities. The authors find that the arrival of refugees in new communities increased support for Marine Le Pen in the 2017 presidential election, consistent with the threat-based account. The Calais Jungle study is particularly informative because the dispersal was both sudden and exogenous to the receiving communities' prior characteristics.
Schaub, Gereke, and Baldassarri (2020) study refugees who were quasi-randomly assigned to communities in eastern Germany. Their findings are more nuanced: on average, they find null effects of refugee exposure on anti-immigrant attitudes, which they interpret as supporting a sociotropic perspective. Beneath these null aggregate findings, however, they observe convergence: local exposure to refugees pulled both right- and left-leaning individuals toward the center, suggesting that direct contact may moderate extreme attitudes in both directions.
Alrababa’h, Dillon, Williamson, Hainmueller, Hangartner, and Weinstein (2020, CPS) extend this natural experiment approach beyond Europe by studying Jordanian attitudes toward Syrian refugees. Jordan received a large influx of Syrian refugees in a short period, creating variation in exposure across communities. The authors find that economic competition concerns and contact dynamics operate in Jordan similarly to European contexts, suggesting that the mechanisms identified in Western settings are not culturally parochial. The study also highlights how hosting refugees in a middle-income country creates distinct pressures related to labor market access and public service provision.
Several additional studies round out the natural experimental evidence. Enos (2014, PNAS) provides one of the earliest causal demonstrations, using a randomized field experiment in which Spanish-speaking confederates were assigned to commuter rail platforms in Boston-area communities. Brief, repeated exposure to visibly different out-group members increased exclusionary attitudes among white riders, establishing that even incidental contact can activate anti-immigrant sentiment. Dustmann, Vasiljeva, and Damm (2019, Review of Economic Studies) exploit the quasi-random dispersal of refugees across Danish municipalities in the 1980s and 1990s, finding that exposure to refugees during childhood and adolescence increased support for anti-immigration parties in adulthood—a long-run socialization effect consistent with the formative-years framework discussed in Section 7.5. Steinmayr (2021, Review of Economics and Statistics) introduces a crucial distinction between transitory exposure and sustained contact. Studying the 2015 refugee movement through Austria, Steinmayr finds that communities where refugees merely passed through became more hostile, while communities that actually hosted refugees—enabling structured, sustained interaction—became more supportive. This finding helps reconcile apparently contradictory results in the exposure literature by showing that the nature and duration of contact matter as much as its occurrence. Tabellini (2020, Review of Economic Studies) extends the temporal frame dramatically by studying immigration to American cities during the Age of Mass Migration (1910--1930), finding that immigration triggered political backlash—increases in anti-immigrant legislation and support for restrictionist politicians—even as it generated positive economic effects. The coexistence of economic benefits and political hostility in Tabellini's historical analysis underscores the sociopsychological primacy documented in the contemporary literature.
These natural experiments have substantially advanced the field beyond the mixed results of observational studies, but they also reveal an important asymmetry: involuntary, large-scale, and unstructured exposure tends to increase hostility, while structured, cooperative contact tends to reduce it. The conditions under which contact occurs matter as much as the fact of contact itself.
While the natural experiments described above study the effects of aggregate exposure, another strand of research examines whether targeted interpersonal contact can reduce prejudice toward immigrants.
The most influential post-2014 study on prejudice reduction is Kalla and Broockman (2020, APSR), who study "deep canvassing"—extended face-to-face conversations in which canvassers encourage voters to share personal narratives related to the topic at hand. They find that ten-minute doorstep conversations about transgender rights durably reduced exclusionary attitudes, and they extend this approach to immigration, finding similarly promising results. The effect persists for months, distinguishing it from the fleeting effects of most persuasion interventions. The mechanism appears to operate through perspective-taking and analogic reasoning rather than factual correction.
Adida, Lo, and Platas (2018, PNAS) study perspective-taking more directly, using an experimental intervention that asks participants to adopt the perspective of an immigrant or refugee. They find that perspective-taking reduces exclusionary attitudes, though the durability of the effect beyond the immediate experimental context is uncertain.
Homola and Tavits (2017, CPS) introduce an important moderator: the effects of contact with immigrants depend on the ideological orientation of the contact-maker. Contact reduces prejudice among individuals who are already predisposed toward tolerance (left-leaning), but has weaker effects—or even backfire effects—among those who are ideologically opposed to immigration (right-leaning). This finding suggests that contact is not a universal prejudice-reduction tool but operates conditional on pre-existing predispositions.
Laurence and Bentley (2018, SSR) provide further evidence of countervailing contact effects, showing that different types of contact can produce different attitudinal outcomes. Positive, cooperative contact reduces prejudice, while negative or competitive contact reinforces it. In diverse communities, individuals may experience both types simultaneously, producing the mixed aggregate effects found in many observational studies.
The experimental record now has a meta-analytic anchor. Paluck, Green, and Green (2018, BPP) synthesize the 27 randomized contact interventions with outcomes measured at least a day after treatment and find that contact generally reduces prejudice, but with notably weaker and less reliable effects for ethnic and racial prejudice among adults—precisely the domain most relevant to immigration attitudes. The optimistic headline number for contact, in other words, rests least securely where this review needs it most.
A persistent challenge in studying the effects of local diversity on attitudes is distinguishing compositional effects—differences in who lives where—from contextual effects—the independent influence of the local environment on residents' attitudes.
Maxwell (2019, APSR) makes an important contribution by demonstrating that cosmopolitan attitudes in diverse urban areas are primarily compositional rather than contextual. That is, diverse cities do not make their residents more tolerant; rather, tolerant people are more likely to move to and remain in diverse cities. This selection effect has been widely suspected but difficult to demonstrate convincingly. Maxwell's study, which leverages within-person panel data and migration histories, provides strong evidence that much of what appears to be a contextual "diversity effect" on attitudes is actually a sorting effect.
Danckert, Dinesen, and Sønderskov (2016, POQ) find that political sophistication moderates the relationship between local immigrant exposure and attitudes. More politically sophisticated individuals—those who pay attention to political information and think about politics in ideological terms—are less affected by local demographic change, potentially because their attitudes are anchored by broader ideological commitments rather than responsive to immediate environmental cues.
Laurence, Schmid, and Hewstone (2018, JEMS) examine the relationship between diversity and social cohesion at multiple geographic scales, finding that the effects of diversity depend on the level of analysis. Neighborhood-level diversity may have different effects than city-level diversity, and individual-level contact may have different effects than aggregate exposure. This scale sensitivity helps explain the apparently contradictory findings in earlier literature.
Kaufmann and Goodwin (2018) conduct a meta-analysis of the diversity-cohesion literature and find that the aggregate relationship between ethnic diversity and reduced social cohesion is modest. The effect is real but smaller than many influential single studies had suggested, and it is moderated by a range of contextual factors including inequality, residential segregation, and the type of social cohesion being measured (trust, participation, sense of belonging).
How do local and national economic conditions shape immigration attitudes? The post-2014 evidence suggests that economic context matters, but not in the straightforward way that simple self-interest models predict.
Carreras, Carreras, and Bowler (2019) examine the role of economic distress in driving support for Brexit. They find that communities experiencing economic decline—as measured by indicators such as unemployment and income stagnation—were more likely to vote Leave, but the mechanism operated partly through heightened immigration concern. Economic distress increased the salience of immigration as a threat, which in turn predicted Leave voting. The finding illustrates how economic conditions can shape immigration attitudes indirectly, by making immigration more salient and threatening, rather than through direct material self-interest.
