Is There a Backlash Against Immigration from Richer Countries? International Hierarchy and the Limits of Group Threat

Authors: Alexander Kustov

Journal: Political Psychology (2019)

Resources

Figures & Tables

Abstract

Why do immigrants from particular countries systematically face more opposition? To resolve inconsistencies of prevailing group threat theories, I re-introduce a long-standing hypothesis stipulating that people have a disposition for maintaining status hierarchy between ethnic groups. Accordingly, independent of perceived economic or cultural threat, natives are more likely to prefer immigrant groups of higher status based on the development level of the group’s national origin. To test this argument, I exploit a substantial provincial variation of immigration flows and attitudes in Spain—one of the only countries that has received immigrants from both less and more developed countries. Consistent with my hypothesis, I demonstrate that anti-immigration attitudes are more widespread in areas with immigrants from less developed countries regardless of their economic and cultural characteristics. I further document that many voters perceive stable group hierarchies and that these preferences are more predictive of anti-immigration attitudes in lower-status immigration contexts. Overall, these results suggest that even culturally similar and economically beneficial immigrant groups from poorer countries can face public opposition due to their lower-status national origin, highlighting the independent role of group status perceptions in politics.

Running head: International Hierarchy and Immigration Threat \ Keywords: Immigration; Public Opinion; National Origin; Ethnic Inequality; Spain

Introduction

With the rise of population movements and a corresponding increase in its salience in Western democracies, public opinion on immigration has gained significant scholarly attention. While existing research predominantly focuses on immigration as a whole, it is increasingly acknowledged that people’s attitudes are largely dependent on contextual factors and the particular immigrant group they have in mind. One of the most robust findings in the literature, for instance, is that the majority of voters prefer immigrants from particular origin countries [@Hainmueller2014].

Why do immigrants from particular countries face more opposition? Relatedly, why do natives oppose immigration in some contexts more than others? As indicated by the recent analysis of the literature, rather than being driven by self-interest, “immigration attitudes are shaped by sociotropic concerns about its cultural impacts—and to a lesser extent its economic impacts—on the nation as a whole” (ibid, 225). Accordingly, many existing explanations assume that immigrants pose a “group threat” related to (intergroup) economic competition for material resources and cultural difference in values. Thus, it is implied that the more competing or different immigrant groups are with regard to the receiving society, the greater is the group threat and thus the backlash against these groups. Similarly, people are expected to be more anti-immigration in contexts with more “threatening” immigrants.

While people may belong to various social groups, immigrants are by definition evaluated with respect to their (non-native) ethnic origin related to country of birth. (Note: As other ethnic group category [@Chandra2006], national origin is an immutable, ascribed group membership based on one’s descent [@Goode2015]. Here and hereafter, I use the term “group” as a shorthand for “ethnic group category based on national origin.”) As indicated by recent evidence from conjoint choice experiments, immigrants from some countries are more disliked than others even conditional on their individual characteristics [@Hainmueller2015a]. Nonetheless, since most immigrants in rich Western democracies are coming from countries that are both poorer and ethnically different, it is often not possible to demarcate which exact characteristic invokes public opposition toward a certain group.

This study unpacks the “black box” of immigrants’ national origin effects by proposing a “group hierarchy” hypothesis. Building on a vast literature on ethnic and class prejudice, the hypothesis stipulates that many people have a disposition for maintaining the stability of status hierarchy between ethnic groups. Accordingly, voters are expected to resist immigration from “lower-status,” but not “higher-status,” national origins independent of perceived “group threat” (or immigrants’ economic and cultural characteristics). Given significant and salient inequalities between receiving and sending countries, public perception of immigrants’ group status is further often based on the development level of their origin country.

To test this argument, I exploit a substantial provincial variation of immigration flows and attitudes in Spain, which is one of the only major countries that received significant amount of immigrants from both richer and poorer countries of diverse cultural and economic backgrounds. In particular, I show that neither linguistic, religious, nor skill-level differences between immigrants and natives can adequately predict public opposition against immigration in a particular local context. Instead, contrary to conventional accounts, I demonstrate that anti-immigration attitudes are more prevalent in provinces with larger shares of immigrants from poorer countries even after accounting for cultural and economic characteristics of the foreign-born population.

To explore the micro-mechanisms of this relationship, I provide evidence that many Spanish voters do perceive stable group hierarchies and that these hierarchical preferences are more predictive of anti-immigration attitudes in contexts with more lower-status immigration. To address robustness and endogeneity concerns, I test a variety of alternative model specifications and employ an instrumental variable approach based on the group-specific settlement patterns of past immigrants and the differences in transportation accessibility across provinces, which overall corroborates the results. Finally, I complement my analysis with the evidence from the European Social Survey, indicating that the relationship between group hierarchy preferences and anti-immigration attitudes is generalizable across countries.

Overall, the evidence supports the idea that people are opposed to immigrant groups whom they consider “inferior” (not just “different” or “competing”). Accordingly, an influx of immigrants from richer countries—who are perceived to be of higher group status—rarely causes backlash among natives regardless of whether it goes against their cultural values or economic group interests. In this sense, my results draw attention to the limits of prevailing group threat explanations and highlights the previously overlooked role of group status perceptions in politics. In doing so, the idea of group hierarchy fruitfully speaks to the currently contested issues in the immigration literature and provides a foundation for future research that moves beyond the standard “economy-or-culture” framework.

Group Threat of Immigration

It has been increasingly acknowledged that public attitudes toward immigration are largely driven by voters’ “sociotropic” concerns about its impacts on their society rather than just their self-interest [@Hainmueller2014]. Similarly, building on Blumer’s group position model Blumer1958 and two major theories in social psychology—realistic group conflict [@Sherif1988] and social identity [@Tajfel1979a]—many scholars have conceptualized the popular perceptions of immigration and its impact as a “group threat.” Although there can be multiple ways to categorize potential sociotropic and threatening stimuli, the growing consensus points to either immigrants’ economic (realistic) threat related to group competition and cultural (symbolic) threat related to group difference [@Ceobanu2010, Hainmueller2014].

