Do Immigrants Ever Oppose Immigration?

Executive Summary

In both the 2016 and 2020 elections, a large number of Latin American immigrants voted for President Trump—despite his strong stance against open immigration. Similar support for anti-immigration candidates was observed in the 2022 midterm elections.

This surprising data exposes an understudied area of research: How do immigrants already settled in a host country feel about other immigrants?

It may seem reasonable to expect that immigrants support immigration as it reinforces group identity, solidarity, and networks. However, as the election data suggests, this is not always the case.

In this paper, researchers Massimiliano Tani and Aflatun Kaeser explore the question of immigrant support for other immigrants. Using US data from the World Value Survey, the authors attempt to better understand the circumstances that affect whether immigrants support immigration as a policy issue.

The authors find that high-skilled, successful immigrants are more likely to oppose immigration. This may stem from a worry that natives around them could have negative experiences with low-skilled immigrants and form stereotypes about all immigrants based on those poor experiences.

For example, researchers find more opposition to immigration among respondents from countries that have wider income and education ranges, making it harder for host natives to distinguish the ‘quality’ (e.g., education, talent, income) of one immigrant over another. However, opposition nearly disappears among respondents in more selective host countries such as Australia, Canada, and New Zealand. In these countries, immigrants must qualify to enter via point systems, and so the ‘quality’ of immigrants is perceived as higher overall.

In the United States, where undocumented immigration dominates public conversations, researchers find greater opposition to immigration among established successful immigrants.

These results indicate that immigration is more likely to be viewed negatively when natives experience a lot of unknowns regarding immigrants. When it is difficult for natives to distinguish the ‘quality’ of immigrants—especially when the inflow of immigrants is poorly managed—negative immigration views and voting choices are more likely to proliferate among natives as well as successful immigrants. Incorporating more selective visa programs for the US may improve attitudes of immigrants towards other immigrants.


After the 2016 and 2020 US presidential elections, many analysts were surprised by the high number of immigrants from Latin America who voted for President Trump despite his strong stance against open immigration (Cadava 2021; Corral and Leal 2020; Garcia 2021; Ostfeld 2019; Russonello and Mazzei 2021). Such political support reemerged in the campaign for the mid-term election of November 2022 (Winknews 2022), suggesting that immigration does not automatically attract every immigrant’s support. Immigration, more generally, is indeed a hot political topic in the United States and other immigration countries. Supporters highlight its role in alleviating critical skill shortages and making a positive contribution to the host country’s economy and growth. Detractors emphasize its potential negative effects on the competition for jobs and housing, crime, and the dilution of the sense of national identity.

These contrasting images of immigration reflect its influence on both labor demand and supply and the mixed evidence about the empirical effects found so far (Tani and Piracha 2022). In theory, new immigrants should adversely affect natives with identical characteristics because they increase the supply of labor among those with the same features. In practice, immigrants are often employed below their actual human capital level, as measured by qualifications and work experience, and so they do not work in the same occupations as similarly skilled natives. As a result, the comparison of immigrants and natives within given job sets yields inconclusive results. Furthermore, natives tend to move “horizontally” by specializing in jobs relying on language skills, an area in which immigrants cannot compete (Peri and Sparber 2009, 2011). This makes it unlikely to find statistically significant effects even when the observed employment outcomes of immigrants and natives are better matched.

Aside from the analysis of its economic impact, a recent and growing literature has focused on natives’ attitudes toward immigration as a key factor of influence underpinning voting preferences (e.g., Hainmueller and Hopkins 2014; Levinson et al. 2010; O’Rourke and Sinnott 2006; Valentino et al. 2013). This literature has pointed out that natives’ resentment toward immigration arises in several areas, which include the perception that immigrants “steal jobs” from natives (Niyimbanira and Madzivhandila 2016), worsen security and increase crime (Ousey and Kubrin 2018), and even contribute to an increased fear of a terrorist attack, especially after the World Trade Center attack on September 11, 2001 (Choi 2021). Such resentment is likely undesirable and costly as it works against social cohesion, cooperation, and upward mobility. It may also contribute to harboring extreme political views and preferences that hinder societal betterments. The literature tends to explain these negative attitudes by pointing to the poor education of respondents and the associated higher likelihood of misinformation from “fake news” spread through social media (Wright et al. 2021). To combat such misinformation and extremism, this line of research recommends, more or less implicitly, supporting educational and “fact checking” initiatives.

Notwithstanding the above, little research exists on how immigrants already settled in a host country feel about other immigrants. A priori, one might expect all immigrants to support immigration, as it reinforces group identity, solidarity, and networks. But this is not the case, as the Latinos’ voting support for President Trump in the 2016 and 2020 US presidential elections has shown. While this evidence signals that it is possible for immigrants to oppose immigration, research on this topic is limited. Some insights are offered by studies highlighting that immigrants are a heterogeneous group (Garcia-Rios et al. 2019), and as such they may respond differently to certain challenges like the perception of relative deprivation (Berry et al. 2022), but immigrants’ views of immigration are otherwise of marginal interest in studies of attitudes of the broader resident population (e.g., Card, Dustmann, and Preston 2005).

Overall, immigrants’ attitudes toward immigration remain under researched: Does being an immigrant carry enough weight to hold only positive views about immigration? Do (some) immigrants, upon settling in the destination country, try to “forget” and hide their origins by taking a strong anti-immigration stance, and if so, why?

We aim to address these questions by offering an explanation capturing the key insight of existing economic and political science research on attitudes toward immigrants―namely that immigration attitudes arise from group and cultural rather than personal and economic concerns (Hainmueller and Hopkins 2014; Javdani 2020)―which we formalize as “cultural unawareness and prejudice” in a theoretical model. We construct such unawareness as a cause of statistical discrimination that natives apply toward immigrants: in other words, we assume that natives view nonnatives first as immigrants and then as Colombians, Italians, Senegalese, British, and so forth. However, immigrants have better understanding of the immigrant origins of other nonnatives.

