Immigration and the Politics of Social Welfare

Scholars of the politics of social welfare in the United States have convincingly demonstrated that racial stereotypes and resentment have played a central role in the attitudes of Americans about these programs (Gilens Reference Gilens1996, Reference Gilens1999; Peffley, Hurwitz, and Sniderman Reference Peffley, Hurwitz and Sniderman1997; Soss and Schram Reference Soss and Schram2007; Winter Reference Winter2008). Specifically, many white Americans have associated means-tested programs like cash “welfare” (formerly Aid to Families with Dependent Children, now Temporary Aid for Needy Families) and food stamps with blacks. This literature offers important insights into welfare politics in the United States, but is narrowly focus on black-white racial dynamics, raising the question of how immigrants fit into the story. Moreover, while it establishes the importance of associating social welfare programs with a specific racial or ethnic group, it offers little insight on the conditions under which people make such associations in the first place.

The work of Schneider and Ingram (Reference Schneider and Ingram1993) on the social construction of target populations clearly compliments the literature on race and welfare, with its emphasis on how perceptions of policy beneficiaries shape the politics surrounding that policy. In contrast to the narrow focus of the race and politics literature, Schneider and Ingram’s theory seeks to explain a much wider range of policies with a wider range of (real or imagined) target populations. The central insight of this literature is that policy design corresponds to social constructions of these target groups. The target population for programs like cash welfare and food stamps is viewed as weak and dependent, unable to make choices for itself, and the policies themselves have paternalistic designs that reflect this. The correspondence to negative stereotypes about racial minorities is obvious.

Recently, public perceptions of the ethnic makeup of social welfare target populations have taken on a new importance in the study of European politics, as increased immigration from non-Western countries seems to have altered the politics of some of the world’s most robust welfare states. Drawing on the American race and welfare literature, numerous scholars of European politics (e.g., Oorschot Reference Oorschot2006; Mau and Burkhardt Reference Mau and Burkhardt2009; Eger Reference Eger2010, Larsen Reference Larsen2011; Schmidt-Catran and Spies Reference Schmidt-Catran and Spies2016; Alesina, Miano, and Stantcheva Reference Alesina, Miano and Stantcheva2019) have documented the ways in which the influx of immigrants has undermined natives’ support for social provision. A related literature in European politics (e.g., Andersen and Bjørklund Reference Andersen and Bjørklund1990; Waal et al. Reference Waal, Achterberg, Houtman, Koster and Manevska2010) examines the phenomenon of “welfare chauvinism,” in which citizens strongly support generous social provision as long as it is targeted to in-groups.Footnote ²

Just as some European scholars have argued that the social welfare politics of non-Western immigrants in Europe are analogous to those of blacks in the United States, scholars of American politics have begun to consider the possibility that immigrants and Hispanic Americans in the United States have also become an important target population.Footnote ³ While stereotypes about black dependence and work ethic may have been more salient to the “welfare reform” debates of the 1990s, concerns about immigration certainly played a role. As Brown (Reference Brown2013a, Reference Brown2013b) notes, the discourse about immigrants (or Hispanics and Latinos) and welfare reform in Southwestern states was distinct from the “welfare mother” rhetoric that pervaded states in the Southeast where blacks were the salient target population; discourse in the former states focused on questioning the legality of welfare receipt by immigrants. Soss et al. (Reference Soss, Schram, Vartanian and O’Brien2001) and Hero and Preuhs (Reference Hero and Preuhs2007) find that large Hispanic and immigrant populations were associated with stricter and less generous state welfare policies following the adoption of the 1996 welfare reform law (also see Hawes and McCrea Reference Hawes and McCrea2017). Zhu and Xu (Reference Zhu and Xu2015) demonstrate that state Medicaid policies also vary widely in terms of how open they are to immigrants, with significant consequences for immigrant populations.

Recent public opinion research on welfare and immigration supports the idea that perceptions of immigrants or Latinos and Hispanics are relevant to welfare attitudes. Using American National Election Study data, Garand, Xu, and Davis (Reference Garand, Xu and Davis2017) find that variables capturing Americans’ sentiments about immigrants and immigration are among the most powerful predictors of attitudes about welfare. Abrajano and Hajnal (Reference Abrajano and Hajnal2015, 11), in their study of how the “threat” of immigration has motivated a range of attitudes and behavior in white American voters, identify social welfare as an area that has been politically reshaped by this dynamic. Knoll and Shewmaker (Reference Knoll and Shewmaker2015) find that “nativism” is an important factor in opposition to the Affordable Care Act, a major piece of social legislation.