The underlying assumption of most sociotropic and group threat accounts of immigration attitudes—regardless of the particular mechanism at work—is that people favor or oppose a certain policy if they believe it poses an opportunity or threat to their national ingroup. Perceived “national interests,” however, are often contested and thus the notion of “group threat” is limited by the vast individual differences in threat perceptions. As aptly noted by @Malhotra2013a, “individuals who are antithetical to immigrants… are likely to describe immigration as harmful on any dimension on which they are asked to assess its merits.” As @Dancygier2014 further contend, “explaining preferences for immigrant restrictions by pointing to survey responses about whether immigrants pose a threat…appears nearly tautological.” These criticisms notwithstanding, however, there is still a substantial literature suggesting that—with regard to immigration—people’s subjective perceptions of their “group threat” could be grounded in the contextually heterogeneous objective conditions of scarce resources and cultural distance [@Ceobanu2010, Hopkins2010, Enos2016].

Perception of Intergroup Economic Competition

The size of the minority outgroup and its change are perhaps the major contextual factors that are hypothesized to shape (or even be equated with) perceived “group threat.” In doing so, many studies argue for the realistic account of group interests related to the group competition for limited resources that effectively determines public attitudes on immigration [@Olzak1994, Quillian1995]. However, given rather mixed evidence for the “group size” argument [@Pottie-Sherman2017], scholars have been compelled to develop a number of additional empirical tests that qualify the popular perception of economic group threat [@Meuleman2010, Hainmueller2014]. Most prominently, natives are expected to oppose immigrants who are perceived to be a net loss for the economy due to their characteristics (e.g., low-skilled). Consequently, the natives who live in areas with more low-skilled immigrants are expected to perceive more group threat and be more opposed to immigration in general [@Hjerm2011].

[Intergroup Economic Threat] Anti-immigration attitudes are higher in contexts with more “economically undesirable” immigrants

Perception of Intergroup Cultural Difference

While scholars do find that immigration attitudes are related to the perceived immigrants’ impact on the economy [@Dancygier2013a], there is a growing consensus that concerns about their cultural impact are much more consequential [@Card2012, Hainmueller2014]. Unlike the economic threat hypothesis that presupposes some realistic group competition [@Sherif1988][1961], the cultural threat hypothesis is often based on an insight from social identity theory that objective rivalry for resources is neither necessary nor sufficient for ingroup bias [@Tajfel1979a]. To explain the contextual and group-specific variation in anti-immigration attitudes, however, the notion of cultural threat also requires an assumption about “the quality and the quantity” of group difference. Accordingly, most studies that find an effect of immigrants’ national origin on people’s perceptions of immigration assume that it is at least partially grounded in certain real cultural differences. This, in turn, leads to the (often implicit) reasoning that the extent of cultural differences explains why certain immigrant groups can be conceived as a bigger threat to the national culture and thus evoke more popular opposition than others.

Despite its notorious ambiguity [@Sewell1999, Jahoda2012], however, “culture” is rarely defined and often viewed as a self-evident construct. This lack of conceptual clarity explains why a lot of empirical evidence can be in principle attributed to the effect of culture. As summarized by @Hainmueller2014, “[t]oo frequently, culture operates as a residual category, describing any noneconomic immigrant attribute. On its own, the claim that culture matters thus has less content than meets the eye.” Nevertheless, immigration scholars rarely admit the difficulty of determining objective differences [@VanOsch2012]. As a result, studies predominantly depart from the assumption that the immigrant-receiving country in question does have a certain “culture” that can potentially be “threatened” by the incoming flows of (more or less culturally different) immigrants.

[Intergroup Cultural Threat] Anti-immigration attitudes are higher in contexts with more “culturally different” immigrants

Perception of Intergroup Status Hierarchy

The widespread understanding of culture as a societally shared meaning system has been challenged by recent studies showing that the variation of cultural values is much greater within countries than between them [@Fischer2011]. Accordingly, empirical research demonstrates little difference between the value orientations of natives and immigrants, as well as its unrelatedness to the cultural tradition of immigrants’ origin country [@Wimmer2014a]. Furthermore, while the idea of omnipresent ingroup bias across groups and contexts is still prevalent in the immigration literature, it has been increasingly contested elsewhere [@Jost1997, Sidanius2004b, Bergh2016].

Although the public preference for skilled immigrants with similar ethnic backgrounds is well-documented, it is often not empirically possible to distinguish these assumed cultural and economic considerations from an expression of “class prejudice” [@Lott2012] against the lower-status groups who are deemed as inferior. At the same time, while less likely on average, most voters are still opposed to even culturally similar and economically beneficial immigration [@Card2005, Naumann2018a]. To account for these unexplained empirical regularities, I propose a new “group hierarchy” hypothesis in which natives prefer immigrants of higher group status over those of lower group status regardless of immigrants’ individual characteristics. Departing from the classical account of social stratification and re-interpreting it in a way consistent with the recent literature on global inequality and political psychology, this hypothesis aims to explain contextuallyand group-differentiated opposition to immigration by serving as a complement to the prevailing “sociotropic” and “group threat” accounts.

The basic idea of group hierarchy is based on the premise that, while different ethnic and national groups need not be unequal in theory, they are often ranked across a variety of consequential characteristics in practice [@Hagendoorn1995]. As opposed to rather ambiguous cultural differences, interethnic and international disparities in resources and power are ubiquitous and strikingly durable [@Tilly1999, Cederman2013, Alesina2016, Milanovic2015, McConaughey2018]. These objective between-group inequalities, however, also give rise to widely shared subjective beliefs about their “social status” position (related to perceived group differences in esteem and respect). In turn, such status beliefs about which groups are “better” constitute an important factor of group attitudes that is independent of actual material inequalities and differences in values [@Ridgeway2013]. Given the ubiquity of various disparities in their lives and status quo bias, people are generally expected to intuitively understand and endorse existing social hierarchies regardless of their own position [@VanBerkel2015a].

