How Do People Perceive Immigrants? Relating Perceptions to Numbers

Abstract

How does the general public perceive immigrants, whom do they think of when thinking about “immigrants,” and to what extent are these perceptions related to the actual composition of immigrant populations? We use three representative online surveys in the United States, South Africa, and Switzerland (total N = 2,778) to extend existing work on the perception of immigrants in terms of geographic coverage and individual characteristics. We also relate these responses to official statistics on immigration and integration patterns. In all three countries, there are significant discrepancies between perceptions of immigrants and their real proportion in the population and characteristics. Although we observe clear country differences, there is a striking pattern of people associating “immigrants” with “asylum seekers” in all three countries. We consider two possible explanations for the differences between perceptions and facts: the representativeness heuristic and the affect heuristic. In contrast to previous research, we find only partial support for the representativeness heuristic, whereas the results are consistent with the affect heuristic. We conclude that images of “immigrants” are largely shaped by pre-existing attitudes.

Introduction

Immigration remains one of the most frequently discussed topics in popular and political discourse, and much research examines attitudes to immigrants, how immigrants are politicized and appear in the news, or citizens’ preferences about policies regulating immigrants, to name just some active research agendas (Dražanová 2022; Green-Pedersen and Otjes 2017; Van der Brug et al. 2015). Reflecting the importance of immigration as a topic, many large social surveys measure relevant attitudes, notable examples being the European Social Survey or the International Social Survey Programme (ISSP). Frequently, surveys ask respondents about their views on “immigrants” in general, without defining the term “immigrant” more precisely (Crawley 2005; Ruedin 2020). For instance, the 2003 wave of the ISSP included the following question: “Do you think the number of immigrants to [country] nowadays should be [increased a lot/increased a little/…]?” Similar formulations have been used in other surveys (Citrin and Sides 2008).

While such generic formulations can have advantages for cross-national research, they are based on the assumptions that respondents have a uniform perception of who an “immigrant” is and that respondents have this perception readily available at the time they respond. Any systematic deviation, for example, due to different levels of knowledge or different stereotypes about the immigrant population among some respondents, would confound measurements and affect results (see Zaller and Feldman 1992 on the problem of “availability bias” in survey research in general). We do not find the assumption that people have a uniform conception of immigrants plausible, not even when restricted to respondents within a single country. First, the term “immigrant” covers a wide array of different categories of internationally mobile persons including labor migrants, asylum seekers, refugees, international students, and diplomats (to name a few relevant categories), and this complexity increases the chances that people think about different groups when relating their conceptions of who an “immigrant” is.

Moreover, we also know that perceptions of immigrants are socially constructed in the sense that “immigrants” do not exist independently of our social world (Blinder 2015; Flores and Schachter 2018). Indeed, perceptions of the number and attributes of immigrants are strongly biased (Citrin and Sides 2008; Herda 2010). Research on whether people can correctly estimate the proportion of immigrants in a specific country typically finds that the indigenous population overestimates the size of the immigrant population, but also that the extent of overestimation varies by the amount of contact with immigrants, exposure to relevant information from television or newspapers, and the perceived level of threat posed by immigrants (e.g., Citrin and Sides 2008; Herda 2010, 2019; Sides and Citrin 2007). Other research focuses on people’s perceptions of specific attributes of immigrants, such as their country of origin, race, religious affiliation, level of education, or work attitudes, and likewise reports widespread differences in understanding immigration and indeed misperceptions of immigrant numbers (Alesina, Miano, and Stantcheva 2023; Blinder 2015; Herda 2015; Peberdy 2001). Alesina et al. (2023) found that people perceive immigrants as less educated, more often unemployed, and more likely to be Muslim than they actually are. Blinder (2015) found that people in the United Kingdom vastly overestimate the degree to which immigrants are asylum seekers, while drastically underestimating the degree to which they are international students.

Looking at the relationship between immigration and support for the welfare state, Negash (2024) provided a different angle, examining how individual characteristics and variations in the welfare institutions available in a place contribute to different perceptions on “immigration.” For example, individuals who feel economically insecure are more likely to oppose immigration (see also Kayran 2024; Pecoraro and Ruedin 2016) and to think that immigrants are a burden on the welfare state (Naumann, Stoetzer, and Pietrantuono 2018; Negash 2024). At the contextual level, Negash (2024) highlighted that places with generous benefits are associated with the view that immigration has a negative fiscal impact. Such differences in perception can be problematic when policy decisions are indirectly based on public opinion surveys, such as when migration policies are made more restrictive or when religious symbols are banned to force immigrant integration in response to what is seen as public demand. In addition, radical and populist right-wing actors may mobilize on the basis of such misperceptions instead of facts, therefore leading to an undermining of liberal democratic norms (Lutz and Bitschnau 2023).