We also assume that immigrants’ utility in the host country (e.g., access to quality goods and services, as well as opportunities) depends on natives’ views about immigration. This feedback loop is critical to generating immigrants’ responses to immigration. In particular, more positive views about immigration among natives will benefit existing immigrants, and the converse also holds. Furthermore, benefits and losses to immigrants will be higher the more immigrants benefit from being disassociated with the immigrant flow that most likely raises negative views about immigration at large among the native population (e.g., undocumented immigrants). We assume that the most economically assimilated and successful immigrants are the most exposed (or at risk) to natives’ negative views about immigration (Zimmermann et al. 2000). These assumptions yield novel testable propositions based on rationality and incentives that may explain (some) Latinos’ anti-immigration vote in the 2016 and 2020 US Presidential elections. We test these hypotheses with data sourced from the World Value Survey (WVS) for the United States as well as other countries.

We find that differences in immigrants’ type and background create incentives for successful immigrants to oppose immigration and associate it with worsening unemployment, crime, and the risk of terrorism. We also show that this result positively correlates with heterogeneity in the countries of origin of immigrant respondents (e.g., in income and human capital differences), as this may negatively affect natives’ ability to distinguish whether immigrants are from the top or bottom tail of the income distribution and negatively correlates with the existence of selective immigration policies in the host country where they now live, as this may reassure natives of immigrants’ “quality.” Immigrants’ anti-immigration views do not arise in host countries that screen immigrants as Canada, New Zealand, and Australia do. In contrast, there is strong opposition to immigration among immigrants when undocumented immigration is high in numbers, as in the United States.

These results are consistent with the hypothesis that the difficulty of distinguishing between skilled and unskilled or legal and undocumented immigrants in the native population can exacerbate views and voting choices against immigration when the inflow of immigrants is highly heterogeneous and poorly managed. The rest of the paper is organized as follows: in Section 2 we review existing literature on attitudes toward immigration. In Section 3, we develop a simple model of statistical discrimination. Section 4 focuses on data and methodology, while Section 5 presents the results. Section 6 concludes.


The United States is known as a prime example of an immigration country, as less than 2 percent of its citizens have indigenous ancestors. However, perceptions about immigrants among the local population vary significantly. A person’s perception is formed by the cognitive interaction with his/her surroundings (Efron 1969), and information and beliefs about immigrants form the perception about them as well as attitudes toward them, regardless of whether these are consciously held. Immigration in the United States generates opposite views (Mears 2001). Historically, anti-immigration attitudes driven by economic (Timmer and Williams 1998) and cultural factors (O’Rourke and Sinnott 2006) have predated the introduction of restrictive immigration policies such as the Chinese Exclusion Act in the United States. Although people do not perceive themselves as consciously having discriminatory attitudes toward those of different races and nationalities when asked directly, unconscious bias is nevertheless likely. Such unconscious bias may occur, as people focus on individual features when looking at the members of an in-group, but they fail to notice these details in out-group people (Levinson et al. 2010). Hence, negative perceptions about immigrants not only arise and persist, but they prevent the realization of similar empathy levels toward immigrants among native adults and children (Rutland et al. 2005; Winkler 2009).

Research on immigration attitudes predominantly focuses on the perception among natives. Two broad strands of literature can be distinguished. The first, which dominates in economics discourse, investigates whether natives’ negative views originate from their fears of additional labor market competition from immigrants (Dustmann, Glitz, and Frattini 2008; Facchini and Mayda 2012; Scheve and Slaughter 2001) or the potential burden than immigrants place on the nation’s finances (Dustmann and Preston 2007; Newman and Malhotra 2019). Empirical evidence on causal effects, however, find that labor market competition between natives and immigrants has not only diminished over time (Card 2005) but that immigration does not carry negative economic consequences for natives (US data: Friedberg and Hunt 2018; Gaston and Nelson 2000; UK data: Dustmann, Fabbri, and Preston 2005), including on wages (Mayda 2006; Niyimbanira and Madzivhandila 2016) and public finances (Hainmueller and Hopkins 2014; Rowthorn 2008).

The second stream of research, drawing from multidisciplinary studies from economics and political science, attributes the roots of natives’ negative attitudes to cultural concerns about loss of group identity and homogeneity (Card, Dustmann, and Preston 2005; Dustmann and Preston 2007; Newman and Malhotra 2019; Sides and Citrin 2007). These studies find that cultural factors have a stronger explanatory power than personal economic motives in explaining negative views about immigration (Javdani 2020; Newman and Malhotra 2019). Therefore, interventions on prejudice, such as better education and opportunities for social interaction, may reduce natives’ reservations about immigration. Furthermore, the existence of selective immigration policies can reassure the resident population that local traditions and customs are valued and will be upheld (Facchini and Mayda 2012), while there will be no tolerance for immigrants’ contribution to crime (Bianchi et al. 2012; Mears 2001; Sampson 2008) or terrorism (Helbling and Meierrieks 2020).

Existing research has also noted that attitudes toward immigration differ by ethnicity, political preferences, and socioeconomic status. In the United States, native whites have more negative views about immigration and refugees than people of other backgrounds (Bodvarsson and Van den Berg 2013; McKeever et al. 2012; Valentino et al. 2013). Women have more positive attitudes than men toward immigration (Rocha et al. 2015; Watson and Riffe 2013). Republican and conservative ideologies tend to be associated with negative views toward immigrants (Gil de Zúñiga et al. 2012). People in occupations where there is a high immigrant share are also more likely to oppose immigration (Mayda 2006).

Despite the extensive literature on immigration, there is little analysis of views among natives. Even when some evidence is empirically reported (e.g., Card, Dustmann, and Preston 2005), there is little follow-up analysis on immigrants’ views. As a result, it remains unclear how positive determinants such as kinship, solidarity, and shared experiences with other immigrants, on the one side, and negative determinants such as the allegiance to the host society, on the other (Just and Anderson 2015), shape immigrants’ views about immigration. We aim to contribute to the literature by developing a model that formalizes the notion of cultural unawareness among natives and immigrants to shed some light on the forces at work.