The literature on the popular perceptions of group hierarchy (as opposed to mere difference or competition) that goes beyond ingroup favoritism can be traced back to the foundations of social science. In his essay on the psychology of race relations, @Thomas1904 was perhaps among the first scholars to differentiate between ethnocentric “race-prejudice” and hierarchical “caste-feeling.” While the former is akin to general antipathy that arises from ingroup preference (i.e., is about “the very fact of difference”), the latter implies the cognitive distinction based on the group’s “mental and economic superiority” [@Thomas1904]. In many ways inspired by Thomas, @Blumer1958 famously described “prejudice as a sense of group position” based on difference and superiority, as well as privilege and fear. (Note: It is worth noting that most of the follow-up theories [@Quillian1995, Bobo1999] predominantly focused on the second part of his argument emphasizing the perceptions of group threat (“they can harm us”) and deemphasizing the perceptions of group superiority (“they should know their place”).) More recently, @Hagendoorn1993 tried to devise a theory of ethnic hierarchy that would account for both the omnipresence of ingroup bias and the remarkable consensus regarding the relative position of different ethnic and national groups. As noted in his later work [@Hagendoorn1995], despite a nearly universal tendency for ingroup preference, “outgroups are ranked as more or less attractive? and there is social consensus about the assigned rank” across different contexts.

How do native voters perceive the ethnic group status hierarchy when it comes to international migration? The literature suggests that, in addition to self-selected characteristics of particular migration flows (e.g., average skill of Indian immigrants to the US), immigrant groups are generally evaluated based on stereotypes about their country [@Lee2006a, Sevillano2012]. The role of cultural difference and economic desirability in this process, however, may be limited since the valence of perceived national images seem to be remarkably stable across different countries. Specifically, national stereotypes are largely determined by countries’ GDP per capita [@McCrae2013], which is especially conducive to producing social hierarchies of relative development levels [@Reese2012]. As recently documented by @Milanovic2015, for instance, more than two-thirds of global variation in individual household income (PPP) is startlingly determined by one’s national origin. Given these enormous and salient international inequalities, immigrants from developing to developed countries may thus be perceived as having lower group status regardless of their individual characteristics [@Ridgeway2013].

Compared to other social processes, international migration makes it especially apparent that some ethnic boundaries (such as those related to national membership) need not be fixed and can potentially be crossed by individuals of any initial group status. Since people overwhelmingly prefer associating with others of equal or higher status [@Weeks2004a], immigration from similarly or more developed countries may not cause a backlash among the natives by itself. At the same time, immigration from less developed countries is more likely to be considered by natives as a “transgression” of group status boundaries, regardless of their perceived group interests. Therefore, I contend that the public perceptions of intergroup hierarchy (rather than mere difference or competition) between sending and receiving countries is a significant predictor of immigration attitudes. Contrary to group threat, the group hierarchy hypothesis implies that natives are more likely to oppose immigration from lower-status groups even when it is culturally similar and economically beneficial.

One can reasonably argue that the perceptions of group hierarchy violations may be considered a type of (symbolic) group threat [@Tolnay1995] and can be explained by a type of social identity theory [@Doosje2002]. At the same time, my account is even more in line with a growing literature on social dominance theory, which asserts that many people—to a various extent—have a disposition for maintaining the stability of group-based social hierarchies regardless of their own position and thus their self- or group interest [@Sidanius2004b]. Despite this general compatibility, my argument significantly diverges from a direct application of social identity and social dominance theories to the study of immigration attitudes [@Esses2005, Thomsen2008]. Most important, I build on national stereotype and global inequality literatures to argue that natives use the development level of sending countries as a heuristic for status hierarchy of immigrant groups. Furthermore, I build on group threat and sociotropic politics literatures to explore potential contextual—rather than mere psychological—factors of public opinion. As a result, unlikely existing accounts, my argument highlights that anti-immigration attitudes are more likely to be widespread in places with more foreigners from poorer countries independent of their other characteristics.

[Intergroup Status Hierarchy] Anti-immigration attitudes are higher in contexts with more immigrants of “lower group status”

Empirical Evidence

Previous research may have not been able to discern the perceptions of group hierarchy from group threat due to the fact that cultural, economic, and status differences largely overlap among most foreign groups in immigrant-receiving countries. Moreover, the popular perception of these differences can be further confounded by the increasingly prominent discourse of radical right populist parties [@Golder2016]. As one of the only advanced democracies that has had significant immigration from both less and more developed countries of diverse cultures and skills (without a major populist backlash), Spain is an ideal case to test these competing hypotheses.

Immigration in Spain

Traditionally conceived as an emigrant country, Spain has recently experienced a dramatic increase in its foreign-born population (from 3% in 2001 to 12% in 2011). (Note: The data on migration flows is derived from the official census conducted by Instituto Nacional de Estadistica. According to the census, foreigners or immigrants are defined as individuals born abroad. Although there was a downward trend in migration after a few years of financial crisis, these data are not considered here. For a review of the change in Spanish immigration flows and policy attitudes, see @CeaDAncona2016.) Largely driven by economic upturn and a corresponding increase in demand for labor, such an extraordinary migration rate has made Spain the second largest absolute immigrant recipient among the OECD countries (next to the US). Another important characteristic of immigration in Spain is its relative diversity, with the largest groups coming from Romania, Morocco, United Kingdom, Germany, Italy, France, Colombia and Ecuador. Most notably, Spain is one of the only countries in the world that had a significant inflow of people from more developed countries (measured as relative GDP per capita). Finally, there seems to be substantial regional variation in migration composition and policy attitudes. In turn, these peculiarities of Spanish immigration processes allow for disentangling the contextual cultural and economic drivers of anti-immigration attitudes from those related to group status.

At the same time, given the absence of nativist discourse in national politics [@Arango2013], Spain has repeatedly been considered to be among the least anti-immigration countries in Western Europe [@Escandell2010]. Relatedly, in contrast to other European countries with similar levels of immigration, Spanish right-wing populist parties have not yet been electorally successful on a national level. Despite all these unique features, on the whole, the majority of Spanish voters still oppose increasing immigration similar to other immigrant-receiving countries [@IOM2015].