Laaker (2023) demonstrates that economic conditions during formative years have long-lasting effects on immigration attitudes. Individuals who came of age during economic downturns hold more negative immigration attitudes decades later, even after their personal economic circumstances have improved. This "impressionable years" effect is consistent with socialization theories that emphasize the lasting imprint of early-life conditions on political attitudes and represents an important challenge to models that assume contemporaneous economic conditions drive attitudes.
Choi, Poertner, and Sambanis (2019, PNAS) study parochial norms and discrimination, finding that communities with stronger norms of parochial cooperation—favoring in-group members—are more likely to discriminate against immigrants. Economic scarcity can activate these parochial norms, creating an indirect pathway from economic conditions to discriminatory behavior that operates through social norms rather than individual self-interest.
Hainmueller and Hopkins (2014) identified the disconnection between immigration attitude research and the study of political partisanship as one of the field's most significant gaps. Their seventh conclusion called for making this relationship "a central issue moving forward." The post-2014 literature has responded emphatically. Immigration has moved from the periphery to the center of partisan conflict in many democracies, and research on the partisan dynamics of immigration attitudes has grown rapidly.
The degree to which immigration attitudes have become sorted along partisan lines is one of the most consequential political developments of the past decade. In the United States, immigration has shifted from an issue that cross-cut partisan lines to one that is strongly predicted by—and, some argue, constitutive of—partisan identity.
Reny, Collingwood, and Valenzuela (2019) provide the most direct evidence for the political centrality of immigration attitudes. Analyzing vote switching between the 2012 and 2016 presidential elections, they find that immigration attitudes were the strongest predictor of switching from Obama to Trump—more powerful than economic anxiety, racial resentment considered in isolation, or evaluations of the economy. The finding was replicated across multiple datasets and model specifications, establishing immigration attitudes as a first-order driver of contemporary American electoral behavior.
Ollerenshaw and Jardina (2023, POQ) document asymmetric polarization on immigration: while both parties' supporters have moved apart on immigration, the movement has been larger among Republicans. This asymmetry has implications for the representation of immigration preferences in the political system, as the party that has moved further holds more intense and electorally motivating views on the issue.
Goodman (2021, CPS) examines the relationship between perceived immigration threat and democratic attitudes, finding that individuals who perceive high immigration threat are more willing to support illiberal policies and leaders. This suggests that the partisan sorting of immigration attitudes is not merely a matter of issue positioning but has implications for democratic norms and institutions.
Dancygier and Margalit (2019, CPS) trace the evolution of the immigration debate across Western democracies, showing that the debate has shifted from primarily economic framing (immigrant workers taking jobs) to primarily cultural and identity-based framing (immigrants threatening national identity and social cohesion). This discursive shift corresponds to the empirical dominance of cultural over economic explanations discussed in Sections 3 and 4.
The two defining electoral shocks of 2016 each acquired a touchstone study. For the United States, Mutz (2018, PNAS) uses panel data to argue that status threat rather than personal economic hardship explains the shift to Trump among 2012-to-2016 vote switchers, a claim that has drawn methodological challenge but framed the subsequent debate. For Britain, Goodwin and Milazzo (2017, BJPIR) show that local increases in immigration and the sentiment that immigration is out of "control" predicted Leave voting, connecting contextual change to the referendum outcome.
How political elites talk about immigration shapes how the public thinks about it. The post-2014 literature has moved beyond this general proposition to identify specific mechanisms through which elite discourse operates.
Flores (2018) examines the effects of elite speech on immigration attitudes using a quasi-experimental design. He finds that anti-immigration rhetoric from prominent political figures—particularly from those perceived as authoritative on the issue—increases public opposition to immigration. The effect operates by legitimizing restrictionist sentiments that may have been suppressed by social norms, effectively shifting the "Overton window" of acceptable discourse.
Schmidt-Catran and Czymara (2022) study elite discourse and attitude polarization across European countries. They find that polarized elite discourse about immigration—where parties take sharply divergent positions—is associated with greater attitudinal polarization among the public. When elites send consistent, moderate signals, public attitudes cluster near the center; when elites diverge, the public follows them apart. This finding underscores the role of partisan elites in structuring mass attitudes.
Czymara (2019, IMR) provides a focused case study of elite discourse about Muslim immigration in Germany, showing that political leaders' framing of Muslim immigrants influences public attitudes toward this specific group. The study illustrates how elite cues can be group-specific—not just pro- or anti-immigration in general, but targeting particular immigrant populations.
Simonsen and Bonikowski (2022) examine the moralization of immigration in elite discourse, finding that when political actors frame immigration in moral terms—as a matter of right and wrong rather than costs and benefits—public attitudes become more extreme, more resistant to change, and more politically consequential. Moralized immigration attitudes are associated with greater willingness to vote on the issue and greater intolerance of opposing views.
Hellwig and Kweon (2016, WEP) study party cues more directly, examining how voters update their immigration attitudes in response to their preferred party's positions. They find that partisan cue-taking is significant: when a voter's party shifts its position on immigration, the voter tends to follow, suggesting that elite positioning is a cause, not merely a reflection, of mass attitudes. This finding helps explain the rapid partisan sorting observed in recent years: as party leaders have taken more polarized positions on immigration, their supporters have moved accordingly.
The rise of populist and far-right parties across democracies has been one of the defining political developments of the post-2014 period, and immigration is widely seen as central to their appeal. The research literature has sought to understand this relationship more precisely.
Art (2020) challenges the common narrative of a "global populist wave," arguing that populist movements in different countries respond to distinct national conditions rather than constituting a unified phenomenon. This is an important corrective to accounts that treat populist success as a simple function of immigration opposition. While immigration is a common theme across populist parties, the specific form and intensity of the immigration-populism connection varies across countries.
Halikiopoulou and Vlandas (2020, EPSR) examine the support coalitions behind far-right parties, finding that these coalitions are more heterogeneous than commonly assumed. Different types of voters are drawn to far-right parties for different reasons—some primarily for immigration restriction, others for economic protection, others for cultural conservatism. This heterogeneity has implications for how we understand the causal role of immigration attitudes in far-right support: immigration is part of the story but not the whole story. Stockemer, Halikiopoulou, and Vlandas (2020, JEMS) further document the heterogeneity of far-right voter bases across European countries, reinforcing the point that a single "far-right voter" profile does not capture the diversity of these electorates.
Schwartz, Simon, Hudson, and van Heerde-Hudson (2020) provide a counterintuitive finding: the Brexit vote actually softened anti-immigrant attitudes among both Leave and Remain supporters. Among Leave voters, a greater sense of control over immigration channeled the effects of the vote onto attitudes; among both groups, efforts to distance oneself from accusations of xenophobia contributed to the softening. This suggests that populist victories can have paradoxical effects on the attitudes they were expected to inflame.
Mader and Schoen (2018, WEP) study how the 2015 refugee crisis reshaped party competition in Germany. They find that the crisis moved immigration to the center of political conflict, benefiting parties that took strong positions on immigration (particularly the AfD) and disadvantaging parties that were perceived as ambiguous. The finding illustrates how exogenous shocks can reshape the partisan landscape by elevating new issues.
Brosius, van Elsas, and de Vreese (2019, EUP) examine the relationship between media coverage and EU trust, finding that negative immigration coverage erodes trust in European institutions. This finding connects the media effects literature (Section 4.5) to the populism literature, suggesting that media framing of immigration can contribute to the broader erosion of institutional trust that populist parties exploit.
The populism literature thus reveals a complex, bidirectional relationship between immigration attitudes and partisan politics. Immigration attitudes drive support for populist parties, but populist parties also shape immigration attitudes through their rhetoric and issue emphasis. The causal arrow runs in both directions, and disentangling these effects remains an active challenge for the field.