An analysis of internet search behavior underscores the increasing importance of immigrants as a perceived target population of social welfare policies. Figure 1 shows the monthly relative search frequency for three Google searches related to “welfare” for the period of June 2013 through June 2019 in the United States.Footnote ⁴ In addition to “welfare immigrants,” I include “welfare black,” in reference to a well-established perceived target population, and “welfare work,” since considerations of work and work ethic have been central to social welfare politics in the United States for decades.Footnote ⁵ The trends are striking⸺at least as measured by Google searches, immigrants have overtaken blacks as a salient target population of “welfare.” It appears that Donald Trump’s presidential campaign, election, and presidency, which have focused heavily on opposition to immigration and occasionally on restricting immigrants’ access to government benefits (e.g., Wegmann Reference Wegmann2019), are an important part of this story. In the twenty-three months prior to Trump’s announcement of his presidential campaign in June of 2015, searches for “welfare immigrants” exceeded searches for “welfare black” only twice.Footnote ⁶ In forty of the forty-eight succeeding months, “welfare immigrants” searches were more common. Prior to Trump’s election in November of 2016, searches for “welfare immigrants” had never exceeded searches for “welfare work” in the entire history of Google Trends data going back to 2004 (155 months); this happened seven times in the succeeding thirty-two months. Searches for “welfare black” have never exceeded searches for “welfare work.” Perceptions about immigrants and social welfare are clearly in need of greater scholarly attention.

Figure 1

Relative monthly frequency of Google search terms in the United States, June 2013 to June 2019

An important component of the question of how immigration affects welfare politics is whether the effects (if any) result from actual exposure to immigrant populations at the local level or the individual’s understanding of national trends. Some work has found that local-level exposure to immigration reduces welfare support, both in the United States (Abrajano and Hajnal Reference Abrajano and Hajnal2015, 14) and in Europe (Dahlberg, Edmark, and Lundqvist Reference Dahlberg, Edmark and Lundqvist2012; Eger and Breznau Reference Eger and Breznau2017), while some find that such contact reduces stereotypes about laziness (Fox Reference Fox2004) and makes citizens more amenable to redistribution (Burgoon, Koster, and Egmond Reference Burgoon, Koster and Egmond2012).

Most work in this vein, like most work that considers social welfare and immigration attitudes separately, focuses on the attitudes of white Americans. The findings of work on immigration that does contrast whites’ attitudes with those of minorities suggest that whites are unique in their views on that issue (Carter Reference Carter2019; Rocha et al. Reference Rocha, Longoria, Wrinkle, Knoll, Polinard and Wenzel2011). It is likely that they differ from minorities in their views of the relationship between immigration and social welfare as well.

The literature has convincingly established immigration and Latino or Hispanic population growth as important factors in American welfare politics. The missing ingredient, however, remains the same as it has been in the predecessor literatures on race and welfare in America, social construction of target populations, and immigration and welfare in Europe. We lack strong causal evidence about the conditions under which individuals come to associate policies with a given group. When and why do Americans assume welfare policy will benefit immigrants? In the following section, I develop a theoretical framework and offer a series of testable hypotheses related to these questions.

Theory and Hypotheses

I begin with a well-founded theoretical argument: that to form an evaluative opinion of a social welfare policy, many if not all individuals first need some idea of who benefits or would benefit from the policy. Without a picture of who will benefit, it will be difficult for many people to render a judgment on whether those people are deserving of assistance, a key consideration in public opinion on such programs. If explicit information about beneficiaries is not readily available, people will fill in the gaps by making educated guesses or assumptions.Footnote ⁷ In doing so, I argue that they are more likely to think about out-groups when they are salient or at the “top of one’s mind” (Zaller Reference Zaller1992), and that rhetoric or information that presents the out-group (specifically immigrants) as a threat, à la Abrajano and Hajnal (Reference Abrajano and Hajnal2015),Footnote ⁸ will raise its salience.