To frame the role of cultural concerns and prejudice toward immigrants, we make recourse to the notion of statistical discrimination―that is, the replacement of individual characteristics with their group averages when the ratio of noise to signal is high. Immigrants’ habits, norms, and culture can be more or less familiar to natives depending on whether or not they have a common language, history, institutional setting, and traditions. In particular, we build on the maintained assumption that the further away immigrants’ set of habits, culture, and behaviors are from natives’ references, the less likely it will be for natives to understand immigrants’ signals (e.g., the quality of their human capital) and, conversely, the more likely that natives will engage in statistical discrimination. Such cultural unawareness and prejudice prevents natives from fully understanding immigrant types, especially those from very different cultures, and makes them rely on an “average immigrant” construct on which the well-being of all immigrants in the host country by assumption depends. As long as immigrants and newcomers are perceived as making a positive economic contribution to the host country, natives will support immigration. In contrast, we assume that immigrants have better understanding of immigrant types. This gives them an edge over natives to anticipate if other immigrants and newcomers enhance or diminish the perception and “value” of the average immigrant held by natives. If immigrants consider some immigrant subgroups to negatively affect such an average immigrant construct, they will respond by opposing immigration and vice-versa.

This construction innovates on traditional models of immigration, as it avoids forcing the introduction of a priori preferences for or against immigration while creating a unity-maximizing mechanism entirely based on incentives about immigrants’ influence on the average immigrant as perceived by natives. By voting for or against immigration, existing immigrants effectively become a “screening device” for immigration in lieu of formal institutional mechanisms enshrined in immigration policy. This theoretical setup yields several relevant, novel, and testable propositions.

Theoretical Setup

Consider immigrant i whose utility in the host country depends on his/her individual characteristics as well as natives’ views about immigration―that is, the combination of awareness and perceptions about immigrants’ contribution to the host economy and society based on beliefs, stereotypes, and observations from experience of existing immigrants.

For simplicity, we assume that immigrants from all over the world are on a continuum in terms of abilities and productivity and that they can “understand” other immigrants’ types, so they can see their quality and productivity better than an equivalent native. Existing immigrants are by assumption endowed with the ability to understand the type of new immigrants arriving and gauge whether they will enhance or reduce natives’ perception of the average immigrant as introduced in the previous section. In contrast, natives do not to have this ability and statistically discriminate against immigrants by associating immigrant types with the average characteristics and behaviors of immigrants already in the host country. Thanks to the feedback link with natives’ perception of immigrants, existing immigrants will rationally decide whether or not to oppose immigration: if their perceived quality of new or existing immigrants is deemed to enhance natives’ view of the average immigrant―for example, as good citizens and law-abiding, productive workers―they will support immigration because better natives’ views of immigrant can generate better/more goods and services for existing immigrants and vice-versa.

The utility of an existing immigrant i can be thus summarized by:

    \[U(i)=f\left(y, X_i\right)\]

where is a function with f^{\prime}>0 and f^{\prime \prime}<0is US natives’ average perception of immigrants (a group indicator of average type), and X is a set of individual i’s observed characteristics. These include gender, age, English language skills, marital status, prior education and work experience, to name a few.

By definition, y affects the well-being of existing immigrants. However, for natives, the variable y is an imperfect indicator of immigrant type, as they observe:


where  is the immigrant’s true type (e.g., productivity) and  is a normally distributed error term with mean zero and constant variance. In contrast, existing immigrants view .

This difference in understanding the type of prospective immigrants has important consequences. For natives, new immigrants are welcome as long as E(q) \geq 0–that is, as long as their expected contribution to the host economy is positive. But the relevant discriminating variable for existing immigrants is instead E(q \mid y): the type of immigrant q conditional on the imperfect signal y. E(q \mid y) arises from the regression equation (Aigner and Cain 1977):

    \[E(q \mid y)=\bar{q}+(y-\bar{q}) \beta=\bar{q}(1-\beta)+\beta y\]

where \bar{q} is the group (i.e., average) indicator of the immigrant type, and \betameasures the reliability of the perceived immigrant type, which depends on the distribution of q and the noise u, that is:


As \frac{\partial E(q \mid y)}{\partial q}>0 it follows that \frac{\partial U(.)}{\partial q}=\frac{\partial U(.)}{\partial E(q \mid y)} \cdot \frac{\partial E(q \mid y)}{\partial q}>0. In other words, the utility of an existing immigrant i depends positively on natives’ perceptions of immigrants and the new immigrant’s type relative to his/her own, as discussed below.


Consider the utility of a high-q existing immigrant and his/her reaction to the immigration of an identical high-q immigrant from the same country. The existing immigrant has a better understanding than natives of the actual type q of the prospective immigrant. She/he also knows that even if natives only observe the imperfect signal y, the new immigrant with q similar or higher than the q of the existing immigrant will raise \bar{q} and E(q \mid y). Should existing immigrant i welcome or oppose the arrival of a high-q immigrant or an immigrant that can pull up the average immigrant? The model predicts welcoming, as noting with \neg q the type of newcomer, if \neg q \geq q then \frac{\partial U(.)}{\partial E(q \mid y)} . \frac{\partial E(q \mid y)}{\partial q}>0.

Consider instead the case of an existing immigrant facing the prospect of a new immigrant with a lower q, that is, \neg q<q. In this case, the existing immigrant’s utility is negatively affected by the new arrival, as immigration will reduce \bar{q}–the group indicator of immigrant types in the host country. The existing immigrant will therefore strongly oppose immigration, as \neg q<q and \frac{\partial E(q \mid y)}{\partial \neg q}<0.