Three further studies sharpen the populism connection. Margalit (2019, JEP) offers an influential corrective on magnitudes, arguing that the share of populist support attributable to economic insecurity is modest and that economic explanations of immigration concern in particular are easily overstated. Norris and Inglehart (2019, Cambridge University Press) provide the broadest cultural-backlash account, placing immigration attitudes at the center of an authoritarian-populist coalition mobilized against long-run value change. On the supply side, Abou-Chadi (2014, BJPS) shows that radical-right electoral success pushes mainstream parties toward more restrictive immigration positions, linking mass attitudes to party-system change through anticipatory elite behavior.
One of the most consequential developments in the post-2014 literature is a fundamental reassessment of how immigration attitudes change—or fail to change—over time. Hainmueller and Hopkins (2014) noted that the field had largely relied on cross-sectional data, leaving open the question of whether attitudes were genuinely responsive to events or instead reflected stable individual dispositions. The subsequent decade has produced a rich body of evidence addressing this gap, with implications that cut across nearly every other topic in this review.
The most direct challenge to event-driven models of attitude change comes from Kustov, Laaker, and Reller (2021, JOP), who analyze nine panel datasets spanning more than two decades to assess the within-person stability of immigration attitudes. Their central finding is striking: individual-level immigration attitudes are remarkably stable over time, with test-retest correlations comparable to those of partisanship and far exceeding those of most policy attitudes. When aggregate-level shifts do occur—such as the modest liberalization observed in U.S. public opinion during the Trump presidency—they tend to reflect thermostatic responses to policy rather than genuine individual-level conversion. The implication is that many apparent "effects" of events on immigration attitudes in cross-sectional data may instead reflect compositional changes or measurement artifacts rather than within-person attitude change.
This finding has far-reaching consequences for the literature. If attitudes are dispositional rather than situational, then the theoretical models that dominate the field—which emphasize economic conditions, local demographic change, and media exposure as drivers of attitude formation—must be reinterpreted. These factors may shape which attitudes are activated or politically salient, but they appear to have limited power to move the underlying predispositions themselves.
Against this backdrop, the literature on event-driven attitude change presents a more nuanced picture than earlier work suggested. Several studies have examined whether dramatic shocks—terrorist attacks, refugee crises, economic downturns—produce lasting attitudinal shifts.
Giani (2020, CPS) investigates the short-run effects of jihadist attacks across six countries (Israel, the Netherlands, Russia, Sweden, France, and Germany) and finds that such events produce significant increases in security fear but, crucially, no corresponding increase in ethnic prejudice—point estimates are, if anything, slightly negative. This pattern of "fear without prejudice" suggests that the cognitive and affective pathways linking threat perceptions to out-group attitudes may be less direct than commonly assumed. The emotional response to terrorism is real and measurable, but it does not automatically cascade into generalized hostility toward immigrants.
The 2015 European refugee crisis has provided the field's most important natural experiment for studying attitude dynamics at scale. Van der Brug and Harteveld (2021, EUP) examine how the crisis affected public opinion across European democracies and find evidence of polarization rather than uniform backlash. Citizens who were already skeptical of immigration became more opposed, while those who were already sympathetic became, if anything, more supportive. This pattern is consistent with the stability thesis: the crisis activated and amplified existing predispositions rather than converting one camp to the other.
Czymara and Schmidt-Catran (2017, ESR) document a more specific dynamic in Germany, where public acceptance of refugees from war-torn countries initially surged in 2015 before declining sharply following the Cologne New Year's Eve assaults. Their analysis suggests that the initial welcoming response reflected a humanitarian norm activation, while the subsequent backlash was driven by specific events that undermined the framing of refugees as deserving recipients of protection. The temporal pattern is important: it suggests that attitude change, when it does occur, may be asymmetric—easier to erode support than to build it.
Brexit offers another instructive case. Schwartz et al. (2020) document what they call a "populist paradox": after the Brexit vote, anti-immigrant attitudes in Britain actually softened among both Leave and Remain supporters. Among Leavers, a greater sense of control over immigration channeled the effects of the vote; among both groups, a desire to distance oneself from xenophobia contributed. Dennison and Kustov (2023, POQ) generalize this insight across 24 European countries over three decades, documenting what they call a "reverse backlash": the electoral success of populist radical right parties is associated with more positive immigration attitudes in the general public, not more negative ones. They argue that populist radical right success triggers negative partisanship, polarization, and a reassertion of anti-prejudice norms among the majority who do not support these parties. The finding challenges the common assumption that populist victories normalize anti-immigrant sentiment and instead suggests a thermostatic dynamic in which publics push back against perceived extremism. Together with the Brexit evidence, these results illustrate a broader pattern: populist victories that signal responsiveness to public concerns may paradoxically dampen rather than inflame the attitudes they mobilized.
The COVID-19 pandemic provided an unexpected test of whether a global health crisis would reshape immigration attitudes. Pickup et al. (2021, Social Science Quarterly) examine this question across multiple countries and find that the pandemic had surprisingly muted effects on immigration attitudes. While some evidence suggests a modest increase in restrictionist sentiment during the early phases of the pandemic, these shifts were generally small, short-lived, and inconsistent across contexts. Dennison, Kustov, and Geddes (2022, IMR) corroborate and extend this finding using Eurobarometer data across 28 European countries, weekly national surveys from the United States, and individual-level panel data from the United Kingdom and Germany. They find no systematic change in immigration preferences and no country-level correlation between the severity of the outbreak and observed attitude shifts. Crucially, however, they document a significant and consistent decrease in the perceived importance of immigration during the pandemic—a salience effect rather than a preference effect. This distinction between preference stability and salience volatility connects the COVID literature to the broader argument developed in Section 7.4: immigration attitudes are dispositionally stable in direction but situationally variable in intensity. Taken together, these findings reinforce the stability thesis: even a shock as unprecedented as a global pandemic was insufficient to produce durable shifts in the distribution of immigration attitudes, though it did temporarily displace immigration from the public agenda.
Perhaps the most consequential development in this literature is the growing recognition that attitude direction (for or against immigration) is only part of the story. Kustov (2022, BJPS) introduces the concept of an "issue importance asymmetry" in immigration politics, documenting across both American and European surveys that those who oppose immigration tend to care about the issue far more intensely than those who support it. In a nationally representative U.S. survey experiment, Kustov (2024, POP) finds that only about one-fifth of voters who prioritize immigration hold pro-immigration preferences. This asymmetry has profound implications for democratic representation. Even when a majority of citizens hold pro-immigration views, the minority who oppose immigration may exert disproportionate political influence because they vote on the issue and punish politicians who deviate from their preferred position.
Kustov (2024) further demonstrates that this asymmetry can be partially addressed: exposure to verifiable arguments about the national benefits of expanding legal immigration raised the issue's importance among pro-immigration voters without producing backlash among opponents. This finding connects the attitude dynamics literature to the persuasion literature and suggests that the political prospects for immigration reform depend not only on changing minds but on changing priorities.
Dennison and Geddes (2019, The Political Quarterly) provide complementary evidence at the European level, showing that the rise of populist radical right parties is better explained by increases in the salience of immigration than by changes in the direction of immigration attitudes. In their account, what changed in European politics was not that publics became more anti-immigrant but that immigration became a more important issue—driven by events such as the refugee crisis and media coverage—allowing parties that "own" the issue to capitalize on pre-existing but previously dormant sentiments. This salience-centered explanation connects the attitude dynamics literature to the literature on party competition and electoral behavior, reinforcing the conclusion that understanding immigration politics requires attention to the intensity and prioritization of attitudes, not merely their direction.