The question is what type of threat is most relevant. There is a debate within broader discussions of immigration politics over whether the more important perceived threat is to values, culture, and demographics, or to job security and wages. In the realm of social welfare, however, the relevant economic threat is not to the labor market, but to public resources⸺the concern is that immigrants will take advantage of the welfare state, and particularly that unauthorized immigrants will do so illegitimately (Brown Reference Brown2013a, Reference Brown2013b). Hainmueller and Hiscox (Reference Hainmueller and Hiscox2010) find that concerns about immigrants consuming social welfare resources drive anti-immigration sentiment,Footnote ⁹ while Hero and Preuhs (Reference Hero and Preuhs2007) present evidence that states with more inclusive welfare policies toward immigrants also tend to have lower benefit levels, suggesting that immigration has the potential to “erode” benefits for all residents. Anti-immigration advocates have used such arguments, highlighting immigrants’ relatively high use of social welfare programs (e.g., Camorata Reference Camorata2015). This is paired with the argument that immigrants, and particularly unauthorized immigrants, pay less in taxes than native-born Americans, as emphasized in a report from the Heritage Foundation, a prominent conservative think tank (Rector and Richwine Reference Rector and Richwine2013). Survey evidence suggests that Americans are concerned about immigrants paying their fair share in taxes (Crepaz Reference Crepaz2008, 244).

Of course, concerns about the changing culture and demographics of the United States⸺sometimes termed “acculturation fear”⸺ is a powerful force in explaining Anglo-Americans’ views about immigration levels (Branton et al. Reference Branton, Cassese, Jones and Westerland2011), and may also be relevant in the domain of social welfare. To the extent that such cultural fears include stereotypes about poor work ethic among Latinos, Hispanics, or immigrants (Fox Reference Fox2004; Branton et al. Reference Branton, Cassese, Jones and Westerland2011), the link to assumptions about welfare use is clear.

However the assumption is triggered, once an individual associates a specific outgroup with a social welfare policy, their attitudes about the policy will become linked to their attitudes about the group. In the present case, opposition to immigration or hostility to immigrants should be much more tightly linked to social welfare policy attitudes among people who assume policy would benefit immigrants than among people who do not.

Drawing on this framework and the insights of the existing literature, I generate a set of hypotheses that I pre-registered, along with the study design, on the Evidence in Governance and Politics (EGAP) website on July 14, Reference Haselswerdt2017, prior to any data collection (Haselswerdt Reference Haselswerdt2017).

The first two hypotheses deal with the effects of “threat” primes on individuals’ assumptions about social welfare beneficiaries. In particular, I am interested in the assumption that such policies will benefit immigrants to the exclusion of native-born Americans. I argue that this is the key assumption citizens can make regarding social welfare and immigration, since it suggests that an out-group is benefiting at the expense of an in-group. If a respondent assumes that both immigrants and natives will benefit, for example, immigration attitudes are presumably less relevant to evaluation of the program than if he or she assumes only immigrants will benefit.Footnote ¹⁰

• H1: Americans who are primed to think about immigrants in a context of cultural or demographic threat will be more likely to assume that social welfare policies benefit immigrants to the exclusion of native-born Americans than will those who are not primed to think about immigrants.

• H2: Americans who are primed to think about immigrants in a context of economic threat or fiscal burden will be more likely to assume that social welfare policies benefit immigrants to the exclusion of native-born Americans than will those who are not primed to think about immigrants.Footnote ¹¹

The next two pairs of hypotheses express expectations about the conditionality of these experimental effects. First, I consider the possibility that these effects will be more pronounced among those with anti-immigration views, since such people should be more susceptible to immigration-related “threat” primes. On the other hand, it is also possible that these primes affect people across the spectrum of immigration attitudes in a similar manner⸺Pérez (Reference Pérez2016) demonstrates that implicit or unstated attitudes and stereotypes about immigrants and Latinos may be more important to public opinion than explicit attitudes expressed openly on survey questions, suggesting that many people with more “liberal” attitudes on immigration may nevertheless harbor implicit stereotypes about immigrants and work ethic.

• H3a : The effect of the cultural/demographic threat prime on the assumption that social welfare policies benefit immigrants to the exclusion of native-born Americans will be strongest among those most opposed to immigration.

• H3b : The effect of the economic/fiscal threat prime on the assumption that social welfare policies benefit immigrants to the exclusion of native-born Americans will be strongest among those most opposed to immigration.