The model also predicts that existing immigrants will oppose immigration more strongly the greater the distance between their type and that (lower) of new prospective immigrants.

Opposition to immigration will also be strong if the reliability of information about immigrants is poor―in other words, if natives have limited or no understanding of the distinction between immigrant types (i.e., if var(u) is large). This would be the case for countries of origin that are culturally, linguistically, and organizationally very different from the host country and do not share the way in which the host economy and society are organized and operate.

This simple application of statistical discrimination also predicts that opposition to immigration can diminish if host countries control undocumented immigration or introduce a selective immigration policy, as that reduces the distance between immigrant types and improves the reliability of \beta. For example, high immigration countries such as Australia, Canada, and New Zealand restrict permanent immigration to particular types of immigrants by giving higher weight to graduate and postgraduate education as well as work experience in jobs that are in high demand. Screening migrants reduces uncertainty about q and u and results in a lower \beta. In contrast, large differences between immigrant types and statistical discrimination, which are more likely when host countries admit immigrants from places of origin with which they have little in common and do not select them on the basis of their qualifications and skills as in the United States, are predicted to generate strong opposition to immigration from existing immigrants who have achieved above average success in the host country―that is, those with the highest q.

The expected probability to support or oppose immigration can be linked to the effect of immigration on personal utility via:

    \[\operatorname{Pr}\left(\mathrm{Y}_{\mathrm{i}}\right)=\{1 \text { (oppose) or } 0(\text { support })\}=\mathrm{E}\left(\frac{\partial U(.)}{\partial q}\right)\]

where Y is the expected probability of individuals i to oppose or support immigration based on the anticipated effect on their personal utility, modeled as a function of individual characteristics, circumstances (including one’s immigration status and relative standing in the host society), institutional settings \left(\mathrm{X}_{\mathrm{i}}\right), and parameters (a, b):

    \[\operatorname{Pr}\left(Y_i\right)=a+X_{\mathrm{i}} \mathrm{b}\]

Testable Hypotheses

This simple application of statistical discrimination about immigrants’ quality in the host country produces some remarkably sharp hypotheses:

H1: high quality immigrants will oppose new immigration if there is a sufficiently high likelihood that new immigrants are of low-q regardless of whether immigrants are documented or undocumented. The opposite occurs for low-q immigrants;

H2: the strongest opposition to immigration will originate from immigrants who have achieved above-average success in the host country.

H3: anti-immigration reactions will be stronger in the case of places of origin where the distribution of q in the emigrant population is highly dispersed;

H4: anti-immigration reactions among high-q immigrants will be stronger in the case of host countries that do not screen immigrants, have large inflows of undocumented immigrants, or admit immigrants from countries with which they have little in common.

These hypotheses can be tested in a linear probability model where immigrants’ standing in the host society (which underpins their wariness of natives’ cultural unawareness and prejudice) is explicitly included \left(S_i\right) in the probability of opposing or supporting immigration. This yields:

    \[\operatorname{Pr}\left(Y_i\right)=a+X_i b_1+b_2 I_i+b_3 S_i+b_4 I_i S_i\]

where the interaction term I_i S_i captures immigrants’ rational assessment of the anticipated effect of immigration on natives’ perceptions of the average immigrant.


We test H1–H4 by studying the responses of immigrants of different socioeconomic status in the United States using data from the WVS. The WVS is a research project run globally to understand people’s values and beliefs and how they change over time using an identical questionnaire, which is internationally comparable. The WVS started in 1981, and it collects data every five years. The WVS has several limitations including its nature and limited sample size, but these disadvantages are balanced by the rich array of questions about opinions and beliefs and the use of the same survey across several countries at broadly similar points in time. To date, seven waves of the survey have been carried out, covering 120 countries. The seventh wave (WVS7) contains specific questions related to immigration attitudes, and table 1 presents the summary statistics of some of these questions for the United States and other regions of the cross-sectional world. Socioeconomic status (variable X045–Q287 in WVS’s questionnaire: file:///Users/mtani/Downloads/F00010562-World_Values_Survey_Wave_7_2017-2020_United_States_v3.0.pdf) is self-assessed and it groups respondents into five categories: upper class, upper-middle class, lower-middle class, working class, and lower class. Education (variable X025A_01–Q275) is measured in eight categories, as per the International Standard Classification of Educational Degrees (ISCED), ranging from early child education only (ISCED 0) to PhD (ISCED 8). Immigration (variable G027A–Q263) separates those born in the host country from those who self-define as immigrants. With reference to the hypotheses tested, WVS reports whether immigrants fill useful jobs (variable G053–Q122), raise the crime rate (G055–Q124), raise the risk of terrorism (G057–Q126), raise unemployment (G059–Q128), and raise social conflict (G060–Q129). Possible answers to each statement are “relevant,” “hard to say,” “not relevant,” “do not know,” and “not answered,” with the last two categories containing only a small proportion of replies (1 percent or less). Answers are reclassified in dichotomous responses equal to 1 (“relevant”) or 0 (“hard to say” and “not relevant”).

Statistically significant differences with US responses, obtained from a Welch’s t-test of mean differences, are indicated with *, **, or *** depending on whether the corresponding p-values of the tests are <.1, <.05, or <.01.

As demonstrated in table 1, WWS’s answers for the United States differ on average from the answers collected in the rest of the world (column 2 of the table) and when this is broken down by individual countries (Canada, Australia, and New Zealand in columns 3–5, respectively) or major regions (Africa, Europe, Asia-Pacific, and Central and Latin America in columns 6–9). US respondents are relatively older (by about one year), predominantly males (WWS respondents are mostly females), and middle class. They are among the most educated respondents, with an average corresponding to postsecondary non-tertiary education and tend to be married with a child. About 10 percent of the sample have immigrant origins, but about 15 percent have an immigrant lineage as either one or both parents were born abroad.