Here we seek to describe public perceptions of “immigrants” in different contexts and ask to what extent these perceptions are related to the actual composition of immigrant populations. The present study contributes to the literature on perceptions of immigrant in two ways. First, it makes a descriptive contribution by mapping perceptions of immigration and how they vary across countries, using survey data from three diverse countries (the United States, South Africa, and Switzerland). We follow earlier work by Blinder (2015) on public perceptions of immigrants in the United Kingdom and expand it by including a larger variety of immigrant groups, by covering the same questions in three countries, and by presenting a descriptive analysis on which social groups tend to have which images of immigrants. The three countries we cover are major immigrant destination countries on different continents, and studying perceptions in this diverse sample allows us to elicit general patterns of perception as well as relevant differences. Resembling earlier findings, we find large discrepancies between public perceptions of immigrants and their factual composition in all three countries.

Second, when it comes to perceptions that can be compared to numbers, we test two explanations for deviations between perceptions and facts, building on psychological theories about heuristics (Bordalo et al. 2016; Kahneman and Tversky 1972; Tversky and Kahneman 1973). We focus on the affect heuristic (which predicts that perceptions of factual attributes of immigrants are influenced by people’s attitudes to them) and the representativeness heuristic (which predicts that people associate immigrants with attributes that are relatively more common in the immigrant population). We discuss how the results relate to both heuristics and why we find support for the affect heuristic but only mixed support for the representativeness heuristic. To conclude, we propose potential avenues for future work on the sources of distortions in the perceptions of immigrants.

Why Perceptions of Immigrants and Migration Vary

While immigration may be a recurring issue in the public consciousness, perceptions of who an “immigrant” is and what they do are constructed and have varied considerably throughout history. For example, while contemporary migration is largely voluntary and consists of people seeking a better future abroad (e.g., Ferrie and Hatton 2015), past waves of migration like those to North America or Australia in the 18^th and 19^th centuries comprised much larger shares of forced migrants (slaves, indentured servants, and convicts). A related example of the changing nature of migration is the fact that in Europe, passports were historically used to constrain the internal mobility of the native population and to limit emigration rather than controlling the entry of foreigners (Torpey 1997). The difficulty of capturing who immigrants are can also be illustrated by the legal distinction between labor migrants and asylum seekers, which is in practice blurred to such an extent that it has been called a “legal fiction” (Hamlin 2021, 3). Indeed, legal and public vocabulary does not align (Saggar and Drean 2001), as can be seen by the legal distinction between asylum seekers and refugees, two groups that are often conflated in the public mind (Blinder 2015).

Research also highlights that forming accurate perceptions of immigrants is not a simple task. As in other areas of political opinion, reasoning about the size and characteristics of the immigrant population requires both information and the capacity to absorb, manage, and transform this information into an educated opinion (Bordalo et al. 2016; Herda 2010, 2015). People generally lack both accurate information about immigration and immigrants, and the time, resources, or inclination to engage in thorough reasoning about the information they have (Bartels 1996; Lupia 1994; Zaller 1992). Negash (2024, 685) provided a framework to integrate different factors that influence perceptions, emphasizing both individual and contextual variables. While the availability of information in the media and the political climate (e.g., Carvalho, Duarte, and Ruedin 2024; Green, Visintin, and Sarrasin 2018), the role of risks such as unemployment (e.g., Kayran 2024), and the composition of the migrant population (e.g., Czymara 2020) have been examined before, Negash (2024) also highlighted how the policy context (e.g., welfare generosity) can play a role. Here, we are less interested in understanding these explanations for (mis)perceptions than in understanding the characteristics of perceptions.

Limited efforts to obtain accurate information do not necessarily mean that the views and opinions of the population are inaccurate: people use heuristics as cognitive shortcuts to develop opinions about all kinds of political issues, including immigration (Fiske and Taylor 1991, 381–94). Heuristics allow individuals to draw conclusions based on limited informational input and limited mental processing. Indeed, the use of heuristics is common in many decision-making situations – especially when information is difficult to obtain or process (Bodenhausen 1990; Kahneman and Tversky 1972; Macrae, Milne, and Bodenhausen 1994; see also Petersen 2012; Petersen et al. 2011, 2012; Tversky and Kahneman 1973; Zaller and Feldman 1992). People may make use of brands, political parties, or recommendations by trusted individuals (e.g., Lupia 1994), to name just a few possibilities. However, because shortcuts are simplifications, there is always a risk that inferences based on heuristics are biased.