Next, I consider the ways that experimental effects may vary by local context. The existing literature has produced conflicting arguments and expectations about whether “exposure” to a given out-group will make one more or less likely to view members of that group as a threat or as a relevant target population for social welfare programs (Abrajano and Hajnal Reference Abrajano and Hajnal2015; Burgoon, Koster, and Egmond Reference Burgoon, Koster and Egmond2012; Dahlberg, Edmark, and Lundqvist Reference Dahlberg, Edmark and Lundqvist2012; Eger and Breznau Reference Eger and Breznau2017; Fox Reference Fox2004). In this case, the focus is not on the direct effects of exposure, but whether such exposure makes one more or less susceptible to primes that evoke a threat narrative.

A number of patterns are plausible here. Perhaps people who live around large numbers of immigrants, Latinos, or HispanicsFootnote ¹² are relatively tolerant and therefore more resistant to primes that cast immigrants as threatening or dependent. Conversely, perhaps frequent contact with such groups makes immigration salient, to the point where people living in such areas already assume social welfare programs benefit immigrants at the expense of native-born Americans, which would also blunt the effects of primes. I argue for a third possibility⸺contact with large immigrant populations should make immigrant target populations relatively accessible as a consideration in the minds of Americans, making them more susceptible to primes. This argument is similar to the “politicized places” hypothesis advanced by Hopkins (Reference Hopkins2010), which posits that local demographics shape immigration attitudes in combination with political information or frames.

• H4a: The effect of the cultural/demographic threat prime on the assumption that social welfare policies benefit immigrants to the exclusion of native-born Americans will be strongest among those who live in regions where Latinos or Hispanics make up a relatively large part of the population.

• H4b : The effect of the economic/fiscal threat prime on the assumption that social welfare policies benefit immigrants to the exclusion of native-born Americans will be strongest among those who live in regions where Latinos or Hispanics make up a relatively large part of the population.

The final hypothesis concerns the impact of assumptions about target populations on policy attitudes. Existing work convincingly establishes that views about immigration are important to Americans’ attitudes about social policy (Garand, Xu, and Davis Reference Garand, Xu and Davis2017). I argue that this relationship should exist only when the individual believes a policy will benefit immigrants, and particularly when it benefits them to the exclusion of native-born Americans. Absent such assumptions, there is no reason nativistic beliefs should influence social welfare attitudes. Indeed, work on so-called “welfare chauvinism” in Europe suggests that opponents of immigration may be especially favorable to social welfare programs if they believe the programs exclude immigrants (Andersen and Bjørklund Reference Andersen and Bjørklund1990; Koster, Achterberg, and Waal Reference Koster, Achterberg and van der Waal2012; Reeskens and Oorschot Reference Reeskens and van Oorschot2012; Waal et al. Reference Waal, Achterberg, Houtman, Koster and Manevska2010).

• H5: When Americans assume a social welfare policy benefits immigrants to the exclusion of native-born Americans, their attitudes about immigration will become more important to their views of that social welfare policy as compared to similar Americans who do not make this assumption.

Experimental Design and Data

I test these hypotheses with a survey experiment focused on a generic social welfare policy. The experiment randomly exposes respondents to primes of fiscal and cultural threat related to immigration, measures whether these primes make them more likely to believe that immigrants (and not native-born Americans) will benefit from a generic social welfare policy, and allows me to determine if these assumptions make attitudes about immigration more important to respondents’ evaluations of the policy.

After exposing respondents to the randomly assigned primes (described later in this section), I present them with a brief description of the hypothetical policy (which would provide “government assistance in the form of extra money to help with living expenses” to “people with lower incomes”) and ask them whether they would approve or disapprove of the program, using a seven-point Likert scale.