Accordingly, polls indicate that anti-immigration attitudes have increased from 1996 to 2010 with the growing foreign-born population [@CeaDAncona2016]. Furthermore, on a subnational level, some autonomous communities seem to have more pronounced exclusionary attitudes than others, which some scholars connect to the prevalence of regional nationalism [@Escandell2009]. (Note: Administratively, Spain is divided into 17 autonomous communities or 50 provinces. Due to their pronounced regional national identity, the three original autonomous communities of Catalonia, Galicia and the Basque Country are sometimes excluded from the analysis of immigration attitudes. The two additional autonomous cities of Melilla and Ceuta (located on the north coast of Africa) are also usually excluded due to the prevalence of irregular immigration and other idiosyncratic factors.) While existing studies have demonstrated that group size is not related to popular attitudes on the level of autonomous communities, there is some corroboration that respondents in provinces with more immigration (especially from Morocco) prefer lower levels of immigration [@Coma2009].

Data

To measure policy preferences on a local level in detail, I use a pooled dataset of five annual consecutive representative surveys on immigration attitudes conducted by Centro de Investigaciones Sociológicas (CIS) in 2008-2012 (n = 13730). In turn, this allows me to conduct multilevel regression analysis on the level of provinces (n = 50)–a significant improvement over previous studies on the level of autonomous communities (n = 17). It should be noted that Spanish provinces are administrative divisions without much political relevance, which arguably helps avoid potential confounding related to elite-driven public opinion.

Furthermore, I use Census Data from 2001 and 2011 provided by Instituto Nacional de Estadistica (INE) to determine the demographic characteristics of various immigrant national groups across Spanish provinces and also trace their change (see Appendix). In further analysis, I only consider the 33 largest immigrant groups, which cover about 92% of all foreign-born in the country.

As for the general indicator of (anti-)immigration attitudes, I follow previous research and look at the survey items regarding respondents’ (dis)contentment with the amount of immigration in the country (on the scale 0-1). Specifically, the question was as follows: “In your opinion, is the number of immigrants currently in Spain: insufficient, acceptable, high, excessive?” For further empirical corroboration of my results, I also use the respective binary version of this variable and two alternative indicators of anti-immigration attitudes related to legal and irregular immigration policy (see Table A3 and A4).

To test the “economic threat” hypothesis (H1), I construct a province-level indicator for the average local level of immigrants’ skill (i.e., education). To test the “cultural threat” hypothesis (H2), I employ two competing variables indicating the proportion of immigrants in a province who are: (1) non-native Spanish speakers [@Escandell2009]; (2) from Muslim countries [@Coma2009].

To test the original “group hierarchy” hypothesis (H3), I follow the national stereotypes literature [@McCrae2013] and approximate the expected group status of immigrants in a province as the average GDP PPP per capita of their origin countries (World Bank). To make the measure more robust to idiosyncratic fluctuations, I take an average GDP estimate across the ten years prior to the census (2001-2011). Alternatively, I dichotomize the measure by calculating the proportion of immigrants from countries that have a higher or lower GDP per capita relative to Spain. While the proportion variable may be easier to interpret, the average measure likely captures the perceived status differences of immigrant groups in more detail (their correlation across provinces is 0.9). (Note: For example, while the both provinces of Almeria and Malaga had 9% of foreign population in 2001, the proportion of immigrants from richer countries was twice bigger in Malaga (46%) than in Almeria (23%). The high-status immigrant groups have the following origin: United States, Switzerland, Netherlands, Belgium, United Kingdom, Germany, France, Italy.)

The ecological validity of these group status variables arguably does not dependent on whether voters themselves deliberately use the country-of-origin GDP per capita as a reason to oppose or support a particular immigrant group (see discussion above). At the same time, the resulting variable of immigrants’ average group status is not only theoretically but also empirically distinct from the average immigrant skills that is used to test the economic threat hypothesis. Most important, the overall correlation of immigrants’ skills and the economic development of their origin countries varies from insignificant to low depending on a particular measure. Accordingly, the correlation of the respective contextual variables is also weak ($<$0.3). Overall, while immigrants from more developed countries naturally tend to have a higher socioeconomic standing than those who come from less developed countries, this relationship (at least in Spain) is not as strong as it may be expected due to well-known selection processes [@Grogger2011a]. For the relevant summary statistics and correlation matrix, see Figure A1 and Table A2 (as opposed to common stereotype, the majority of UK and other Western European migrants are not retirees).

All model specifications include fixed effects for each survey year, an average provincial GDP per capita, as well as the overall provincial foreign-born population share. As for control variables on the individual level, I employ a number of sociodemographic indicators proved to be significant in the previous research: gender, age, nativity, experience abroad, education, unemployment, and socioeconomic status [@Ceobanu2010]. To account for potential confounding variables, I also look at the proportions of poor and non-white immigrants, which may arguably be among the omitted variables that explain the observed relationship between the immigrant status and anti-immigration attitudes. The measure of poor is approximated with the percentage of individuals who live in unsatisfactory housing as defined in the Census (direct income data is not provided in the Census). All contextual predictors are scaled from 0 to 1. For details on variable construction, see Appendix.

Analysis

My analysis exploits the fact that Spanish foreign-born population has drastically changed over a relatively short time period, so that different provinces experienced diverse immigration inflows of various size. According to the conventional “group threat” hypotheses, anti-immigration attitudes should be more widespread in areas with more unskilled, Muslim and non-Spanish speaking immigrants. According to the “group hierarchy” hypothesis, however, anti-immigration attitudes should also be more widespread in areas with more immigrants from poorer countries even after controlling for major individual and contextual variables.