A final strand of this literature examines the developmental origins of immigration attitudes. Laaker (2023, BJPS) investigates whether formative experiences during adolescence and early adulthood leave lasting imprints on immigration attitudes. Drawing on cohort-level variation in exposure to immigration-related shocks, Laaker finds evidence consistent with an impressionable years hypothesis: cohorts that came of political age during periods of high immigration salience or economic disruption hold systematically different attitudes later in life. This finding complements the individual-level stability documented by Kustov, Laaker, and Reller (2021) by suggesting that while attitudes are stable within individuals once formed, the conditions under which they crystallize matter considerably for the aggregate distribution of opinion.
Taken together, the attitude dynamics literature offers a revised picture of immigration opinion. Attitudes are more stable than early cross-sectional evidence suggested, more resistant to dramatic events than threat-based theories predicted, and more consequential in their intensity than in their direction. The field has moved from asking "what changes attitudes?" to asking "what activates them?"—a shift with significant implications for both theory and political strategy.
The 2015 European refugee crisis thrust the question of attitudes toward refugees to the center of the immigration attitudes research agenda. While Hainmueller and Hopkins (2014) primarily addressed attitudes toward economic immigrants, the subsequent decade has produced a substantial body of work on how publics evaluate asylum seekers and refugees—a category of migrants that activates distinct moral, legal, and political considerations.
The landmark study in this subfield is Bansak, Hainmueller, and Hangartner (2016, Science), who fielded a conjoint experiment across 15 European countries during the height of the refugee crisis. Asking nearly 18,000 respondents to evaluate pairs of hypothetical asylum seekers with randomly varied attributes, they find a remarkably consistent preference structure across countries. Asylum seekers who are more vulnerable (e.g., victims of torture), have higher employability, are consistent with the religious and cultural majority of the host country, and have stronger asylum claims receive substantially more support. The cross-national consistency of these preferences is notable: despite the considerable variation in national contexts, the same attributes drive evaluations from Sweden to Hungary.
This preference structure suggests that citizens evaluate asylum seekers through a dual lens of deservingness and perceived integration potential. Asylum seekers who are seen as both genuinely in need and likely to contribute are favored, while those who appear to be economic migrants or who are perceived as culturally distant face greater opposition. De Coninck (2019, JEMS) provides further evidence for this categorization dynamic, showing that the labels attached to migrants—"refugee," "asylum seeker," "economic migrant"—significantly shape public evaluations even when the underlying profiles are similar.
Jeannet, Heidland, and Ruhs (2021, EUP) extend the conjoint approach to examine European preferences over asylum policy design, finding that citizens care not only about who receives protection but about how the asylum system is organized. Preferences for burden-sharing, processing speed, and conditions of reception vary systematically across countries and demographic groups, suggesting that the institutional architecture of asylum policy is itself an object of public opinion.
A growing body of work examines whether correcting misperceptions about refugees can reduce opposition. Sijilmassi et al. (2025, PNAS Nexus) investigate whether information about immigrants' deservingness—their intent to contribute and efforts to overcome hardship—can counter anti-immigration prejudice. Using factorial survey experiments, they find that deservingness cues significantly reduce opposition, particularly for low-status immigrants. The finding is consistent with a conditional solidarity framework: publics are willing to extend support to outsiders, but that willingness is contingent on perceptions that the recipients merit assistance and intend to reciprocate.
Cappelen, Sicakkan, and Van Wolleghem (2023, EUP) examine attitudes toward burden-sharing in the European asylum system, finding that citizens are generally willing to accept some responsibility for hosting refugees but that support depends heavily on perceptions of fairness. When citizens perceive that their country bears a disproportionate share of asylum claims, support erodes substantially. This finding highlights the importance of institutional design in shaping public acceptance.
Deservingness judgments are also becoming directly measurable. De Coninck et al. (2022, JIRS) validate a five-dimension Migrant Deservingness Scale—control, attitude, reciprocity, identity, and need—across nine countries, giving the field a common instrument for the evaluative dimensions that conjoint experiments repeatedly recover and a bridge between the refugee-specific and general immigration attitudes literatures.
One of the most debated questions in the refugee attitudes literature concerns the effect of direct exposure to refugees on host-community attitudes. Hangartner, Dinas, Marbach, Matakos, and Xefteris (2019, APSR) exploit a natural experiment in the Aegean Sea, where Greek islands close to the Turkish coast experienced sudden and massive refugee arrivals while similar islands slightly farther away did not. Using distance to Turkey as an instrument, they find that direct exposure to refugee arrivals produced sizable and lasting increases in hostility toward refugees, immigrants, and Muslim minorities, as well as support for restrictive asylum policies. Because refugees were only transiting through these islands—minimizing labor market competition and fiscal burden—the finding challenges standard economic explanations and suggests that mere exposure to large-scale arrivals can generate durable hostility.
Other studies paint a more varied picture. Alrababa’h et al. (2020, CPS) study Jordanian host communities and find that proximity to Syrian refugees does not uniformly increase hostility; instead, the effect depends on the nature and quality of contact. Where interactions are cooperative or structured, proximity can reduce prejudice, consistent with classic contact theory. This contrast with the Greek islands case underscores the importance of context: the conditions under which exposure occurs—whether brief and chaotic or sustained and structured—likely determine whether contact promotes tolerance or backlash.
Vertier, Viskanic, and Gamalerio (2022, PSRM) study the dismantling of the Calais "Jungle" refugee camp in France and the subsequent dispersal of its residents to reception centers across the country. They find that the arrival of refugees in new localities increased support for the far-right Front National in nearby municipalities. The effect was concentrated in areas with little prior exposure to immigration, suggesting that sudden, involuntary exposure—especially when it is politically framed as an imposition—generates stronger backlash than gradual demographic change.
Getmansky, Sinmazdemir, and Zeitzoff (2018, JPR) examine the relationship between refugee presence and xenophobic attitudes in Turkey, finding that proximity to Syrian refugees is associated with increased hostility. Von Hermanni and Neumann (2018, JEMS) similarly find that perceived threats associated with refugee arrivals predict reduced acceptance in Germany. Together, these studies suggest that the direction of exposure effects depends critically on the speed, scale, and framing of refugee arrivals—factors that policy design can in principle influence.
Lazarev and Sharma (2015, PSRM) offer a more optimistic perspective, demonstrating that intergroup contact interventions can reduce prejudice toward Syrian refugees in Turkey. Their experimental evidence suggests that even brief structured interactions can shift attitudes, although the durability of such effects remains an open question.
The Aegean natural experiment also extends from attitudes to electoral behavior. Dinas, Matakos, Xefteris, and Hangartner (2019, PA) show that quasi-random exposure to the refugee crisis on Greek islands increased vote shares for the extreme-right Golden Dawn, demonstrating that the hostility documented in attitude surveys translated into ballots where exposure was most direct.
An important counterpoint to the backlash literature comes from evidence on the temporal stability of refugee preferences. The Bansak, Hainmueller, and Hangartner (2016) conjoint design has been replicated across time periods, including during the 2022 Ukrainian refugee crisis (Bansak, Hainmueller & Hangartner, 2023, Nature). Repeat conjoint experiments conducted during the 2015--2016 and 2022 refugee crises, encompassing over 33,000 citizens in 15 European countries, reveal that preferences for asylum seekers with specific attributes remained remarkably stable over time. General support for refugees, if anything, increased slightly. Ukrainian asylum seekers were welcomed in 2022, but this welcome did not come at the expense of support for other marginalized refugee groups, including Muslim refugees. The finding is striking: despite the enormous political upheaval of the intervening years, the underlying preference structure that governs refugee evaluations appears resistant to erosion.