The following question asks respondents what “groups in American society” would be most likely to benefit from the proposed program. Respondents were allowed to select multiple choices from a list of sixteen groups. To avoid forcing respondents to see the question strictly in terms of race, ethnicity, or citizenship, the list includes groups based on gender and economics (e.g., “Women,” “Poor people,” “Big business,” “Labor unions”) as well as race, ethnicity and national origin. The choices of interest are “Immigrants” and “Americans born in the United States.”Footnote ¹³ Recall that I am interested in the assumption that the policy will benefit immigrants and not Americans born in the United States, so the dependent variable in the analyses is a dichotomous variable coded “1” if the respondent selected “Immigrants” without also selecting “Americans born in the United States.”Footnote ¹⁴

A different potential approach would be to offer “illegal,” “unauthorized,” or “undocumented” immigrants, or perhaps “non-citizens,” as one of the categories, since so much rhetoric on immigration focuses on legality and citizenship. It is this very salience, however, that motivates my decision not to add any descriptors to the “immigrants” option. By avoiding loaded language in these response options, I hope to avoid blatantly priming respondents, especially those in the control group, to think about immigration if it is not already on their minds. Nevertheless, “illegal immigrants” are obviously a politically important target population, and an important consideration for future research in this direction.

Prior to these key questions, respondents were randomized into three treatment groups and presented with different combinations of priming questions. The “cultural threat prime” treatment group was asked the following question about driver’s license tests:

The focus on languages other than English in this question is meant to prime concerns about the changing demographics of the United States.

The “fiscal threat prime” treatment group was posed the following question about tax enforcement: “A recent report by a policy research organization found that unauthorized immigrants pay much less in taxes than the average taxpayer.Footnote ¹⁶ Do you agree or disagree with the idea that the IRS should have greater resources to improve tax enforcement?”

The intention of both the cultural and fiscal threat primes is to evoke a sense of threat in the respondent. Recall that the theoretical argument of the study is that concerns about the threat posed by immigration will make immigrants more salient or accessible in a person’s mind when they think about social welfare programs. Neither prompt directly deals with immigrants’ use of social welfare policies, but both could evoke the connection: the cultural prime via its reference to driver’s licenses (a form of government identification that confers benefits and privileges), and the fiscal prime by raising concerns about immigrants failing to contribute to a state that may benefit them in unstated ways.

The goal of this experiment is to determine the impact of primes that relate to immigration specifically, but both treatments have the potential to raise other considerations about drivers’ licensing in one case and fiscal or tax issues in the other. To more precisely measure the impact of information that is related to immigration per se, the study includes “placebo” questions on driver’s licensing (about raising the minimum age for learner’s permits) and taxes (about generic tax enforcement) that have no content relevant to immigration. In the cultural threat treatment group, respondents also answered the tax enforcement placebo question, while in the tax treatment group, they answered the learner’s permit placebo question. This ensures that the differences between each of the two treatment conditions and the control group offers as clean an estimate as possible of the impact of the immigration-related prime.Footnote ¹⁷

Tests of H3a, H3b , and H5 require a measure of anti-immigration sentiment. I construct an index to tap these attitudes using a battery of five questions about immigration issues drawn from the 2012 American National Election Study (ANES).Footnote ¹⁸ The questions deal with deportation policy for unauthorized immigrants, allowing people who entered the United States illegally as children to become citizens, allowing immigration status checks on suspected unauthorized immigrants, the likelihood that immigrants will take jobs away from current citizens, and the appropriate level of legal immigration. Importantly, none of these questions deal with social welfare policy, making the index plausibly exogenous to social welfare attitudes. I generate the scale by standardizing the responses to each question to adjust for the varying size of answer scales, reversing scales as necessary so that higher values indicate more opposition to immigration, then adding the responses together and rescaling the resulting variable to range from 0 to 1.Footnote ¹⁹

Including a battery of immigration policy questions carries with it the risk of unintentional priming effects. If the immigration policy items make immigration a salient consideration for all respondents, it could interfere with the intentional priming effects of interest in the experimental treatments. To avoid this, I placed the battery of immigration questions at the end of the survey, after the experiment, the beneficiary group question, and the social welfare policy question. This choice carries with it an important trade-off: the potential for post-treatment bias (Montgomery, Nyhan, and Torres Reference Montgomery, Nyhan and Torres2018), where a variable used as a moderator is potentially influenced by the treatment questions. I conducted a follow-up study with a different survey flow to explore the implications of these ordering choices (refer to online appendix F).

Another possible approach to measuring immigration-related opinion would be to directly measure respondents’ attitudes about the relevant target group (immigrants or Latinos and Hispanics) rather than about immigration policy. Immigration policy attitudes may indirectly tap implicit attitudes about Latinos in any case (Pérez Reference Pérez2016), but it is possible that different patterns in assumptions and social policy attitudes would emerge if a more direct measure were used. I use immigration policy attitudes here because I am especially interested in the linkage between immigration policy and social policy in Americans’ minds, but explore an alternative measurement strategy (directly tapping positive or negative feelings about immigrants) in the follow-up study (refer to online appendix).