In order to test the competing hypotheses, I run a series of multilevel logit models in which I regress immigration attitudes on respondents’ personal characteristics and group-level covariates. First, I fit a random-intercept model with only individual-level demographic predictors taken from CIS (2008-2012). (Note: All covariates have a rather expected relationship (see Table tab:CDHrobust): older and unemployed people are more likely to have anti-immigration sentiments. At the same time, higher education, urban residence, higher socioeconomic status and any experience abroad (including immigration itself) have a negative relationship with individual proclivity to hold anti-immigration attitudes.) Since the intercept greatly varies across provinces, individual attitudes towards immigration are highly dependent on a particular local context in Spain. As seen from Figure fig:odds (based on the binary outcome for illustrative purposes), the province-level differences are quite striking and they certainly go much beyond the mere variation in regional economy and nationalism. For instance, after controlling for individual covariates, the odds of being anti-immigration in Tarragona (70%) are on average two-three times higher than in Alava (27%).

Figure. Variation of Anti-immigration Attitudes Across Spanish Provinces

What can explain such substantial regional variation in immigration attitudes? In line with existing research, people seem to be less anti-immigration in provinces with fewer immigrants (and better economies), but this effect can be attributed to a variety of underlying causes as discussed earlier. From previous findings, for instance, we know that natives discriminate between immigrants based on their country of origin and related characteristics.

To test the cultural threat hypothesis, I include the proportion of non-native Spanish speakers and those from Muslim countries (in addition to GDP per capita and the overall share of immigrant population). Although I cannot fully rule out the possibility that this is due to the variable construction, none of these variables seem to have an impact either separately (not shown) or together (see Table tab:DCHmain, 1). With regard to a possibility of economic threat (2), I do find a relationship with the average immigrant skill indicating that people are less restrictive in the contexts of “economically desirable” immigration. These effects, however, cease to be statistically significant once I account for group status.

Specifically, in the model (3) I include the proposed contextual indicator for capturing the average group status of immigrants in a province. As expected, this variable has a strong negative relationship with anti-immigration attitudes. Substantively, it implies that people are more likely to be anti-immigration if they live in provinces with greater proportion of immigrants from origin countries that are perceived to be lower-status. I also test a simpler alternative indicator of the proportion of immigrant population from richer countries with no change in the results. The predicted probability of considering the current immigration as excessive increases from about $262%$ to $422%$ once one moves from the observed higher to lower group status composition of foreign population. Quite equivalently, this implies moving either from 60% of immigrants from less developed countries to 97% or from the average origin’s GDP per capita of 10,000$ to 22,000$.

While I cannot claim that the coefficient for group status is larger than other contextual effects, the model (4) shows that the stipulated relationship is robust to including measures of economic or cultural composition. In other words, natives’ expression of immigrant threat in a province may be more a function of immigrants’ average relative standing in the ascribed international group hierarchy rather than their actual cultural or economic characteristics.

Table. Anti-immigration Attitudes: The Role of Difference, Competition, Hierarchy

Instrumental Variables and Robustness Checks

I take possible robustness concerns seriously and run a variety of different model specifications including the instrumental variable estimation to corroborate my findings. In sum, these alternative models confirm previous results and some even indicate a greater coefficient size for the group status variable. First of all, I test a number of alternative models with the use of logistic regressions on the binary dependent variable (see Table tab:CDHrobust), as well as ordinal logit models (not shown). Second, I confirm the findings are not driven by (a) the municipal proportion of two prominent national groups from Morocco or Romania and that they are robust to the inclusion of (b) the percentage-point change (from 2000) and its interaction with the share [@Newman2013]; (c) economic growth (from 2000); (d) urbanization; and (e) the controls for three national communities or their exclusion (not shown). (Note: While one can speculate that the effect of all tested compositional measures should depend on how large the immigrant population or its change are in general [@Hopkins2010, Newman2013], there is no significant interaction effects of group size with any other variable in the model. The possible reasons for this non-finding include a relatively low variation of group size itself and the low amount of contextual observations (i.e., 50 provinces) in the study.) Third, while various threat indicators may be significant in some models, I corroborate that the uncovered relationship consistently holds for other indicators of anti-immigration attitudes including the ones on legal and irregular immigration policy (See Table tab:CDHrobust2).

Finally, reverse causality is arguably unlikely in this case (i.e., immigrants from poorer countries are deliberately moving to more anti-immigration provinces even after accounting for local economy and urban residence). Due to the nature of cross-sectional analysis presented here, however, I cannot fully rule out the possibility that there is some omitted factor that predicts both anti-immigration attitudes and immigrants’ average group status in a province. To address this concern, I adapt an established technique in labor economics and exploit the transportation accessibility of different localities (“gateways”) and the group-specific settlement patterns of past immigrants (“ethnic networks”) [@Gonzalez2013]. While these variables have been used to predict the foreign-born population share across provinces in general, they can also instrument for the share of foreign-born population from poorer countries in particular. In short, the idea behind these two instruments is (1) that provinces are more accessible by land to some national groups than others and (2) that the current location decisions of migrants are influenced by the past decisions of their co-nationals. In turn, these instrumental variables should significantly affect the recent group status composition and should only affect the current levels of immigration attitudes through the recent immigrant composition.

Accordingly, I find that—after controlling for the overall share of immigrant population, the economy, and various individual covariates—both measures are strong instruments for group status in the first stage regression (F $> 10$) and are significant predictors of immigration attitudes in the second stage (see Table tab:CDHinst). Of course, there can be reasonable doubts about the exclusion restriction in instrumenting average group status of immigrants in 2011 with the one in 2001 even after controlling for major confounders. Nonetheless, the main purpose of the instrument is to corroborate the evidence obtained with the use of the more robust estimation based on transportation accessibility. Furthermore, some argue that the pull and the push factors of immigration in Spain during its unprecedented growth over 2001-2011 and before that were very distinct [@Gonzalez2013].

So far, I have established that, regardless of multiple contextual characteristics, people who live in provinces with a lower (greater) relative proportion of foreigners from richer (poorer) countries are more likely to express a variety of anti-immigration sentiments. It is still not clear, however, what is behind this macro-level relationship: namely, (1) whether or not people perceive group hierarchies, (2) how these perceptions are related to anti-immigration attitudes; (3) whether this relationship is affected by contextual group status composition and also (4) how generalizable this relationship is across countries. Although this may require a separate investigation, below I provide some initial empirical evidence concerning these questions. To do that, I first utilize the pooled eighteen annual consecutive representative surveys of the ASEP data (1991-2007) which gives a more detailed group-specific account of immigration attitudes in Spain. I then use the European Social Survey data (2002) to corroborate my account using a different set of group-specific items across countries (for detailed data description, see Appendix).