US respondents score the highest average on questions related to the usefulness of immigration but have mixed views about immigrants’ influence on crime rates, terrorism, unemployment, and social conflict in general, as their average is higher than in other high-immigration English-speaking countries (and lower than in remaining parts of the world): we focus on Canada, Australia, and New Zealand as they take part in the WVS.

Table 2 focuses on the key characteristics of US respondents, by immigrant status. Natives tend to be older, have fewer years of education, and are more likely to live in rural centers than immigrants. Natives’ views about immigration are less favorable than those of immigrants, especially with reference to immigrants’ possible influence on crime, unemployment, and terrorism. When data are further dissected by socioeconomic status, some of the natives’ concerns about immigration arise also in the wealthier immigrant group. In particular, immigrants that self-report as most successful (i.e., in the upper class of the self-assessed socioeconomic scale) have the strongest opinions against immigration: notwithstanding its economic contribution, they do not view immigrants as useful (average: .25 vs. natives’ .783) and are concerned that immigration brings other socioeconomic problems, such as terrorism (average: .625 vs. natives’ .521) and crime (.75 vs. .478).

Those results give some preliminary support to the hypotheses emerging from the model: notably, that natives statistically discriminate against immigrants and their benefits, but this does not apply (at least to the same extent) to immigrants already in the United States, for whom less successful immigrants and newcomers may represent a threat to the benefits they enjoy from natives’ current perception of the average immigrant.

H1–H4 are tested using only the seventh wave of WVS, which, for the United States, was collected in 2017 and focuses on opinions about immigration.

H1 suggests that immigrants who have been successful in the host country (those self-assessing in the top quintile of the socioeconomic scale) oppose immigration relative to those whose well-being is less at risk if natives have a worse view of immigration: namely, those who have not been successful in the host country and are lower quintiles of the socioeconomic (SOE) scale.

H2 suggests that the strongest opposition should arise from those who have been most successful, that is, those who self-assess to be in the top SOE quintile. It should be emphasized that SOE is used here as a proxy of immigrants’ relative success in the host country rather than their actual income in local currency. In fact, this variable in WVS is a self-assessment of respondents’ relative rank in the host society, structured in quintiles. As such, it is a more useful instrument to capture immigrants’ perceived “wellbeing at risk” from natives’ views of immigration, as formalized in the theoretical description, than other indicators such as income or education.

H3 is tested by studying the sign of the relationship between the SOE of immigrants in the United States and the income of human capital dispersion of their respective countries of origin. H3 suggests that immigrants of countries with the greatest income disparity or human capital dispersion should hold the strongest views against immigration because they know best, while natives know least, the immigrant types originating from these countries.

Finally, H4 is tested by comparing the dispersion of the education acquired by natives and immigrants in the United States, which underpins natives’ statistical discrimination toward immigrants and existing immigrants’ fears of a worsening view of immigration among natives, with that of three English-speaking countries that instead formally screen immigrants as part of the immigration process: namely, Canada, Australia, and New Zealand.

Empirical Specification and Results

H1 and H2 are tested from the same linear probability model, which applies Ordinary Least Squares (OLS) to the functional form:

(1)   \begin{equation*} y_i=\alpha+X_i \beta+\gamma I_i+\delta S O E_i+\theta I_i \operatorname{SOE}_i+\varepsilon_i \end{equation*}

where y is the for/against attitude of respondent i toward immigration (reference: for), X is a vector of demographic, employment, and characteristics such as age, education, labor force status, marital status, state of residence. . . . I is a dichotomous variable capturing the immigration status of the respondent (immigrant/native), SOE is a self-reported measure the respondent’s socioeconomic status (upper, high, average, below-average, and low income), which proxies for immigrants’ exposure to natives’ views about immigration, and \varepsilon_i is an error term. The coefficients of interest are \gamma, the link between the immigration status of the respondent and outcome variable y relative to a comparable native, and \theta, the coefficients of the interaction between immigrant and socioeconomic status (ref: low socioeconomic status).

Since the United States attracts immigrants from both ends of the skill distribution (top: H-1 visas; bottom, more common: family reunification and undocumented immigrants), H1 advances that existing immigrants of higher SOE will oppose additional immigration to the United States believing that new unqualified foreign labor may worsen US natives’ perception of all immigrants in the country.

We use five outcome binary indicators: namely, whether respondents believe that immigrants (i) fill useful jobs in the United States, (ii) raise crime, (iii) increase the risk of a terrorist attack, (iv) increase the unemployment rate, and (v) lead to social conflicts.

Table 3 presents the baseline results across the five indicators. Prior to discussing the variables of interest, it is worth noting that women have a more benign view of immigration than men (negative and statistically significant coefficients), as do highly educated respondents. In contrast, opposition to immigration is higher in rural centers and towns with smaller population size. Singles show less opposition to immigration while those with family, and especially with a large number of children, tend to oppose new immigrant arrivals―perhaps for fear of competition for jobs and housing for them and especially for their children (the indicator for singles is statistically no different from zero).

With respect to the variables of interest, the coefficients for the immigrant dummy variable are not statistically different from zero, implying that natives and immigrants who perceive themselves to be middle class share similar views about the effects of immigration, on average, when controlling for other observable characteristics. The indicators of socioeconomic status also indicate relatively few differences across social groups. Respondents in the higher socioeconomic status view immigrants as contributors to raising the crime rate, though the coefficient is statistically significant at only 10 percent. Respondents in the lower socioeconomic groups are less likely to suggest that immigrants carry out useful jobs and raise social conflicts in the United States.

However, when immigrant and self-assessed relative success are joined, very strong results in support of H1 emerge. The strongest opposition arises in the case of immigrants in the highest socioeconomic status, in line with H2. For those respondents, immigrants positively contribute to increased crime, the likelihood of a terrorist attack, and the unemployment rate. The coefficients are all statistically significant at the 1 percent level and their magnitude is very large. In fact, the coefficient \theta is the largest among all explanatory factors, suggesting that such views are strongly and uniformly shared among immigrants in the highest socioeconomic group. Interestingly, there is no such view on questions about the economic effect of immigration or the contribution to social unrest in the United States, revealing that the opposition of immigrants to immigration is limited to specific areas: those that perhaps most directly affect the perception of immigrants held by US natives: crime, terrorism, unemployment.