Research has identified a range of heuristics that shape how views and opinions about people and objects are formed and can explain perceptions and misperceptions in cases where population numbers are involved (Bordalo et al. 2016). These heuristics are nonexclusive, meaning that people can rely on multiple heuristics simultaneously. The heuristics produce testable predictions of the shape of perceptions about immigrants and therefore provide potential explanations for misperceptions of immigrants and their variation between countries. In the present study, we focus on two heuristics: the representativeness heuristic and the affect heuristic, because they reflect factors identified in earlier work on perceptions (e.g., Negash 2024), and because they are testable with our data.

Representativeness Heuristic

The representativeness heuristic is the first heuristic we examine as a mechanism people use to form judgments about social groups. The core idea is that people associate a given group with attributes that are relatively representative of that group (Bordalo et al. 2016). In other words, people associate those attributes with the group which are relatively more frequent in the group than in a relevant comparison group. For example, if the share of 18- to 29-year-olds is larger in the immigrant population than in the majority population (say 20% among the immigrants and 15% among the majority), the attribute “young” is relatively representative of the immigrant population – which would then lead people to focus on this attribute and associate immigrants with being young or younger than the majority population.

Importantly, while such perceptions may originally be based on fact – real compositional differences between groups – they tend to become exaggerated and thus lead to distorted perceptions. To stay with the previous example, the representativeness heuristic could lead people to see the immigrant population as overwhelmingly or even entirely comprised of young persons – even if the young are in fact a minority among immigrants, though larger than the proportion of young persons in the majority population.

Representativeness heuristic: The representativeness heuristic leads people to associate immigrants with attributes that are relatively more common among immigrants than among the majority population. Knowing the demographic composition of the immigrant and general populations in different countries, we expect that attributes that are relatively more common among immigrants inform perceptions of immigrants.

As outlined in further detail below, we test this generic expectation with four specific indicators: (1) We expect that in countries where the share of 0 to 24-years-old among immigrants is higher than among the majority population, people are more likely to perceive immigrants as young. (2) We expect that in countries where linguistic diversity is higher, the perception that immigrants speak with a foreign accent is less widespread. (3) We expect that in countries where the unemployment rate for immigrants is higher than that of the majority population, immigrants are more likely to be perceived as a burden to society. (4) We expect that in countries where the employment rate for immigrants is higher than that of the majority population, immigrants are more likely to be perceived as paying taxes.

Affect Heuristic

When people form perceptions about immigrants, they are also likely to rely on the affect heuristic (Herda 2015), the second heuristic we examine. Related to motivated reasoning (Taber and Lodge 2006), the affect heuristic is used when people base their perceptions of the specific attributes of a group on whether they generally have positive or negative views of that group. In the current case, people with generally negative views of immigrants will be more likely to associate immigrants with specific negative attributes (e.g., unproductive, threat to internal security), as this leads to an internally coherent perception where their general negative attitudes to immigration are justified and cognitive dissonance can be avoided.

Affect heuristic: The affect heuristic leads people to associate immigrants with attributes conforming to their existing views. We, therefore, expect a positive correlation between general anti-immigrant attitudes and associations of immigrants with negative specific attributes. This implies, for example, that individuals with general anti-immigrant attitudes tend to associate immigrants with negative behaviors (e.g., being unemployed).

Data and Methods

To investigate people’s perceptions of immigrants and test our hypotheses, we drew on data from three online surveys administered on Qualtrics. As is commonly done in studies with online panels, we used quotas in respondent recruitment to approximate a representative sample of the general population in the United States, South Africa, and Switzerland. Findings based on such online samples have been shown to generalize to the underlying population (Berinsky, Huber, and Lenz 2012; Coppock, Leeper, and Mullinix 2018; Weinberg, Freese, and McElhattan 2014). The survey in the United States was conducted between 16 November and 12 December 2017 and was available in English. Quotas were set for gender, age, state, and race/ethnic group (N = 1,150). The survey in South Africa was conducted between 25 July and 17 August 2017 and was available in English – consultation with survey partners suggested that translating the questionnaire into additional local languages was unnecessary. Quotas were set for gender, age, education, province, and population group/race (N = 1,260). The survey in Switzerland was conducted between December 5 and December 12 2016 and was available in German and French. For reasons of research economy, the Italian-speaking part of the country was not covered (4.1% of the population). Quotas were set for gender, age, and language region (French, German; N = 368). Participants were remunerated in cash or through vouchers.