Exploring Mechanisms

According to the ASEP data, some immigrant groups in Spain—conceptualized in terms of their regional origin—are systematically preferred to others across both time and space (Figure fig:hierarchy). Although the relative position of Black Africans and Eastern Europeans gradually flipped from 1991 to 2007 (Note: While this may be a result of various processes such as divergent migration flows or strengthening of anti-racist norms, the reasons behind these patterns deserve further research.), both groups have always been less liked than Asians, Latin Americans, and Western Europeans. Nonetheless, despite the structural stability of the intergroup ranking, there is a lot of individual heterogeneity in the intensity of these group-differentiated perceptions. While 36% of the public does not make any intergroup distinctions at all, the rest discriminate between different groups to a various degree. To capture this dynamic, I created an individual-level measure of “group hierarchy preference” (GHP) indicating the variability (i.e., standard deviation) in likability for different groups or the extent to which people think that some groups are better than others.

As indicated by regression analysis, GHP—an indicator which likely relates to social dominance orientation—appears to be a significant determinant of anti-immigration attitudes (Table tab:hierarchycontext). Indeed, its coefficient is comparable to that of education—perhaps the biggest predictor in the literature. This is true even after accounting for absolute evaluations of different groups themselves and other covariates. In other words, this analysis demonstrates that, overall prejudice aside, those who make distinctions between groups in likability are more likely to oppose immigration than those who do not.

Finally, a multi-level model with a cross-level interaction indicates that the effect of GHP is moderated by local group status composition, so that it is more consequential in provinces with more immigrants from poorer countries (see Figure fig:interaction). Overall, this suggests a possibility that people’s group hierarchy preferences are more salient in contexts of lower-status immigration. Consequently, in such contexts voters may be more likely to use these preferences when formulating their opinion on immigrants and immigration more generally. At the same time, it implies that the contextual relationship uncovered earlier is more relevant to people who tend to perceive and endorse group hierarchies.

Figure. Anti-immigration Attitudes and Group Hierarchy Preferences: The Role of Context

Exploring Generalizability

As demonstrated above, inasmuch as group hierarchy is only one of the factors driving immigration attitudes, its explanatory power depends on a particular context. Unfortunately, other immigrant-receiving countries may not be as well-suited as Spain for testing the contextual implications of my argument since economic, cultural, and status characteristics of immigrant groups are largely collinear there. At the same time, one can argue that the Spanish context is unique in the sense that voters there are particularly likely to respond to group status of immigrants’ national origin due to the novelty of large-scale immigration in the country. Consequently, it is conceivable that national origin status may be a less applicable factor in other national contexts with established domestic racial hierarchies (e.g., United States), far-right nativist politics (e.g., France), or significant political shocks (e.g., the influx of refugees). To address the generalizability concerns to some degree, I rely on the European Social Survey data with a special immigration module (2002). While the ESS data has been used for testing the conventional cultural/economic threat dichotomy, it can still shed some light on the role of group hierarchies by providing cross-national evidence on the individual attitudes toward different immigrant groups.

Overall, the ESS provides six admission policy items differentiating between immigrants from richer and poorer countries inside and outside of Europe, as well as people of the same or different ethnicity. The “group hierarchy” hypothesis implies that immigrants of higher group status are generally preferred to those of lower group status. Despite the potential for social desirability bias, the ESS indicates that most respondent indeed prefer richer European immigrants of the same ethnicity across most countries [@Hainmueller2007]. While it is impossible to know what exactly drives these preferences, it is suggestive that the tendency to differentiate between groups based on different dimensions is significantly correlated. Consequently, similar to the analysis before, I created an GHP index summarizing the extent to which one makes group distinctions for admission policy.

As in the Spanish data described earlier, about half of the general European public explicitly discriminates between various immigrant groups. Using various distinct indicators of anti-immigration attitudes, I then run a number of multilevel OLS regressions with GHP as the main explanatory variable. As can be seen from Table tab:hierarchyEU, GHP remains a strong determinant of various anti-immigration attitudes even after accounting for the average attitudes toward different immigrant groups themselves. At the same time, a random slope model reveals that GHP is a similarly important predictor of anti-immigration attitudes across all European countries (not shown). Although one may object to regressing one set of immigration attitudes on another, this analysis?based on intergroup attitude variability—makes it clear that many people do perceive group hierarchies which are (at least partially) based on national development and that these perceptions are robustly related to anti-immigration attitudes across Europe. (Note: Although a small fraction of people (13%) even prefer lower to higher status immigrant groups, their removal from the analysis does not change the results. As an additional robustness check, I also exclude people who universally oppose or support immigration of any group with no change in the results (not shown).)

Table. Anti-immigration Attitudes and Group Hierarchy Preferences in Europe

Would my argument hold for immigrant-receiving developing countries? While I do not aim to provide a direct test here, there is an ample evidence consistent with my conjecture that people are less eager to oppose immigration from rich countries regardless of their characteristics. In addition to the occasional rejection of co-ethnic migrants, for instance, @Carr1996 documents the widespread outgroup favoritism of Western expatriates in Africa. Western foreigners are also viewed more favorably than culturally proximate Chinese immigrants in Hong Kong and, what is especially intriguing, sometimes even more favorably than the natives themselves [@Lim2003a, Cuddy2008]. @Coates2005 further show a similar pattern in New Zealand.

Discussion

Politicians and their electorates around the world are often willing to express their worries regarding the potential negative impact of immigration on their countries. The literature on the topic has closely followed these popular economic and cultural discourses, but a diversity of critics are skeptical whether such lay perceptions of threat are exactly what causes anti-immigration attitudes and whether such justifications of exclusion should be taken at face value and used as analytical categories in research. In line with these ideas, the analysis presented here suggests that the explanatory power of group threat is limited. At the same time, it indicates that people may also form immigration attitudes by relying on the development level of immigrants’ country of origin as a group status heuristic for their relative worthiness.