To verify whether US immigrants who self-assess as upper class in the United States originate from countries where the quality of immigrant types is highly diverse (i.e., high dispersion of income or human capital), we first present descriptive results using two distinct indicators of such likely heterogeneity in table 4a: (i) the dispersion of economic opportunity, as measured by the share of income accruing to the poorest 10 percent of the population of their respective countries of origin and (ii) the human capital of the population of their countries of origin, as measured by the Human Capital Index. For both indicators, higher values imply greater harm if immigrants experience statistical discrimination in the United States. Both indicators (i) and (ii) are sourced from the World Bank’s World Development Indicators database.

Tables 4a and 4b suggest the existence of a positive relationship between immigrants’ perception of a being member of the upper class in the United States and heterogeneity in immigrant types in their respective countries of origin (socioeconomic group and inequality in income and human capital). In Table 4a, US-based immigrant respondents in the top SOE quintile originate from countries with the highest heterogeneity in immigrant types (lowest average share of income going to the poorest 10 percent of the population or the lowest human capital index).

Table 4b presents more formal evidence obtained by regressing each of the five immigration-related questions of the WVS on the socioeconomic status of immigrant respondents joined with the inequality indicator (i.e., heterogeneity in immigrant types). The regression follows the statistical model:

(2)   \begin{equation*} y_i=b_0+X_i b_1+b_2 I N E Q_i+b_3 S O E E_i+b_4 I N E Q_i * S O E_i+e_i \end{equation*}

where y_i and SOE are defined as in (1) and INEQ is the share of income going to the poorest 10 percent of the population. The resulting coefficient of the interaction term is positive and statistically significant only for the highest SOE group. A similar result arises when INEQ is replaced by the human capital index.

The results from tables 4a and 4b support the hypothesis that stronger opposition to immigration arises in the case of immigrants from countries of origin characterized by inequality in income and human capital development―that is, countries where heterogeneity in possible immigrant types is high in the eyes of native residents and hence where statistical discrimination is likely to be stronger and where immigrants from the same countries are most susceptible to natives’ views of immigration.

To assess whether the results obtained are specific to the case of the United States, and to test H4, we run model (1) on various countries that have similar or more controls on the range of immigrant types that settle in. In particular, we compare the coefficients of the key interaction variables obtained for the United States with those of the three other English-speaking immigration-screening countries as well as other regional groups.

The results are reported in table 5. For simplicity, only the results obtained on the highest and lowest socioeconomic group are reported, as the other categories fall in between these two extreme cases. The rows show the countries, while the columns report the coefficients of interest: namely, immigrant status (\gamma, socioeconomic status (\delta, and their interaction (\theta. As shown, \theta, the coefficient identifying the responses of immigrants in the upper class (which we interpret as the category that is most sensitive to natives’ views about immigration), is positive and statistically significantly different from zero only in the case of the United States. In fact, in no other English-speaking country do immigrants of higher SOE oppose immigration to the same extent as immigrants living in the United States. A similar result applies when the regression is performed on other regional country groups.

H4 interprets these results as a consequence of the high skill dispersion of immigrants resettling into the United States relative to the three other English-speaking countries covered by the WVS that apply more selective immigration policies. As highlighted by the theoretical model, the lack of institutional screening of immigrants entering the United States, and the presence of undocumented immigration, creates incentives for immigrants who identify as most sensitive to natives’ view of immigration (those in the top SOE quintile) to signal their quality and differences (to US natives) relative to other immigrants from the same countries of origin through strong opposition to immigration especially from those places of origin. Supporting evidence for this interpretation is reported in table 6, which shows the average dispersion of education completed among natives and immigrants by main country of destination. In particular, it shows that such dispersion is wider among immigrants than natives in the United States―a possible incentive for immigrants of higher education (and SOE) to signal their quality to US natives relative to less-educated immigrants from the same country of origin.


As these baseline results are obtained on cross-sectional data, they may be affected by omitted variable bias. To test for this possibility, we apply the Oster (2019) method to test the stability of the coefficient when control variables are progressively added in a regression under the assumptions that (i) the relationship between treatment and unobservables can be recovered from that between treatment and observables and that (ii) the hypothetical model that includes treatment, observables, and unobserved variables produces a R_{\max } that can be less than unity―for example, because of measurement error. We apply Oster’s model to calculate the ratio of unobserved/observed selection (“delta”) required to nullify the statistical significance of the coefficient \theta, the estimate of the interaction between immigrant and socioeconomic status, reported in table 3. Robustness to omitted variable bias occur if delta is greater than 1 (the benchmark).

The results of this calculation, reported in table 7, indicate robustness to omitted variable bias, as the values of delta in the case of \mathrm{R}_{\max }=1.3 \times \mathrm{R}^2, the benchmark suggested by Oster, range from 7.14 (do immigrants raise crime?) to 15.21 (do immigrants increase unemployment?). These values imply that selection on unobservables should be 7.14 and 15.21 times the selection of observables, respectively―a very unlikely scenario (Oster suggests a benchmark ratio of 1). As a result, it is unlikely that the results presented suffer from severe omitted variable bias. Overall, the results of the robustness tests make the point estimates reported in table 3 credible.


To check if the results obtained for the United States apply to other high immigration countries, the model is run on pooled data restricted to responding immigrants only. The results, displayed in table 8a, focus on the five questions of interests and report the coefficient of the interaction term between the highest SOE and the country where WVS data were collected. The model applied is analogous to that specified in equation (1) where the immigrant status is replaced by a host country indicator, using Europe as a reference. The regression is performed on observations restricted to immigrants. The coefficients clearly point to a statistically significant opposition of immigrants toward immigration only in the case of the United States. For every other case, namely Canada, Australia, New Zealand, as well as regions (Africa, Americas, and Asia-Pacific), immigrants most exposed to natives’ views of immigration (highest SOE quintile) appear to be supportive or at least indifferent to immigration, as the coefficients are either negative and statistically significant or statistically equivalent to zero.