The refugee attitudes literature thus reveals a tension between two sets of findings. At the aggregate level, preferences for which refugees to admit are remarkably stable and structured by consistent criteria. At the local level, sudden exposure to refugee arrivals can produce real and lasting backlash. Reconciling these findings requires attention to the distinction between evaluative criteria (which refugees are preferred) and acceptance levels (how many refugees are acceptable)—a distinction that mirrors the stock-versus-flow framework that Margalit and Solodoch (2021, BJPS) have applied to immigration attitudes more broadly.
No methodological development has reshaped the study of immigration attitudes more profoundly than the rise of conjoint experiments. Since Hainmueller and Hopkins (2014, AJPS) introduced the approach to the immigration context—asking respondents to evaluate pairs of immigrant profiles with randomly varied attributes—the conjoint design has become the dominant paradigm for studying how citizens weigh the characteristics of prospective immigrants. The technique's appeal is straightforward: by randomizing attributes such as education, occupation, country of origin, language skills, and reason for migrating, researchers can estimate the causal effect of each attribute on support for admission while holding other factors constant.
The foundational conjoint finding—that citizens across skill levels prefer high-skilled immigrants, that employment prospects matter more than ethnic background, and that there exists a broad "hidden consensus" on admission criteria—has been replicated and extended in numerous contexts. Bansak, Hainmueller, and Hangartner (2016, Science) scaled the design to 15 European countries and nearly 18,000 respondents, confirming the cross-national generalizability of the core preference structure. Asylum seekers who are employable, vulnerable, and consistent with the host country's cultural majority receive the most support, and this pattern holds across countries with very different immigration histories and policy regimes.
Naumann, Stoetzer, and Pietrantuono (2018, EJP) replicate the skills premium across 15 European countries, reinforcing the conclusion that preference for high-skilled immigrants is a near-universal feature of public opinion in developed democracies. Ford and Mellon (2019, JEMS) probe this finding further in the British context, asking whether the preference for skilled immigrants reflects a genuine economic assessment or an "ethnic premium" that correlates with skill. They find that while skills matter independently, the preference for skilled immigrants partially reflects preferences for the demographic groups associated with high-skill migration, complicating a purely economic interpretation.
The conjoint literature has moved well beyond the initial skills-versus-culture debate. Ward (2018, APSR) focuses specifically on attitudes toward young immigrant men—a category that has been central to public anxieties during the European refugee crisis. Using conjoint experiments across multiple countries, Ward finds that young men face a penalty in admission evaluations relative to other demographic profiles, though this penalty varies across contexts and interacts with other attributes such as country of origin.
Newman and Malhotra (2019, JOP) exploit the conjoint framework to investigate the role of racial appearance, finding that darker skin tone reduces support for immigrant admission even after controlling for country of origin and other attributes. This finding complicates the "hidden consensus" narrative by showing that racial cues operate as an independent dimension of preference, not merely as a proxy for cultural distance or skill level.
Kustov (2020, CPS) uses conjoint experiments to test the "parochial altruism" hypothesis, examining whether citizens who express humanitarian concern for immigrants extend that concern broadly or restrict it to co-ethnics and co-nationals. The results suggest that altruistic preferences in the immigration domain are indeed parochial: they are extended more readily to immigrants who share ethnic or national ties with the evaluator, even among citizens who express universalist values.
Cogley, Doces, and Whitaker (2018, PRQ) demonstrate the portability of the conjoint approach by applying it to immigration preferences in Cote d'Ivoire, finding that attribute preferences in this non-Western context differ in important ways from those observed in Europe and North America. Their study illustrates both the generalizability of the method and the context-specificity of the substantive findings.
An important extension of the conjoint literature examines not just who gets admitted but on what terms. Neureiter (2021, IMR) investigates how integration policy expectations shape admission preferences, finding that citizens are more willing to admit immigrants when the host country has strong integration policies in place. This suggests that admission and integration preferences are linked: the perceived institutional capacity to manage the consequences of immigration shapes willingness to accept it.
Lutz (2024, EPSR) examines how post-entry rights—such as access to welfare benefits and permanent residence—affect public opposition to immigrant admission. Using factorial experiments in the United States and Switzerland, Lutz finds that restricting immigrants' welfare rights significantly decreases opposition to admission, while restricting residence rights does not have a consistent effect. The finding suggests that fiscal concerns, rather than cultural threat per se, may drive the conditional acceptance that citizens extend to immigrants.
Margalit and Solodoch (2021, BJPS) draw attention to a conceptual distinction that the broader literature has largely overlooked: the difference between attitudes toward the immigrant stock (those already present) and the immigrant flow (those seeking to enter). Using experimental evidence, they show that citizens hold systematically different preferences regarding these two groups. Concerns about the stock are driven more by cultural integration anxieties, while concerns about the flow are driven more by perceived control over borders. This distinction has important implications for interpreting survey measures of immigration attitudes, many of which conflate these two dimensions.
The rapid proliferation of conjoint experiments has been accompanied by methodological refinements. Ham, Imai, and Janson (2024, Political Analysis) develop machine learning methods for analyzing conjoint data, allowing researchers to detect heterogeneous treatment effects and complex interactions among attributes that traditional analysis may miss. Egami et al. (2022, Science Advances) introduce methods for causal inference with text-based treatments in conjoint-like designs, expanding the approach beyond discrete attribute categories to richer, more naturalistic stimulus materials.
These methodological contributions address some of the criticisms that have accumulated around the conjoint paradigm. Critics have noted that the forced-choice format may overstate consensus by compelling respondents to make trade-offs they would not voluntarily make. The hypothetical nature of the evaluations raises questions about external validity: do stated preferences for immigrant attributes predict actual behavior, such as voting or community-level acceptance? And the attribute space is necessarily constrained by the researcher's choices, potentially omitting dimensions that matter to respondents but do not appear in the experimental design.
Despite these limitations, the conjoint revolution has produced several durable insights. The preference for skilled immigrants is genuinely robust and not easily reducible to ethnic preferences. Citizens evaluate immigrants through multiple dimensions simultaneously rather than relying on a single heuristic. And the striking cross-national consistency of preference structures suggests that the demand side of immigration politics is more structured and less chaotic than media narratives imply. The challenge for the next generation of conjoint research is to connect these carefully estimated preferences to actual policy outcomes—a task that requires bridging the gap between individual-level evaluations and collective political processes.
The methodological foundations of this literature deserve explicit citation. Hainmueller, Hopkins, and Yamamoto (2014, PA) supply the causal-inference framework—the average marginal component effect estimand and its diagnostics—on which the conjoint revolution rests. Hainmueller, Hangartner, and Yamamoto (2015, PNAS) then validate stated conjoint preferences against real voting in Swiss naturalization referendums, establishing the external-validity benchmark against which subsequent designs are judged.
The decade since Hainmueller and Hopkins (2014) has seen substantial methodological innovation in how immigration attitudes are measured, modeled, and analyzed. These developments extend beyond the conjoint paradigm discussed in the previous section to encompass meta-analytic synthesis, improved survey design, behavioral outcome measures, and more sophisticated causal identification strategies.