I test H4a and H4b by including the Hispanic population percentageFootnote ²⁰ in the respondent’s ZIP (postal) code (captured as embedded data), derived from American Community Survey (ACS) ZIP Code Tabulation Area (ZCTA) five-year estimates from 2017, divided by its standard deviation to ease interpretation. I chose this measure, which is included in the pre-registration details, since the size of the local Hispanic or Latino population is likely to be the most visible or salient clue for most citizens about how many actual immigrants live near them. When considering this variable, it may be appropriate to think of individual survey respondents as being nested within geographic regions rather than as independent observations, so I employ state-clustered standard errors and state fixed effects in some specifications.Footnote ²¹

Of course, anti-immigration attitudes and geographic factors are likely correlated with other factors that could influence policy assumptions and approval, most notably general liberal-conservative ideology, party identification, and attitudes about non-immigrant minority groups. The survey includes measures of ideology and party identification (using seven-point scales), as well as respondent race and ethnicity,Footnote ²² gender, age, income, education, and the national origin of the respondent and his or her parents. I also include the standard battery of four questions used to measure “symbolic racism” or “racial resentment” (Henry and Sears Reference Henry and Sears2002) among whites toward blacks, since this scale is a consistently powerful predictor of welfare attitudes.Footnote ²³ Responses to these questions are used as control variables where appropriate.Footnote ²⁴

The completed sample of 2,209 self-reported U.S. citizens was recruited from an online panel with the aid of Qualtrics, a survey research firm. Quotas were used to ensure national representativeness in terms of gender (52% female), age (13% 18–24, 35% 25–44, 35% 45–64, 17% 65 and older), and education (13% some high school or less, 30% high school graduate, 29% some college, 18% college graduate, and 10% graduate school degree). The sample is either representative or close to representative of the American public on key attitudinal dimensions (including anti-immigration attitudes), though the respondents are less partisan than the public. In terms of demographics, Hispanics or Latinos are somewhat underrepresented.Footnote ²⁵ The survey was conducted July 19–25, 2017. The median time to completion was 5 minutes and 39 seconds, and 30 respondents who completed the survey in less than one-third of that time were dropped to ensure the quality of responses.Footnote ²⁶

Discussion and Conclusion

Americans’ perceptions of who benefits from government aid are clearly relevant to the politics of social welfare. This study demonstrates that these perceptions are malleable: certain primes or information can lead some Americans to assume that a social welfare policy will benefit immigrants to the exclusion of native-born Americans. In a pre-registered survey experiment with a national sample that is representative on key dimensions, I find that a common anti-immigration talking point–that unauthorized immigrants pay less in taxes than American citizens–makes respondents more likely to assume that a described social welfare policy benefits immigrants but not native-born Americans.Footnote ⁴² Information about linguistic changes meant to prime cultural threat may have had a similar effect, but only for certain respondents.Footnote ⁴³ Importantly, these effects were not conditional on attitudes about immigration⸺respondents who expressed general hostility to immigration were not more or less receptive to these primes than pro-immigration respondents. This is consistent with the arguments of Pérez (Reference Pérez2016) on the importance of implicit attitudes or stereotypes in immigration opinion⸺even outwardly “liberal” Americans may harbor such unexpressed stereotypes.

There was evidence that effects varied by local context, however. The effect of the economic threat prime on assumptions was strongest among those in areas with low concentrations of Hispanic residents, and weakened among those in high-Hispanic-population areas. This suggests that it may be easier to shape the perceptions of those with little actual exposure to immigrants (or ethnic populations likely to be perceived as immigrants) with arguments and information about the fiscal impact of immigration. By contrast, while the cultural threat prime had no effect on most respondents, it may have made those near large Hispanic populations more likely to assume that the policy would benefit immigrants to the exclusion of native-born Americans. It is possible that exposure to ethnic (or perhaps linguistic) outgroups makes considerations about cultural change more accessible to citizens, making their assumptions about policy more susceptible to cultural threat primes. These findings are especially interesting in light of the arguments and findings of Hopkins (Reference Hopkins2010) and Abrajano and Hajnal (Reference Abrajano and Hajnal2015) on immigration and threat, and to the larger debate about how exposure to outgroups shapes social welfare attitudes.