The three countries are major destination countries for immigrants, lie on three different continents, and have different immigrant populations and welfare institutions (compare Negash 2024). What unites these countries is that all of them are relatively prosperous in the context of their respective regions and attract labor migrants, including highly skilled immigrants. By contrast, the share of asylum seekers as part of the immigrant population varies: while asylum seekers constitute a small part of the immigrant population in the United States and Switzerland, they are a relatively large part of the immigrant population in South Africa (Amit 2015; Ruedin 2019). Looking at the media environment, in the United States and South Africa, irregular migration is highly politicized, but this is not the case in Switzerland. At the institutional level, workers in the three countries are exposed to competition to a different degree. Taking the OECD mean as a reference, workers in the United States are less protected and more exposed, while workers in Switzerland are more protected and less exposed (OECD 2024). South Africa has a large informal sector and incomplete protection (Fall and Steenkamp 2020). The inclusion of South Africa also allows us to test the scope of theories developed in the Global North in a context of the Global South (compare Smith 2021).

The surveys used two kinds of questions to ask about people’s perceptions of immigrants (Figure 1). The first block of questions was adapted without changes from Blinder (2015, 7): “When you think about immigrants coming to and living in [country], which of these groups do you normally think about?” For each specific question, respondents could tick multiple answers; an (exclusive) option of “none of these” was also provided. The first block of questions covered aspects of citizenship, motives, and country characteristics. The tables in the appendix outline the percentage of the population who picked a specific answer (online Appendix 1).

Figure 1. Question wording and design of the outcome variables, taken from Blinder (2015; left panel) or adapted using sliders (right panel).

The second block of questions follows the same principles but approaches perceptions differently to capture a wider set of attributes: “Here is a list of people. For each person, do you think this person is an immigrant?” Here respondents were provided with a slider from 0 (“clearly not an immigrant”) to 4 (“clearly an immigrant”). The slider scale was marked with the numbers 0, 1, 2, 3, and 4 and allowed for a precision of .1. These questions covered aspects of nationality and skin color, specific jobs, and personal characteristics. The tables in the appendix provide the mean and standard deviation (online Appendix 1). In the United States and South Africa, we randomly varied the “immigrants” part in both blocks of questions, asking about “immigrants”, “refugees”, or “expats” (survey experiment).

Analytical Strategy

Because we fielded new questions about people’s perceptions of the immigrant population that had not been asked in that way before (adding items to the first block of questions and adding a second block of adapted questions), the first part of our analysis will provide a description of the patterns we found in the three countries, going over the available data on immigrant perceptions. Tables are presented in the main text that summarize the data; further data tables are included in online Appendix 1. The data and replication code are available on OSF (https://osf.io/jp7u5/).

The second part of the analysis is concerned with testing the relevance of the two heuristics for (mis-)perceptions about the immigrant population. Because our data are observational, our general approach is to use multivariate regression to estimate the effects of relevant predictors of affect (to test the affect heuristic) and objective immigrant characteristics (to test the representativeness heuristic) on people’s perceptions of immigrants in the three countries, while controlling for a set of relevant variables (see following sections).

To analyze the role of the representativeness heuristic, we test whether people’s perceptions of immigrants correlate with known features of the immigrant population – specifically, whether people associate immigrants with attributes that are relatively more widespread in the immigrant population than in the general population. Here, obtaining cross-national comparative data about the objective features of the immigrant populations that match our micro-level data is an important challenge. We only found relevant macro-level information for four micro-level indicators: being young, speaking with a foreign accent, paying taxes, and a proxy of being a burden on the welfare state. We therefore (and due to limited space) focus on these four indicators.

The United Nations provides data on the share of young persons (0–24 years of age) among immigrants and in the overall population in all three countries, which directly relates to the first attribute (being young). Overall, the share of young persons (0–24) is lower in the immigrant populations than in the overall populations (see also Figure A1 in the online Appendix), reflecting the fact that many immigrants are working-age adults (i.e., in their mid-20s and older) rather than children or teenagers. ¹ Still, the difference in the share of young persons between immigrants and the overall population is somewhat less pronounced in South Africa than in the other two countries. This leads us to expect that South Africans will be less likely to perceive immigrants as young compared to Swiss or Americans.

Measuring the extent to which immigrants really are fiscal burdens or pay taxes is exceedingly difficult (e.g., Boeri 2010), and we use two macro-level indicators as proxies (that are also likely to be available to the general public): the unemployment rate (to measure the “burden” attribute) and the employment rate (to measure the “taxes” attribute; see d’Albis, Boubtane, and Coulibaly 2018 for the use of this indicator). We acknowledge that there are individual-level differences in perceptions on welfare (Negash 2024) and interpret the results for this proxy jointly with the other indicators when drawing conclusions. These data are available from the OECD for Switzerland and the United States (see Figure A1 in the online Appendix). Starting with the data on unemployment rates, we can see that the unemployment rate for immigrants is much higher than that of the overall population in Switzerland. By contrast, there is only a small difference between the unemployment rate of immigrants and the overall population in the United States. This leads us to expect that immigrants are strongly associated with unemployment – as a proxy of being seen as a “burden on the welfare state” – in Switzerland but not so in the United States. We can add, based on qualitative assessments, that the situation in South Africa is similar to that in the United States in that immigrants are less likely to be unemployed than the native-born (Cornwell and Inder 2004) and should therefore be less strongly associated with being seen as a “burden” than in Switzerland.