One of the most important developments has been the effort to synthesize the accumulated evidence through systematic meta-analysis. Kaufmann and Goodwin (2018, SSR) conduct a meta-analysis of studies examining the relationship between local diversity and attitudes, finding that the widely cited negative association between community-level ethnic diversity and social cohesion is smaller and less consistent than often assumed. The heterogeneity of findings across studies depends heavily on how diversity is measured, which outcomes are examined, and the geographic scale of analysis.
Dražanová et al. (2023, JEMS) advance the meta-analytic approach to the individual level, synthesizing data from dozens of studies to assess the relative importance of different predictors of immigration attitudes. Their analysis confirms the dominance of cultural and identity-based predictors over economic self-interest but reveals important variation across national contexts and time periods. Individual-level meta-analysis of this kind represents a significant improvement over narrative literature reviews, providing more precise estimates of effect sizes and more systematic assessment of moderating variables.
Weber et al. (2025, BJPS) provide the most comprehensive meta-analysis of the immigration attitudes literature to date, synthesizing findings across studies to assess the robustness of the field's core conclusions. Their results confirm several of the patterns identified in this review—the weakness of labor market competition explanations, the strength of cultural and identity-based predictors, and the importance of education—while also identifying areas where the evidence is more equivocal than individual studies suggest.
The largest synthesis to date pushes this integration to the level of raw data. Mendoza Aviña et al. (2026, Science Advances) reanalyze roughly one hundred conjoint experiments covering 1.47 million profile evaluations from more than 140,000 respondents in 36 countries, confirming the consensus structure of immigrant preferences while documenting a gradual rise in the relative weight of economic considerations over the past decade.
Sniderman (2018, ARPS) reviews advances in survey experiment design that have particular relevance for the immigration attitudes literature, including factorial designs, list experiments for reducing social desirability bias, and vignette-based approaches that increase ecological validity. These innovations have enabled researchers to ask more precise questions about the mechanisms underlying attitude formation—moving beyond "do you favor more or less immigration?" to structured evaluations of specific trade-offs.
The concern about social desirability bias remains central. Citizens may understate their opposition to immigration in survey settings, particularly in countries where anti-immigrant sentiment carries social costs. The conjoint format partially addresses this concern by embedding evaluations in a comparative task rather than asking for direct attitude reports, but the degree to which stated conjoint preferences correspond to behavior remains an open question.
Choi, Poertner, and Sambanis (2019, PNAS) address the external validity concern directly by developing behavioral measures of attitudes toward immigrants. Rather than relying solely on survey self-reports, they use field experiments that measure actual discriminatory behavior—for instance, differential response rates to rental inquiries from applicants with immigrant-identifying names. Their findings suggest that behavioral discrimination can exceed what survey measures would predict, reinforcing concerns about social desirability bias in standard surveys.
The behavioral measures agenda complements rather than replaces survey-based approaches. Surveys remain essential for assessing population-level distributions of attitudes and for understanding the cognitive structure of preferences. But behavioral measures provide a critical check on the correspondence between stated and revealed preferences, particularly in contexts where social norms discourage the expression of anti-immigrant sentiment. The most informative research designs combine both approaches, using surveys to map the landscape of opinion and behavioral measures to test whether that landscape predicts real-world outcomes.
Together, these methodological developments have substantially strengthened the evidentiary foundation of the immigration attitudes literature. The field has moved from relying on cross-sectional correlations to deploying a diverse toolkit that includes panel data, natural experiments, conjoint designs, meta-analyses, and behavioral measures. This methodological pluralism reduces the risk that any single design artifact drives the field's conclusions and increases confidence in the robustness of the core findings reviewed in this article.
One of the starkest limitations of the literature reviewed by Hainmueller and Hopkins (2014) was its near-exclusive focus on North America and Western Europe. The authors themselves acknowledged this constraint, noting that their review covered primarily quantitative scholarship from developed democracies. In the intervening decade, the geographic scope of immigration attitudes research has expanded substantially, generating evidence from South and East Asia, the Middle East, sub-Saharan Africa, and Latin America. This expansion is not merely a matter of replication; it has produced genuinely novel findings that challenge, qualify, and extend theories developed in the Western context.
Gaikwad and Nellis (2016, AJPS) provide one of the earliest rigorous studies of immigration attitudes in a developing country, examining preferences over internal migrants in Mumbai, India. Using a conjoint design, they find that residents evaluate migrants based on their perceived economic contribution and cultural proximity—dimensions that parallel the Western literature. However, the relevant cleavages are defined by caste, religion, and regional origin rather than the national-origin and skill categories that dominate Western studies, illustrating how the structure of preferences may be universal even when the content of group boundaries varies.
Shao et al. (2023, JEMS) examine immigration attitudes in China, a context where immigration is minimal but where attitudes toward internal rural-to-urban migrants may parallel anti-immigrant sentiment in other countries. They find that urban residents in China hold structured preferences over migrant characteristics that echo the skill-based and cultural-distance patterns observed in Western democracies, suggesting that the psychological processes underlying immigration attitudes may transcend specific migration regimes.
Lee and Chou (2020, Political Psychology) study attitudes toward immigrants in Hong Kong, where the highly salient distinction between mainland Chinese immigrants and other groups provides a unique context for testing cultural threat hypotheses. Their findings suggest that perceived cultural threat operates even within nominally same-ethnicity migration flows, challenging the assumption that cultural threat is primarily about visible ethnoracial difference.
Chang (2019, Asian Journal of Political Science) examines attitudes toward immigrants in South Korea, while Denney and Green (2023, JEMS) focus specifically on attitudes toward co-ethnic migrants (ethnic Korean returnees from China) in the same country. Their combined work reveals that co-ethnicity provides only a limited buffer against restrictionist attitudes: South Korean citizens prefer co-ethnic returnees to non-co-ethnic immigrants, but this preference is modest compared to other attributes such as skill level and cultural adaptation. The finding challenges simplistic versions of ethnic solidarity theory and suggests that host-society attitudes are driven more by perceived integration capacity than by shared descent.
Kawakami et al. (2025, JEMS) offer a direct cross-national comparison between England and Japan, two countries with very different immigration histories and cultural contexts. Their study reveals both commonalities (the universal preference for skilled immigrants) and instructive differences (variation in how cultural distance is defined and weighted), illustrating the value of comparative designs that hold the methodology constant while varying the context.
Shockley and Gengler (2023, PSRM) study immigration attitudes in the Gulf states, where migrant workers constitute a majority of the population in several countries but enjoy few political or social rights. They find that citizens in these contexts hold attitudes shaped by concerns about cultural preservation and economic competition, but the specific dynamics differ from Western democracies in important ways. The kafala (sponsorship) system and the near-total exclusion of migrants from citizenship create a political economy of attitudes that has no direct parallel in the democratic contexts that dominate the literature.
Alrababa’h et al. (2020, CPS) study attitudes toward Syrian refugees in Jordan, finding that the nature and quality of intergroup contact moderates the direction of exposure effects. Their work demonstrates that the findings from the European refugee crisis literature cannot be straightforwardly exported to Middle Eastern host countries, where the political, economic, and social context of refugee reception differs fundamentally.
Ariely (2019, JEMS) examines immigration attitudes in Israel, a context where national identity, security concerns, and the ethno-religious composition of immigrant flows interact in ways that differ markedly from both Western European and Gulf state settings.
Whitaker and Giersch (2015, JEMS) provide one of the first systematic studies of immigration attitudes in Africa, documenting patterns that both parallel and diverge from the Western literature. In the African context, attitudes toward immigrants are shaped by concerns about economic competition and cultural difference, but the relevant categories—defined by tribal, linguistic, and national boundaries—differ from those in Western studies. Harris et al. (2017, PRQ) study attitudes toward immigrants in South Africa, where xenophobic violence has been a recurring feature of the political landscape, finding that economic threat perceptions and national identity are powerful predictors of hostility.