Whatever may trigger assumptions about immigrant beneficiaries, there is little doubt that the assumption matters for policy attitudes. Attitudes about immigration appear unrelated to social welfare attitudes (all else equal) when the respondent does not assume that immigrants will benefit, but when they do, opposition to immigration becomes a powerful predictor of opposition to welfare. There is also evidence that anti-immigration attitudes are correlated with social welfare support, all else equal, especially when the respondent assumes native-born Americans will benefit. This pattern is consistent with “welfare chauvinism,” a phenomenon mostly associated with European politics (e.g., Andersen and Bjørklund Reference Andersen and Bjørklund1990; Koster, Achterberg, and Waal Reference Koster, Achterberg and van der Waal2012; Reeskens and Oorschot Reference Reeskens and van Oorschot2012; Waal et al. Reference Waal, Achterberg, Houtman, Koster and Manevska2010). These findings and those of Jardina (Reference Jardina2019) suggest that scholars of American welfare politics should pay attention to beliefs about in-groups as well as prejudice towards out-groups.

How well does this survey experiment generalize to the real-life politics of immigration and social welfare? While there is always some artificiality in the presentation of primes in this sort of study, the primes were selected to be reflective of arguments and data presented by real political actors, such as the Heritage Foundation. The question tapping assumptions of target population was carefully designed to avoid obtrusively forcing considerations of race, ethnicity, or immigration on the respondent. Moreover, the work of Garand, Xu, and Davis (Reference Garand, Xu and Davis2017) and the web search analysis presented in Figure 1 indicate that the patterns discussed in this study are highly relevant to social welfare attitudes in the modern United States.

This research speaks to a larger question: has increasing immigration undermined American political support for the welfare state? These results suggest that it may have, but only for a minority of the population. Recall that the key assumption of the policy benefiting immigrants to the exclusion of those born in the United States was made by only 21.5% of respondents overall, and only 26% of those who received the fiscal threat prime. Overall, support for the generic social welfare policy was moderately strong, with a mean of 4.6 and a median of 5 (“approve somewhat”) on the 7-point scale. Only 27.5% of respondents expressed disapproval, and fewer than 10% strongly disapproved, compared to 21% that strongly approved (the modal answer). Among non-Hispanic whites, these approval numbers are only slightly weaker. Nevertheless, we should not underestimate the importance of these dynamics among the subset of the public that does associate these policies with immigrants and links their social welfare attitudes to their views about immigration. Recent history demonstrates that the most conservative subset of voters and activists wield real power in American politics due to high levels of organization and mobilization, particularly in Republican primary elections. As opposition to immigration becomes a more and more salient part of the Republican platform and conservative ideology, these patterns are likely to become more important to public policy.Footnote ⁴⁴

For advocates and leaders seeking to promote tolerance of immigrants or support for social welfare programs, the findings suggest an important focus: the perception of immigrants as contributing less than their fair share in taxes. Countering this perception with information about the contributions that documented and undocumented immigrants do make to the tax base (e.g., Gee et al. Reference Gee, Gardner, Hill and Wiehe2017; Denhart Reference Denhart2017) could help to both combat stereotypes about immigrants as a drain on government and distance social welfare from the immigration issue.

These findings also suggest several avenues for future work. Are economic or fiscal arguments generally more powerful than cultural ones in the politics of social welfare and immigration, or is this finding specific to the treatments tested here? What are the mechanisms that link local-level exposure to immigrant populations, rhetoric about threat, and assumptions about social welfare? Do these patterns work differently when Americans are asked to consider “undocumented” or “illegal” immigrants, as opposed to immigrants in general? How real, and how consequential, is American welfare chauvinism, given that so many Americans who are hostile to immigration are also ideologically conservative and supportive of Republican politicians? Future work should also examine where American-born racial and ethnic minorities, including Hispanic or Latino citizens, fit into this story. The findings of the present study suggest that the experimental effects are confined to white non-Hispanic respondents, but studies more squarely focused on minority populations are needed to probe the ways that nonwhites view immigrants relative to social welfare.

A list of permanent links to Supplemental Materials provided by the authors precedes the References section.