To measure the final attribute, speaking with a foreign accent, we rely on a qualitative assessment. Switzerland is characterized by a high degree of linguistic diversity in general, but there is also important variation in the local accents and dialects spoken within the linguistic regions, such as the differences in the Swiss–German dialects within the German-speaking part of Switzerland (Steiner, Jeszenszky, and Leemann 2023). The situation is similar in South Africa, which has eleven officially recognized languages and, although English is the dominant language due to its association with the anti-apartheid struggle, there are still recognizable variations in spoken English between people of different classes or regions (e.g., Collins 2017; Trimbur 2009). The United States is arguably less diverse in this respect, although there is noticeable variation in accents across regions (e.g., the accents associated with Boston, New York, or the Southern States) and racial or ethnic groups (e.g., Labov 2012). Overall, we would therefore expect that speaking with a foreign accent is more of a distinguishing feature of immigrants – and therefore more strongly associated with immigrants – in the United States compared to Switzerland and South Africa.

To measure affect toward immigrants, we use a measure of nativist sentiment: the degree to which respondents agree that employers should prioritize hiring citizens over immigrants when jobs are scarce (measured on a Likert scale of 0 to 4, ranging from “strongly disagree” to “strongly agree”). A high score on this scale indicates a willingness to discriminate against immigrants in favor of citizens and therefore a stronger anti-immigrant attitude. Based on the affect heuristic, we would expect stronger anti-immigrant attitudes to correlate with perceptions of immigrants as more negative and distant. More specifically related to our four indicators (being a burden, being a taxpayer, having a foreign accent, and being young), we would expect that people who dislike immigrants more strongly see them as more negative (more likely to be a burden, less likely to be a taxpayer) and distant (having a foreign accent), but we would not expect an association with the fourth, neutral item (being young).

We tested the representativeness and affect heuristics empirically via a multivariate regression analysis where our four variables on people’s perceptions – immigrants as someone with a foreign accent, as someone who is a burden on the welfare system, as someone who pays taxes, and as someone who is young – are the outcome variables. To assess the representativeness heuristic – which predicts between-country differences in the extent to which immigrants are associated with certain attributes – we estimated pooled regression models with country-fixed effects and assessed the country intercepts (and derived predicted values). To test the affect heuristic, we included our measurement of nativism (“when jobs are scarce”) as a predictor in the regression models and assessed its coefficient. As control variables, we included factors that are thought to reflect cognitive resources, contextual factors, and motivated reasoning. In particular, we included demographic variables to account for potential composition effects (Alesina, Miano, and Stantcheva 2023; compare Negash 2024): age in groups, gender, employment status, household type, years of education, and being foreign born.

Perceptions of Immigrants and Their Variation Across Countries

Given that we present new data, we begin with a description of perceptions of “immigrants” in the three countries before we return to the two heuristics in the subsequent section. In the surveys, we used a series of questions to ask who respondents had in mind when thinking about immigrants. We have provided a list of countries, motives, nationality, country characteristics, and personal characteristics (full tables are included in online Appendix 1). Starting with questions on citizenship, we can see clear differences between the responses in South Africa, the US, and Switzerland. In South Africa, nearly 9 out of 10 respondents think about citizens from (other) African countries, with only 4 in 10 thinking about non-African countries. South African citizens born abroad, children of non-South Africans born in South Africa, and those naturalized were mentioned by around 1 in 10 respondents. By contrast, in the United States around 4 out of 10 respondents did not consider the categories provided fitting and picked “none of these” – a response that was virtually absent in South Africa and Switzerland. Between 2 and 3 out of 10 respondents think about naturalized US citizens, US citizens born abroad, or US-born children of non-US citizens. In Switzerland, more than 6 out of 10 respondents picked citizens from non-European countries, with only 3 out of 10 choosing citizens from European countries. Similar to South Africa, naturalized citizens, Swiss-born children of non-Swiss citizens, and Swiss citizens born abroad were chosen by only around 1 in 10 respondents. At the same time, within the countries, there was little difference whether we asked about “immigrants”, “refugees”, or “expats” (Figure 2; Tables A1, A7, A14 in online Appendix 1).