It is true that people may dislike certain immigrant groups from poorer countries simply because they are (perceived as) likely to be lower-skilled [@Hainmueller2014], which is still in line with some of the economic group and individual threat arguments [@Hainmueller2015]. However, it appears hard to reconcile this idea with the non-significance of immigrants’ average education (i.e., a proxy for individual status) on the one hand and the robust impact of immigrants’ GDP of national origin (i.e., a proxy for group status) on the other. Moreover, the inclusion of skill variable does not significantly change the coefficient size of group status variable, indicating that anti-immigration attitudes cannot be easily reduced to economic self- or group interests.

It is also possible that some of the national origin prejudice may be attributed to the major alternative form of social stratification related to class distinctions [@Blalock1967]. Indeed, studies show that poor people are usually as enthusiastically disliked and scorned as some of the most disadvantaged ethnic groups [@Lott2012]. As a result, both poor and ethnic minorities (including most immigrant groups) are often conceived as less warm, less competent, and even less human [@Cuddy2008, Loughnan2014]. Although there has been little research on class prejudice and its interaction with ethnic prejudice, some literature shows that the availability of unambiguous ethnic stereotypes in the context of significant ethnic inequalities makes the distinction between class and ethnic prejudice meaningless unless there is some incongruence [@Calavita2000, Weeks2004a]. (Note: Under the conditions of extremely low social mobility, class—a ranked category by definition—becomes more descent-based and thus reminiscent of ethnicity. At the same time, under the conditions of extremely high international inequality and low international mobility, ethnicity/nationality—a non-ranked category—becomes ranked and thus reminiscent of class.) Therefore, regardless of whether the antecedents of anti-immigration attitudes observed in this paper are in fact class- or ethnicity-based, they are at least partially related to the perceived status of immigrant groups. After all, reacting to the transgression of existing group hierarchies by lower-status migrants is both conceptually and, as shown here, empirically distinct from promoting economic group interest or feeling threatened by cultural difference.

Finally, another possible interpretation of the results is that people simply do not have groups of higher-status in mind when they consider the social category of “immigrants.” The Spanish case supports this idea: while immigrants from more developed countries constitute at least 20% of the foreign-born population, only 1% of respondents mention any of them in the open-ended survey items that specifically ask people to name various immigrant groups. Referring to Spain as a prototypical example of such innumeracy, @Braun2013 similarly notes that “[i]t is particularly with regard to EU-15 migrants that the representation of objective migrant-group sizes in the minds of respondents seems to fail.” Such interpretation, however, can only reinforce the argument proposed here since such demarcation is clearly based on immigrants’ group status rather than culture or skills per se.

Conclusion

There has been increasing scholarly agreement that native attitudes toward various immigrant groups are based on perceived sociotropic or group threat related to how economically undesirable and culturally different these groups are. In line with this, many contextual theories also predict that the quantity and the quality of local foreign populations should be consequential for immigration attitudes. Taken together, these arguments presumably explain the backlash to Hispanic immigrants in the US and to Muslim immigrants in Europe.

Since its foundation, the group threat theory has provided the literature with an important insight that opposition to immigration cannot be fully reduced to self-interest and has a significant social component. However, it has also become increasingly apparent that identity politics may be based on more than just group difference and competition. The present study suggests that the “group threat” metaphor may not be the only way to capture the relevant group dynamic.

While misperceptions about the size of the immigrant population are robustly linked to anti-immigration attitudes, for instance, recent experiments find no consistent evidence that people exposed to correct information change their views on the issue [@Hopkins2018]. Moreover, it is possible that even genuine agreement with such statements as “immigrants undermine our culture” or “immigrant take our jobs” may simply be post-hoc rationalizations of a prejudicial gut feeling [@Talaska2008]. Accordingly, there is a growing literature that looks at the attribution of group difference (or competition) as an expression of prejudice [@Vala2009] and at the legitimizing role of perceived threat in general [@Pereira2010, Hartman2014]. Recent experimental evidence has confirmed that “threat perception can be used as a way to explain the experience of prejudice, rather than forming the source of the prejudice itself” and that “[i]f negative feelings for a group already exist, beliefs that the group is threatening are likely to follow” [@Bahns2017].

Consistent with this growing literature, I demonstrate that the very appeal of the group threat account may largely be an artifact of the study of immigration attitudes in highly developed countries where the overwhelming majority of lower-skilled immigrants come from not only culturally distinct but also much poorer countries that may be considered “inferior” by the better-off native populace. In turn, this makes it impossible to discern whether certain immigrant groups are perceived “threatening” because of their cultural distinctiveness and economic unattractiveness or rather due to their lower group status derived from a stereotyped image of its relative standing in the global ethnic hierarchy of national origin. As demonstrated by the case of Spanish provinces, immigration attitudes can indeed be predicted by immigrants’ group status in addition to the cultural or economic composition of the immigrant population. The independent role of group status is further corroborated by the instrumental variable analysis and various robustness checks.

Of course, the presented empirical evidence is not without some limitations. For instance, the coding of cultural variables inferred from the country of origin may be imprecise in some instances, which may potentially lead to measurement error. Furthermore, the analyzed cross-sectional survey data with a main independent variable on the contextual level can provide only indirect evidence for the proposed argument due to the potential of ecological fallacy. Finally, despite the unique advantages of using Spain to test my argument, one can have concerns about generalizability of the proposed contextual relationship and the underlying mechanisms. That said, the additional micro-level evidence of differentiated anti-immigration attitudes in Spain, as well as its contextual determinants and generalizability across Europe help mitigate these concerns to some extent.

Future studies could address some of these issues further by creating more fine-grained contextual measures, as well as exploring the role of group hierarchy in developing countries (where natives may even exhibit outgroup favoritism toward high-status foreign groups). Moreover, the research on differentiated immigration attitudes might consider utilizing conjoint experimental designs that manipulate group status independent of difference and competition. Finally, social scientists more generally would benefit from richer qualitative accounts of how voters think about group status hierarchies of national origin.