Acevedo and Meseguer (2022, Migration Studies) examine immigration attitudes in Mexico, a country that has traditionally been studied as a sending rather than receiving society. As Mexico increasingly hosts Central American and Caribbean migrants, public attitudes have become an object of scholarly attention. They find that Mexican citizens' attitudes are shaped by many of the same factors identified in the Western literature—economic concerns, cultural distance, and media exposure—but that the experience of being a sending country also shapes how citizens think about immigration, producing a more complex attitudinal landscape.
The geographic expansion of the literature has produced several theoretical insights that would not have emerged from Western-only research. First, the preference for skilled immigrants appears genuinely universal, observed across every context in which it has been tested. This near-universal pattern is consistent with a sociotropic economic interpretation—citizens everywhere prefer immigrants who are perceived as likely to contribute rather than to compete—but may also reflect status-based prejudice that transcends cultural boundaries (Helms, 2023, AJPS).
Second, the content of "cultural threat" varies dramatically across contexts even as the form remains constant. In Western Europe, cultural threat is defined primarily in terms of religion (especially Islam) and ethnoracial appearance. In East Asia, it may center on regional or linguistic identity. In the Gulf states, it reflects concerns about demographic dilution of the citizen population. Fouka and Tabellini (2021, APSR) provide a theoretical framework for understanding how in-group boundaries shift in response to immigration, showing that the categories defining who counts as an "insider" are themselves endogenous to immigration flows. Carter and Pérez (2015, Political Psychology) examine how racial hierarchy structures shape attitudes toward different immigrant groups, finding that existing domestic status hierarchies condition which immigrants are seen as threatening.
Third, the institutional and political context of immigration varies so dramatically across regions that theories developed in democratic welfare states may not travel to authoritarian regimes, rentier economies, or countries with porous borders and limited state capacity. The Gulf states literature, in particular, suggests that attitudes are shaped not only by individual-level predictors but by the macro-institutional framework that defines immigrants' legal and social position. This insight has implications even for the Western literature, where variation in welfare state design, labor market regulation, and naturalization regimes may moderate the individual-level relationships that researchers have treated as universal.
The geographic expansion of the field is far from complete. Large parts of South Asia, Central Asia, and Southeast Asia remain understudied, as do many African countries beyond South Africa. But the evidence accumulated to date is sufficient to conclude that the core psychological processes underlying immigration attitudes—the preference for perceived contributors, the sensitivity to cultural distance, the role of threat perceptions—have deep cross-cultural roots, even as their specific expression varies with context.
This review has surveyed the immigration attitudes literature published in the decade following Hainmueller and Hopkins's (2014) influential synthesis. The volume of research has been remarkable: the original review has accumulated over 3,000 citations, and the literature it catalyzed has produced hundreds of additional studies spanning new methods, new geographies, and new substantive questions. In this concluding section, I revisit the seven conclusions offered by Hainmueller and Hopkins, assessing how each has fared in light of subsequent evidence, before identifying five new conclusions that emerge from the post-2014 literature.
Conclusion 1 (original): The labor market competition hypothesis has repeatedly failed to find empirical support, making it something of a "zombie theory."
Updated assessment: This conclusion has been strongly reinforced. The post-2014 literature has produced additional evidence against the labor market competition hypothesis from conjoint experiments (Bansak, Hainmueller & Hangartner 2016; Naumann, Stoetzer & Pietrantuono 2018), panel data (Kustov, Laaker & Reller 2021), and cross-national surveys. Citizens at all skill levels continue to prefer high-skilled immigrants, and individual-level economic vulnerability remains a weak predictor of immigration attitudes. However, Pardos-Prado and Xena (2018) offer a refinement, showing that skill specificity—the transferability of a worker's skills across occupations—is a better predictor than skill level per se. Workers whose skills are occupation-specific and thus harder to transfer are modestly more opposed to immigration, suggesting that the relevant economic mechanism may be asset specificity rather than simple factor competition. The labor market competition hypothesis remains a zombie, but a somewhat better-specified version of economic threat—focused on asset vulnerability rather than skill level—retains some explanatory power. Moreover, Colantone and Stanig (2018) demonstrate that trade globalization shocks can fuel anti-immigration sentiment and nationalist politics, though the operative mechanism runs through perceived cultural and identity threats rather than direct labor market competition. Similarly, Wu (2022) and Gamez-Djokic and Waytz (2020) find that economic anxieties from trade and automation are often misattributed to immigration, reinforcing the pattern in which economic disruptions activate sociopsychological rather than self-interest channels.
Conclusion 2 (original): Education is consistently associated with more pro-immigration views, but the mechanism is unclear.
Updated assessment: The education-attitudes relationship remains one of the most robust findings in the field, and the post-2014 literature has made progress on the mechanism question. Cavaillé and Marshall (2018) exploit six compulsory schooling reforms in five Western European countries (Denmark, France, Great Britain, the Netherlands, and Sweden) as natural experiments, finding that exogenous increases in education produce more tolerant attitudes toward immigrants. This quasi-experimental evidence strengthens the case that education has a causal effect on attitudes rather than merely proxying for pre-existing values or cognitive ability. The finding is consistent with socialization and cognitive sophistication explanations: more schooling appears to foster values of tolerance and cosmopolitanism that translate into more favorable immigration attitudes.
Conclusion 3 (original): Sociotropic theories are less clearly differentiated from one another, and future scholarship should identify critical tests to distinguish them.
Updated assessment: Progress on differentiating sociotropic mechanisms has been incremental rather than transformative. Conjoint experiments have helped by isolating the effects of specific attributes (cultural background, economic contribution, legal status), but the fundamental challenge remains: sociotropic economic and cultural concerns are empirically correlated and theoretically intertwined. Ford and Mellon (2019) demonstrate the entanglement by showing that the apparent "skills premium" partially reflects ethnic preferences. The geographic expansion of the field has contributed by showing that the content of cultural threat varies across contexts while the form remains constant, suggesting that the underlying mechanism may be a general aversion to perceived out-groups rather than a specifically cultural or economic concern. Fully disentangling these pathways remains an open challenge.
Conclusion 4 (original): Attitudes toward social aggregates—the nation or ethnic groups—commonly find empirical support, but endogeneity concerns leave the direction of causation unclear.
Updated assessment: The post-2014 literature has made significant progress on the causal identification front. Natural experiments on refugee exposure (Hangartner et al. 2019), compulsory schooling (Cavaillé & Marshall 2018), and policy shocks (Schwartz et al. 2020) provide much stronger causal evidence than the cross-sectional designs that dominated the earlier literature. Panel data analyses (Kustov, Laaker & Reller 2021) have addressed within-person dynamics that cross-sectional data cannot capture. The field has heeded this call and moved substantially toward designs that support causal inference.
Conclusion 5 (original): Prejudice and ethnocentrism are consistently connected with restrictive attitudes, but existing research disagrees about how group-specific such attitudes are.
Updated assessment: The conjoint literature has provided new purchase on the group-specificity question. Newman and Malhotra (2018) show that racial appearance operates as an independent dimension of preference, not merely as a proxy for other attributes. Kustov (2020) finds that altruistic preferences in the immigration domain are parochial, extended more readily to co-ethnics. At the same time, the geographic expansion of the field reveals that the relevant group boundaries vary across contexts—race in the Americas, religion in Western Europe, caste and region in India, language in parts of Africa—suggesting that ethnocentrism may be a general disposition that takes locally specific forms.