Figure 2. Selected perceptions of immigrants in South Africa, Switzerland, and the United States, 2017.

Notes: The thin line indicates 100% of respondents, the length of the thick line indicates the percentage of respondents who agree with a particular image. The first two items were not asked in the United States. Example: 83% of respondents in South Africa agreed that immigrants refer to someone from an African country. Multiple responses were allowed except for “none of these,” which was a single response. The numbers include all labels where they were varied; see online Appendix 1 for numbers and additional tables.

Regarding the motive of migration, we again observe little difference within countries between “immigrants”, “refugees”, and “expats”, but responses to different motives vary (Tables A2, A8, A15 in online Appendix 1). This could suggest that labels to describe “immigrants” (such as “expats”, “refugees”, and “migrants”) may be less influential than the general images individuals have in relation to people who have migrated, images that are often racialized (Scheel and Tazzioli 2022). The most common motive respondents think about, however, varies between countries. Work is chosen most often in the United States and South Africa, whereas in Switzerland, asylum is the dominant motive people think about when asked why immigrants have come and settled in their country. Compared to citizenship, there is less variance across different motives, and we observe some differences between “immigrants”, “refugees”, and “expats” regarding fleeing economic hardship (less for expats) and fleeing bad government (more for refugees). These differences are small and run counter to the supposedly clear distinction between immigrants and asylum seekers in immigration law (compare Hamlin 2021). In none of the three countries was the image of a person fleeing climate change common. Looking at the characteristics of the countries of origin, we did not find large differences between “immigrants” and “refugees”. In general, “expats” seemed more difficult to classify, and fewer characteristics were picked across the board, but the relative frequency is not fundamentally different from the other two categories. Images of poor countries, conflict, and dictatorships dominate (chosen by around 6 out of 10 respondents), while religious differences seem less salient (indicated by around 2 to 3 out of 10 respondents; Tables A3, A9, A16 in online Appendix 1). These images could readily be racialized (Scheel and Tazzioli 2022), though with the data at hand we cannot examine this directly.

In a separate question, we provided a list of potential “immigrants”, “expats”, or “refugees” and asked whether respondents thought that a person was an immigrant/expat/refugee. Again, we found a clear difference between the three countries concerning the images of immigrants, but surprisingly little variance across the terms (Tables A4, A11, A17 in online Appendix 1). For instance, in South Africa, the majority thought that a person with Somali nationality is an immigrant (mean 3.20 [1.04] on a four-point scale), with a minority choosing expat (2.60 [1.26]) or refugee (2.93 [1.13]). In general, the scores for “expats” are somewhat lower, but the ranking is largely the same. Compared to the question on citizenship, this question identifies more variance between the three terms (“immigrants”, “expats”, “refugees”), but in none of the cases can we identify the clear distinctions that immigration law (immigrants versus refugees) and social habitus (expats) would suggest.

Using a list of jobs and personal characteristics, we further assessed the perception of immigrants in the three countries. In all three countries, low-skilled jobs or unemployment were part of the more common images, but we did not observe strong associations: most of the values are below the midpoint, indicating that jobs are not a central part of the images people have of immigrants (Tables A5, A12, A18 in online Appendix 1). In this sense, it is not surprising to find hardly any differences between “immigrants”, “expats”, and “refugees”. This lack of clear differentiation between the three groups also applies to individual characteristics. The two most common characteristics were speaking with a foreign accent and being a burden on the welfare system (Tables A6, A13, A19 in online Appendix 1). In this regard, we observed no strong differences between the countries. Once again, most characteristics were rated below the midpoint, indicating no distinct perceptions.

Relevance of Heuristics for the Formation of Stereotypes

In this section, we examine to what extent different heuristics play a role in shaping perceptions of immigrants. Table 1 provides descriptive information about the extent to which perceptions of immigrants are associated with nativism (“when jobs are scarce, preference should be given to natives”). Overall, people higher in nativism are more likely to perceive immigrants as different and problematic. In particular, higher levels of nativism are associated with perceiving immigrants as having a foreign accent in South Africa and the United States, less likely to pay taxes (especially in Switzerland), more likely to be a fiscal burden (especially in Switzerland and the United States), and more likely to be young (especially in the United States).

Table 1. Correlations between perceptions of immigrants and nativism (prioritizing natives when jobs are scarce) in South Africa, Switzerland, and the United States, 2017

Notes: N = 430 in South Africa. N = 368 in Switzerland. N = 387 in the United States. Only answers to questions concerning “immigrants” were included. *p ≤ .05, **p ≤ .01, ***p ≤ .001.