In sum, this study goes beyond the prevailing “economy versus culture” dichotomy by questioning the explanatory power of group threat and highlighting the role of group status in politics. While perceived cultural difference and economic competition are certainly important drivers of popular opposition to certain immigrant groups, my results challenge the conventional interpretation of this opposition solely through the prism of threatened interests and values. Instead, the evidence suggests that immigrant groups of higher status are generally preferred to those of lower status independent of other economic and cultural factors. As a result, even culturally similar and economically beneficial immigrant groups from poorer countries can face public opposition due to their lower-status national origin.

References

See source bibliography.

References

See source bibliography.

Appendix

Data Sources

Individual variables: Centro de Investigaciones Sociológicas (CIS), “ACTITUDES HACIA LA INMIGRACIÓN,” 2008-2012
Contextual variables: Spanish Census (INE) 2011, CIA World Factbook, World Bank

Variable Descriptions

Anti-immigration attitudes (scale 0-1 or binary, higher values indicate more restrictive attitudes)
Immigrant number: “In your opinion, the number of immigrants currently in Spain is?” [Insufficient (more needed); Acceptable; High; Excessive]
Legal (policy restrictiveness): “In your opinion, do you believe that the laws governing the entry and stay of foreigners in Spain are too tolerant, rather tolerant, correct, rather harsh or too harsh?”
Irregular (policy restrictiveness): “Considering the immigrants who are already in Spain but who have not regularized their situation (they do not have the required residence permits), what do you think would be better?” [Regularize them all; Regularize only those who have been living in Spain for several years, whether or not they work; Regularize only those who have work today, regardless of the time they have been in Spain; Let them remain as they are; Deport them to their country of origin]
Contextual predictors (province level for 2011, scale 0-1):
Cultural group threat: Percentage of Non-native Spanish speakers among foreign-born population; Percentage of foreigners from (predominantly) Muslim country background among foreign-born population
Economic group threat: Average education among foreign-born population
Group status hierarchy: Average GDP per capita of sending countries among foreign-born population; Percentage of foreigners from richer countries among foreign-born population
Instrument 1: The distance from province capitals to the French border along highways
Instrument 2: Percentage of foreigners from richer countries among foreign-born in 2001
Robustness checks: Percentage of foreigners from (predominantly) non-white countries among foreign-born population; Average poverty (poor housing conditions) among foreign-born population;
Control variables
Individual level: Age, Gender (Female == 1), Nativity, Experience abroad, Education (college or higher == 1), Socioeconomic status, Unemployment, Year fixed effects
Province level: Province’s GDP pc in 2010, Overall percentage of foreign-born population
Additional data for “Exploring mechanisms”
Source: Centro de Investigaciones sobre la Realidad Social (CIRES), “Attitudes of Spaniards towards Immigration, 1991-2007” (ASEP)
Anti-immigration attitudes 2-item index:
“In general terms, what would you say with respect to the number of people of a different nationality, race, religion, or culture who live in our country? Do you think there are too many, many but not too many, or not many?”
“Could you tell me if, basically, you very much agree, agree, disagree or very much disagree with each of the following sentences that I’m going to read to you:? Citizens of any country should have the right to settle in in any other country, without any kind of limitations.”
Group hierarchy preference (GHP): “I will now read a list of people from various areas of the world. In a scale from 0 to 10, please tell me how much do you like each one of them, where 0 means you don’t like them at all and 10 means you like them very much.” [calculated as the sd in the evaluation of Asians, Eastern Europeans, North Americans, Russians, Arabs and Muslims, Western Europeans from the EU, Gypsies, Jews, Latin Americans, Black Africans]
Mean attitudes: the same as above [calculated as the average]
Control variables: individual and contextual covariates are similar to CIS
Additional data for “Exploring generalizability”
Source: European Social Survey (ESS) 2002
Anti-immigration qualifications 8-item index: “Please tell me how important you think each of these things should be in deciding whether someone born and raised outside [country] should be able to come and live here.” [education, family, language, religion, race, wealth, skills, commitment]
Anti-immigration rights 5-item index: e.g., “Please say how much you agree or disagree with each of the following statements. If people who have come to live and work here are unemployed for a long period, they should be made to leave” [imsmrgt, imunplv, imscrlv, imacrlv, stimrdt]
Anti-immigration beliefs index: e.g., “Would you say that people who come to live here generally take jobs away from workers in the UK, or generally help to create new jobs?” [imtcjob, imbleco, imbgeco, imueclt, imueclt, imwbcnt, imwbcrm]
Anti-immigration prejudice 4-item index: e.g., “Now thinking again of people who have come to live in the UK from another country who are of the same race or ethnic group as most people in the UK, how much would you mind or not mind if someone like this was appointed as your boss?” [imsetbs, imsetmr, imdetbs, imdetmr]
Group hierarchy preference (GHP): “ To what extent do you think [country] should allow people from [group] to come and live here?” [calculated as the average difference between admission attitudes toward immigrants from richer, European, ethnically similar countries and poorer, non-European, ethnically different countries]
Mean attitudes: the same as above [calculated as the average of all admission attitudes]
Control variables: individual covariates are similar to CIS

Tables and Figures

Atable

Table. Summary Statistics for Contextual Covariates (2011)

Afigure

Figure. Correlation Matrix for Major Contextual Covariates (2011)

Atable

Table. Socioeconomic Differences of Natives and Immigrants of Different Origin (2011)

Atable

Table. Anti-immigration Attitudes and the Role of Group Hierarchy: Robustness Checks

Atable

Table. Various Forms of Anti-immigration Attitudes: The Role of Group Hierarchy and Individual Covariates

Atable

Table. Various Forms of Anti-immigration Attitudes: Instrumental Variable Estimation

Afigure

Figure. Likability Hierarchy of Foreign Groups in Spain Across Time and Space

Atable

Table. Anti-immigration Attitudes and Group Hierarchy Preferences: The Role of Context