Conclusion 6 (original): Information environments and elite rhetoric play central theoretical roles, but their measurement has been very coarse.
Updated assessment: This remains a significant gap. While the persuasion literature has advanced—most notably with Kalla and Broockman's (2020) work on deep canvassing, which shows that sustained interpersonal conversations can durably reduce exclusionary attitudes—the measurement of information environments remains challenging. The post-2014 period has seen the dramatic rise of social media as a vehicle for immigration discourse, but rigorous research on how social media exposure shapes immigration attitudes is still in its early stages. Kustov (2024) extends this by showing that verifiable arguments about the benefits of immigration can raise the issue's importance among sympathetic voters, suggesting that the information environment affects not just attitude direction but attitude intensity.
Conclusion 7 (original): Research on immigration attitudes has been surprisingly divorced from research on political partisanship and ideology.
Updated assessment: This conclusion has been decisively addressed. The post-2014 period has seen immigration become one of the defining issues of partisan politics in virtually every Western democracy. Reny, Collingwood, and Valenzuela (2019) demonstrate that attitudes toward immigration were a central predictor of vote choice in the 2016 U.S. presidential election, with immigration attitudes predicting support for Donald Trump more strongly than economic concerns. Immigration has become deeply embedded in partisan conflict, and the question is no longer whether immigration attitudes and partisanship are connected but rather how they became so tightly linked and what consequences this alignment has for democratic representation and policy.
The post-2014 literature has generated findings that go beyond updating Hainmueller and Hopkins's original framework. Five new conclusions emerge from the accumulated evidence.
New Conclusion 1: Immigration attitudes are dispositional, not situational. The most important single finding in the post-2014 literature may be the remarkable within-person stability of immigration attitudes documented by Kustov, Laaker, and Reller (2021). When combined with evidence that dramatic events—terrorist attacks (Giani 2020), refugee crises (van der Brug & Harteveld 2021), pandemics (Pickup et al. 2021)—produce modest and often transient effects on attitudes, the weight of evidence favors a dispositional model. Immigration attitudes crystallize relatively early in the life course (Laaker 2023), remain stable thereafter, and are resistant to updating by new information or events. This does not mean attitudes are immutable—deep canvassing (Kalla & Broockman 2020) and compelling arguments (Kustov 2024) can produce measurable shifts—but it does mean that event-driven theories of attitude change require substantial qualification.
New Conclusion 2: Attitude intensity matters as much as attitude direction. The discovery of an issue importance asymmetry in immigration politics (Kustov 2022; 2024) fundamentally reframes the relationship between public opinion and policy. Even when majorities hold pro-immigration preferences, policy does not follow because the anti-immigration minority cares more about the issue, votes on it, and punishes politicians for deviation. This asymmetry may explain the persistent representation gap that Hainmueller and Hopkins (2014) noted as an understudied topic. The implication is that persuasion alone—changing the direction of attitudes—is insufficient for producing policy change. Mobilization—increasing the intensity of pro-immigration sentiment—may be equally or more important.
New Conclusion 3: The conjoint method reveals structured and cross-nationally consistent preferences. The proliferation of conjoint experiments has established that citizens everywhere prefer immigrants who are perceived as economically productive, culturally proximate, and genuinely in need of protection (in the refugee context). These preferences are strikingly consistent across countries, time periods, and respondent characteristics. The hidden consensus identified by Hainmueller and Hopkins (2014) has been confirmed at a much larger scale. At the same time, this consensus operates at the level of evaluative criteria rather than policy outcomes—citizens may agree on which immigrants to prefer while disagreeing sharply on how many to admit, creating an apparent paradox that the field has yet to resolve fully.
New Conclusion 4: Geographic expansion reveals universal processes with locally specific content. Studies from South and East Asia (Gaikwad & Nellis 2016; Lee & Chou 2020; Shao et al. 2023), the Middle East (Shockley & Gengler 2023; Alrababa’h et al. 2020), Africa (Harris et al. 2017; Whitaker & Giersch 2015), and Latin America (Acevedo & Meseguer 2022) confirm that the psychological processes underlying immigration attitudes—economic threat perception, cultural distance sensitivity, status-based evaluation—operate across the globe. But the specific groups that are targeted, the cleavages that define in-group and out-group, and the institutional contexts that shape political expression vary dramatically. Theories developed in the Western context are partially portable, but full generality requires attention to local social structures and political institutions.
New Conclusion 5: The attitudes-policy gap is the field's most important remaining puzzle. Despite the enormous growth in our understanding of immigration attitudes, the field has made far less progress on connecting attitudes to policy outcomes. We know a great deal about what citizens think about immigration and why, but we know surprisingly little about when and how those attitudes translate into policy. The representation gap—the persistent tendency of immigration policy to diverge from median-voter preferences—remains poorly understood. Addressing this gap requires bridging the divide between public opinion research and comparative policy analysis, incorporating the role of organized interests, institutional veto points, judicial constraints, and international obligations that mediate between public preferences and policy outputs.
Several pressing questions remain for the next generation of research. First, the causal mechanisms through which education fosters pro-immigration attitudes deserve continued investigation. Cavaillé and Marshall (2018) have established that the relationship is at least partially causal, but the specific pathways—cognitive sophistication, value socialization, intergroup contact, economic security—remain underspecified.
Second, the dynamics of immigration attitudes in the context of rapid demographic change merit sustained attention. As immigrant-origin populations grow and naturalize, the distinction between "native" and "immigrant" attitudes becomes increasingly blurred. How immigration attitudes evolve as ethnoracial composition changes—and how existing citizens respond to the prospect of becoming demographic minorities—is a question of first-order political importance.
Third, the field needs to develop better theories and evidence about the supply side of immigration politics: how elites, parties, and media shape the information environment that activates, amplifies, and channels pre-existing attitudes. The demand side of immigration opinion is now relatively well understood; the supply side is not.
Fourth, the behavioral consequences of immigration attitudes—for voting, community engagement, intergroup interaction, and policy compliance—deserve more systematic investigation. The conjoint literature has produced precise estimates of stated preferences, but the link between stated preferences and consequential behavior remains an area where the evidence is thin.
Fifth, the emergence of climate-induced migration as a category distinct from both economic migration and conflict-driven displacement raises new questions about public attitudes. Early evidence is mixed: some studies suggest publics view climate migrants with intermediate sympathy, but the first major U.S. conjoint on the question (Adman, Lajevardi & Seligsohn, 2024, PS) finds that climate migrants are preferred no more than economic migrants and clearly less than political refugees. This literature is in its earliest stages. More broadly, the proliferation of migrant categories (economic, refugee, climate, family reunification, student) calls for integrative deservingness frameworks that identify the common evaluative dimensions underlying preferences across these diverse flows, a need the validated deservingness scale discussed in Section 8.2 begins to meet.
Finally, the attitudes-policy nexus should become a central research agenda. The conditions under which public opinion influences immigration policy—and the conditions under which it does not—are among the most consequential questions in democratic theory. The field has laid the groundwork by mapping the landscape of attitudes; the next step is to trace the pathways through which that landscape shapes political outcomes.
The study of immigration attitudes has matured remarkably in the decade since Hainmueller and Hopkins's review. The field has stronger methods, broader geographic coverage, and a more sophisticated understanding of the psychological, economic, and political processes that shape how citizens think about immigration. What it does not yet have—and what it most urgently needs—is a systematic account of when and why those attitudes matter for the policies that governments enact. Until that gap is closed, the field's impressive empirical accomplishments will remain somewhat disconnected from the political outcomes they are meant to illuminate.