To verify whether the US case is symptomatic of a country-specific effect, we carry out a complementary regression identical to that presented in table 8a with the exception of restricting the observations to native respondents only. The results are summarized in table 8b and reveal that US natives do not have different views toward immigrants relative to most other countries and regions of the world. The coefficients are mostly statistically no different from zero. The one exception is represented by Canada, whose natives display a strong opposition to immigration: most coefficients to questions about immigrants’ contribution to unemployment, crime, and terrorism are positive and statistically significant and point to strong negative views and perceptions about immigration.

Immigrants in the United States therefore emerge as a unique group opposing additional immigration. While the hypotheses developed by the theoretical model are tested by means of correlational analysis, the estimates are consistent with the view that immigrants’ opposition to further immigration stems from natives’ inability to discern legal and undocumented immigrants’ quality and contribution to the host country’s economy and social compact.

These results are consistent with the hypothesis that, when faced with statistical discrimination, immigrants respond by acting as screens for new immigration via their support/opposition to it. This is especially the case when the host country has no institutionalized form of population management or large inflows of undocumented immigrants as is the case in the United States.

The results raise important questions about the role of immigration policy in enhancing immigrants’ integration in the host country, including their acceptance and support from existing immigrants. They support efforts to control undocumented immigration as a way to ensure orderly population management and gain support, especially among natives. The results also support the use of screening tools to limit the dispersion of skills distribution supplied by immigrants or at least their tighter control to prevent the possible rise of statistical discrimination toward immigrants (operated by natives) and anti-immigration stances among existing immigrants themselves.


This paper fills an important role in the immigration literature by studying the role of statistical discrimination as a cause for (some) immigrants’ opposition to immigration in cases where immigrants are not screened or there is a large amount of undocumented immigration. The results support this hypothesis for the case of the United States, which stands out among other immigration countries. In the United States, immigrants who are most exposed to natives’ views about immigration (those in the highest socioeconomic quintile) strongly oppose immigration, as shown during the most recent presidential elections, against the expectations of immigration observers. Rather than an emotional response or a preference, we argue that such a behavioral response is consistent with a rational utility-maximizing decision when newcomers may compromise existing immigrants’ perception among natives in the host society.


Aigner, D., and G. Cain. 1977. “Statistical Theories of Discrimination in Labor Markets.” Industrial and Labor Relations Review 30, no. 2: 175–87.

Berry, J. A., C. Cepuran, and S. Garcia-Rios. 2022. “Relative Group Discrimination and Vote Choice among Blacks, Latinos, Asians, and Whites.” Politics, Groups, and Identities 10, no. 3: 410–29.

Bianchi, M., P. Buonanno, P. Pinotti. 2012. “Do Immigrants Cause Crime?” Journal of the European Economic Association 10: 1318–47.

Bodvarsson, Ö. B., H. Van den Berg. 2013. The Economics of Immigration. New York: Springer.

Cadava, G. 2021. “The Deep Origins of Latino Support for Trump.” The New Yorker, December 29.

Card, D. 2005. “Is the New Immigration Really So Bad?” Economic Journal 115, no. 507): F300–F23.

Card, D., C. Dustmann, and I. Preston. 2005. “Understanding Attitudes to Immigration: The Migration and Minority Module of the first European Social Survey.” CReAM Discussion Paper Series 03(05), 1–43.

Choi, S.-W. 2021. “Immigration Policy and Terrorism: An Empirical Analysis.” Defence and Peace Economics 32: 271–95.

Corral, Á. J., and D. L. Leal. 2020. “Latinos for Trump? Latinos and the 2016 Presidential Election.” Social Science Quarterly 101, no. 3: 1115–31.

Dustmann, C., F. Fabbri, and I. Preston. 2005. “The Impact of Immigration on the British Labour Market.” Economic Journal 115, no. 507: F324–F41.

Dustmann, C., A. Glitz, and F. Frattini. 2008. “The Labour Market Impact of Immigration.” Oxford Review of Economic Policy 24, no. 3: 478–95.

Dustmann, C., and I. P. Preston. 2007. “Racial and Economic Factors in Attitudes to Immigration.” B. E. Journal of Economic Analysis and Policy 7, no. 1.

Efron, R. 1969. “What Is Perception?” In Proceedings of the Boston Colloquium for the Philosophy of Science 1966/1968, 137–73. New York: Springer.

Facchini, G., and A. M. Mayda. 2012. “Individual Attitudes towards Skilled Migration: An Empirical Analysis Across Countries.” World Economics 35: 183–96.

Friedberg, R. M., and J. Hunt. 2018. “The Impact of Immigrants on Host Country Wages, Employment and Growth.” In Interdisciplinary Perspectives on the New Immigration, edited by M. M. Suárez-Orozco, C. Suárez-Orozco, and D. Qin-Hilliard, 89–110. New York: Routledge.

Garcia, E. 2021. “Trump, My Dad and the Rightward Shift of Latino Men.” The Washington Post, March 22.

Garcia-Rios, S., F. Pedraza, and B. Wilcox-Archuleta. 2019. “Direct and Indirect Xenophobic Attacks: Unpacking Portfolios of Identity.” Political Behavior 41, no. 3: 633–56.

Gaston, N., and D. Nelson. 2000. “Immigration and Labour-Market Outcomes in the United States: A Political-Economy Puzzle.” Oxford Review of Economic Policy 16 no. 3: 104–14.

Gil de Zúñiga, H., N. Jung, and S. Valenzuela. 2012. “Social Media Use for News and Individuals’ Social Capital, Civic Engagement and Political Participation.” Journal of Computational Communication 17: 319–36.