In our regression analysis, we examine the association between different perceptions or images of immigrants and the two heuristics more closely. As mentioned above, we focus on between-country differences to assess the representativeness heuristic (see Panel (A) in Figure 3), and on the effects of nativist sentiment (“prioritize natives when jobs are scarce”; see Panel (B) in Figure 3) to assess the affect heuristic.

Figure 3. Results of regression analysis: Panel (A) shows predicted scores for each outcome variable and country to assess the representativeness heuristic. Panel (B) shows the estimated effect of nativism (“prioritize natives when jobs are scarce”) on each outcome variable. All results are based on pooled models including all controls. See online Appendix 4 and 5 for the detailed regression results and additional estimations by country.

Panel (A) in Figure 3 presents the between-country differences in immigrant perceptions in the form of predicted mean scores on each outcome variable per country based on pooled regression models including all control variables. These results provide only limited support for the relevance of the representativeness heuristic. On the one side, we expected that immigrants are seen as more of a burden on the welfare state in Switzerland compared to the United States and South Africa, since immigrants are more commonly unemployed in Switzerland compared to the other two countries ² – and that is indeed what we find. As is immediately visible, Swiss respondents associate immigrants more commonly with “being a burden” than in the other two countries, and the differences between Switzerland and the others are each statistically significant (Δ CH-SA = .374 [SE .090]; p < .01; Δ CH-US = .390 [.092]; p < .01).

This being said, we find no corresponding patterns when it comes to the perception of immigrants as taxpayers. Here, we expected that Swiss respondents would be less likely to have this perception compared to those from the United States and South Africa, but we find only a significant difference between the United States and South Africa (Δ US-SA = −.174 [.077]; p = .024). We also expected that immigrants should be less commonly seen as young in South Africa than in the other countries, and the predicted values correspond somewhat to this expectation – but only the difference between South Africa and Switzerland is statistically significant (Δ CH-SA = .187 [.078]; p = .022).

Finally, we find substantial and statistically significant differences between all three countries when it comes to perceptions of immigrants as speaking with foreign accents (Δ CH-SA = −.535 [.086], p < .01; Δ CH-US = −.221 [.088], p = .012; Δ SA-US = .314 [.085], p < .01). This perception is clearly most common in South Africa and least common in Switzerland, with the United States taking a middle position. This finding is not in line with our expectations, however, since we had originally expected that foreign accents should be more strongly associated with immigrants in the United States compared to the other two countries.

The case is clearer regarding the affect heuristic. Here, we predicted that people with stronger nativist sentiment would associate immigrants more with negative and distant attributes and less with positive ones, and this is what we find. More specifically, and as shown in Panel (B) of Figure 3, we find that stronger nativism is significantly and negatively linked to the perception of immigrants as taxpayers but positively to the perception of immigrants as burdens on the welfare state or as having foreign accents. Nativism has no significant effect on the neutral attribute of being young.

Discussion and Conclusion

The aim of this study was to assess people’s perceptions of immigrants in the United States, South Africa, and Switzerland. The guiding research question was: Which groups are perceived by the public as “immigrants,” and to what extent are perceptions of immigrants related to actual numbers? Despite covering three different national contexts on three continents, we find similarities: in all three countries, there are large discrepancies between perceptions of immigrants and the factual compositions of these immigrant groups in the population (see also Paquet and Lawlor 2022). These differences are not simply a reflection of the well-known tendency to overestimate the share of immigrants in the population – which reflects general human biases (Landy, Guay, and Marghetis 2018). Instead, survey respondents tend to associate immigrants strongly with asylum seekers, irrespective of the fact that asylum seekers represent only a small part of the immigrant population in two countries (the United States and Switzerland) but a larger one in South Africa (Amit 2015; see also Blinder 2015).

This study’s first contribution is a broad descriptive overview of immigrant perceptions across three national contexts in the Global North and the Global South. We not only covered different contexts, but in the United States and South Africa, we also varied labels to capture possible differences concerning perceptions of “immigrants”, “refugees,” and “expats”. While we found differences between countries in how “immigrants” are perceived (in line with research emphasizing the role of context in perceptions and attitudes, e.g., Green, Visintin, and Sarrasin 2018; Negash 2024), there were surprisingly few differences within countries between the labels (“immigrants”, “expats”, “refugees”). This suggests not only context-specific perceptions but also images that do not differentiate clearly or systematically between different kinds of immigrants – a stark counterpoint to immigration law that makes a clear distinction based on citizenship and the motive of migration through legal immigration statuses, supporting the claim that this distinction could be considered “legal fiction” (Hamlin 2021).