Hainmueller, J., and D. J. Hopkins. 2014. “Public Attitudes toward Immigration.” Annual Review of Political Science 17: 225–49.

Hassner, R. E., and J. Wittenberg. 2015. “Barriers to Entry: Who Builds Fortified Boundaries and Why?” International Security 40: 157–90.

Helbling, M., and D. Meierrieks. 2020. “Terrorism and Migration: An Overview.” British Journal of Political Science 52, no. 2: 1–20.

Javdani, M. 2020. “Public Attitudes toward Immigration—Determinants and Unknowns.” IZA World of Labor 473, doi: 10.15185/izawol.473.

Just, A., and C. J. Anderson. 2015. “Dual Allegiances? Immigrants’ Attitudes toward Immigration.” Journal of Politics 77: 188–201.

Levinson, J. D., H. Cai, and D. Young. 2010. “Guilty by Implicit Racial Bias: The Guilty/Not Guilty Implicit Association Test.” Ohio State Journal of Criminal Law 8: 187.

Mayda, A. M. 2006. “Who Is against Immigration? A Cross-Country Investigation of Individual Attitudes toward Immigrants.” Review of Economics and Statistics 88: 510–30.

McKeever, B.W., D. Riffe, and F. D. Carpentier. 2012. “Perceived Hostile Media Bias, Presumed Media Influence, and Opinions about Immigrants and Immigration.” Southern Communication Journal 77: 420–37.

Mears, D. P. 2001. “Immigration and Crime: What’s the Connection.” Federal Sentencing Reporter 14: 284.

Newman, B. J., and N. Malhotra. 2019. “Economic Reasoning with a Racial Hue: Is the Immigration Consensus Purely Race Neutral?” Journal of Politics 81, no. 1: 153–66.

Niyimbanira, F., and T. S. Madzivhandila. 2016. “Myth That Immigrants ‘steal jobs’: An Economic Blame Game?” International Journal of Social Sciences and Humanities Studies 8: 165–79.

Ousey, G. C., and C. E. Kubrin. 2018. “Immigration and Crime: Assessing a Contentious Issue.” Annual Review of Criminology 1: 63–84.

O’Rourke, K. H., and R. Sinnott. 2006. “The Determinants of Individual Attitudes towards Immigration.” European Journal of Political Economy 22: 838–61.

Ostfeld, M. C. 2019. “The New White Flight?: The Effects of Political Appeals to Latinos on White Democrats.” Political Behavior 41, no. 3: 561–82.

Peri, G., and C. Sparber. 2009. “Task Specialization, Immigration, and Wages.” American Economic Journal: Applied Economics 1, no. 3: 135–69.

Peri, G., and C. Sparber. 2011. “Highly Educated Immigrants and Native Occupational Choice.” Industrial Relations: A Journal of Economy and Society 50, no. 3: 385–411.

Preston, I. 2014. “The Effect of Immigration on Public Finances.” Economic Journal 124, no. 580: F569–F92.

Rocha, R. R., B. R. Knoll, and R. D. Wrinkle. 2015. “Immigration Enforcement and the Redistribution of Political Trust.” Journal of Politics 77: 901–13.

Rowthorn, R. 2008. “The Fiscal Impact of Immigration on the Advanced Economies.” Oxford Review of Economic Policy 24, no. 3: 560–80.

Russonello, G., and F. Mazzei. 2021. “Trump’s Latino Support Was More Than Thought: Report Finds.” New York Times, April 2.

Rutland, A., L. Cameron, A. Milne, A., and P. McGeorge. 2005. “Social Sorms and Self‐Presentation: Children’s Implicit and Explicit Intergroup Attitudes.” Child Development 76: 451–66.

Sampson, R. J. 2008. “Rethinking Crime and Immigration.” Contexts 7: 28–33.

Sides, J., and J. Citrin. 2007. “European Opinion about Immigration: The Role of Identities, Interests and Information.” British Journal of Political Science 37, no. 3: 477–504.

Tani, M., and M. Piracha. 2022. “Migration and Education.” In Handbook of Labor, Human Resources and Population Economics, edited by K. F. Zimmermann, 256–306. New York: Springer.

Timmer, A.S., and J. G. Williams. 1998. “Immigration Policy Prior to the 1930s: Labor Markets, Policy Interactions, and Globalization Backlash.” Population and Development Review: 739–71.

Valentino, N. A., T. Brader, and A. E. Jardina. 2013. “Immigration Opposition among US Whites: General Ethnocentrism or Media Priming of Attitudes about Latinos?” Political Psychology 34: 149–66.

Watson, B. R., and D. Riffe. 2013. “Perceived Threat, Immigration Policy Support, and Media Coverage: Hostile Media and Presumed Influence.” International Journal of Public Opinion Research 25: 459–79.

Winknews. 2022. “GOP Senate Push Focuses on Latinos with ‘Vamos’ Initiative, Title 42.” April 6.

Winkler, E. N. 2009. “Children Are Not Colorblind: How Young Children Learn Race.” PACE Practical Approaches in Continuing Education 3: 1–8.

Wright, C., R. Brinklow-Vaughn, K. Johannes, and F. Rodriguez. 2021. “Media Portrayals of Immigration and Refugees in Hard and Fake News and Their Impact on Consumer Attitudes.” Howard Journal of Communication 32: 331–51.

Zimmermann, K. F., T. K. Bauer, and M. Lofstrom. 2000. “Immigration Policy, Assimilation of Immigrants and Natives’ Sentiments towards Immigrants: Evidence from 12 OECD-countries.” Swedish Economic Policy Review 7, no. 2:22–53.

CGO scholars and fellows frequently comment on a variety of topics for the popular press. The views expressed therein are those of the authors and do not necessarily reflect the views of the Center for Growth and Opportunity or the views of Utah State University.