As a second contribution, we examined possible reasons for why perceptions vary between countries. We considered the representativeness heuristic and the affect heuristic as potential explanations for the difference between perceptions and reality (Bordalo et al. 2016). With this, we complement research showing differences in perceptions depending on the context (e.g., Green, Visintin, and Sarrasin 2018). The representativeness heuristic should lead people to associate immigrants with attributes that are relatively more common among immigrants than among the majority population. Based on the demographic composition of the immigrant population and the general population in different national contexts, we expected a positive correlation between relative importance in the population and perceptions of immigrants across national contexts. However, we found only limited evidence for the representativeness heuristic. There are some characteristics for which we find associations in line with the representativeness heuristic, for example, that immigrants are perceived as a burden in Switzerland but not in South Africa or in the United States. Mostly, however, our results do not conform with our expectations here. For example, there was no similar association with respect to immigrants being seen as taxpayers. Or, while demographic data would suggest that “being young” should be relatively more strongly attributed to being an immigrant in Switzerland than in the United States or South Africa, we found no such association in the regression analysis. Given the role of mass media in shaping public perceptions of immigrants (Blinder and Allen 2016; Bolet and Foos 2024; van Klingeren et al. 2015), future research should examine the link between individual perceptions and media coverage to complement studies considering variation in context (e.g., Green, Visintin, and Sarrasin 2018; Negash 2024).

The affect heuristic leads people to perceive immigrants in relation to their existing attitudes and feelings to immigrants. Thus, people with generally negative attitudes to immigrants associate immigrants more with negative and distant attributes (e.g., “burden”, “accent”) and less with positive ones (e.g., “taxpayer”) but have no similar tendency when it comes to neutral attributes (e.g., “young”). We found evidence for the affect heuristic across the three national contexts. Specifically, persons with stronger anti-immigrant sentiments see immigrants as more likely to be a burden on the welfare state, as less likely to pay taxes, and as more likely to have a foreign accent, but not as younger or older. Thus, the perceptions people have of immigrants relates strongly to their attitudes and feelings toward immigrants. This suggests that perceptions of “immigrants” are largely shaped by existing attitudes, and following previous research, such perceptions could potentially even reinforce pre-existing attitudes and reinforce racialized images of immigrants (De Coninck 2020; Flores and Azar 2022). However, with cross-sectional data, we cannot disentangle the causal paths between attitudes and perceptions.

Further research could explore alternative heuristics to understand differences in perception, such as the availability heuristic (Herda 2010, 2015; Tversky and Kahneman 1973), which is at work when people base their opinions or perceptions on the information currently available to them – what is “at the top of their head” (Schmidt-Catran and Czymara 2023; Zaller and Feldman 1992, 590). The kind of information, considerations, and facts available to people depends greatly on their environment. For example, where mass media reporting about immigration is focused on asylum seekers and other humanitarian migrants as opposed to higher-skilled “expat” labor migrants or exchange students (Blinder and Allen 2016), people associate immigrants more readily with asylum seekers simply because this association is more easily accessible to them. As Herda (2010, 2015) highlights, people’s immediate environment can influence the availability of certain considerations over others. Those living in areas with relatively large numbers of immigrants from certain regions – say, the Middle East – will more easily associate an “immigrant” with someone from the Middle East. As a result, the share of immigrants from that region or with a certain attribute relatively common within that region will be overestimated.

Given that the media not only shape which kinds of immigrants are mentioned more frequently but also how they are referred to (e.g., Van der Brug et al. 2015), the availability heuristic may be associated with positions on immigration (Carvalho, Duarte, and Ruedin 2024). Blinder (2015) noted that media discussion of asylum seekers in the United Kingdom tends to be negative (compare Leveson 2012). Media coverage thus not only encourages people to think about asylum seekers when they think about immigrants but also encourages people to have negative views and thus to prefer reduced immigration levels. By contrast, Blinder (2015) found that international students did not feature prominently in the media, and negative opinions were rarely expressed. The availability heuristic can explain differences both at the country level and at the individual level, where contact with immigrants may increase the availability of relevant information. At the cross-national level, we would expect differences in media reporting about immigrants to correlate with average differences in stereotypes about immigrants. At the individual level, we would expect that exposure to certain media sources or exposure to immigrants in daily life to correlate with differences in perceptions of immigrants. Apart from such theoretical considerations, we can only call for further research to cover additional contexts and seek ways to test the influence of various heuristics on the formation of stereotypes and the perception of immigrants, including work on racialized images of immigrants and who in the population is more likely to subscribe to them. Here we have shown that attitudes play an important role in shaping perceptions, and given that, it may not be surprising that perceptions of different immigrant groups do not correspond to reality and are therefore often blurred.