In September 2022, Florida Governor Ron DeSantis drew national attention when it was revealed that his administration had orchestrated the transportation of more than four dozen asylum seekers from San Antonio, Texas, to the affluent and predominantly liberal island of Martha’s Vineyard, Massachusetts. For all intents and purposes, this effort was a political stunt aimed at exposing “liberal hypocrisy” over immigration (Bump Reference Bump2022). In DeSantis’ own words:
The minute even a small fraction of what those border towns deal with every day is brought to their front door, they all of a sudden go berserk. And they’re so upset that this is happening. And it just shows their virtue signaling is a fraud (Kam Reference Kam2022).
Similar efforts have been undertaken by the administration in Texas with Governor Greg Abbott’s “busing program,” which aims to shift “the burdens imposed by open-border advocates in other parts of the country.” As of February 2024, this has cost Texans a collective $148 million (Martínez-Beltrán Reference Martínez-Beltrán2024). In the two years since its inception, the program has transported more than 100,000 migrants to various liberal cities across the country.Footnote 1
The core argument made in the justification of these programs is similar to the broader concept of NIMBYism (not-in-my-backyard). This phenomenon describes the tendency for people to support policies in the abstract but to resist their implementation when the policy may affect their own community (Dear Reference Dear1992; Marble and Nall Reference Marble and Nall2021). In this context, individuals may be favorable toward immigrants when their community is unlikely to receive immigrants. Their attitudes, however, may become less favorable if they believe immigrants will settle in their neighborhood. In other words, spatial scale may alter their considerations.
Despite a NIMBY argument being used in support of these transportation programs, extant scholarship on immigration has yet to provide direct causal evidence of the NIMBY effect. Much research explores the economic and cultural determinants of immigration attitudes (for a review, see Hainmueller and Hopkins Reference Hainmueller and Hopkins2014), and scholars debate the extent to which exposure to immigrants affects anti-immigrant hostility (Hangartner et al. Reference Hangartner, Dinas, Marbach, Matakos and Xefteris2019; Kotzur, Schäfer, and Wagner Reference Kotzur, Schäfer and Wagner2019). However, fewer studies explore how proposed spatial scale alters an individual’s evaluation of prospective immigrants, as argued by Republican elites. Moreover, this omission leaves no expectation of how important any variation is relative to other determinants of immigration attitudes. Stated another way, to what extent do immigration preferences vary by spatial scale?
We conduct a conjoint experiment with a nationally representative sample and find causal evidence that Americans exhibit a substantively small but statistically significant preference against immigrants posited to move into their neighborhood but are indifferent to all other spatial considerations. Although spatial scale influences mass evaluations of immigrants, the effect is modest compared to other prominent determinants of immigration attitudes. Moreover, despite elite claims of liberal hypocrisy—that Democrats seem to prefer immigrants in the abstract but actually oppose them in their community—we find little evidence that the effect of spatial scale on immigrant evaluation varies by the partisanship of the respondent. Taken together, our results suggest that despite popular rhetoric underpinning the justification for politicized migrant transportation programs, the NIMBY effect pales in comparison to individual-level attributes of an immigrant.
We conduct a conjoint experiment with a nationally representative sample and find causal evidence that Americans exhibit a substantively small but statistically significant preference against immigrants posited to move into their neighborhood but are indifferent to all other spatial considerations.
This article focuses primarily on immigration, but this represents only one instance in which the broader NIMBY effect poses challenges for public support of political decisions and policies that involve land use. Across many domains (e.g., environmental policy, housing policy, and support for and attitudes toward the unhoused), the presence of a NIMBY effect would suggest that support toward certain public policies is conditional on spatial considerations, or how it affects an individual’s livelihood. Measures of spatial scale, however, are often omitted from studies of public opinion. Although we hesitate to generalize our findings to these other cases, our results suggest that studies that ignore issues of spatial scale may miss potential variation in individual-level support and attitudes. It also suggests that despite being seen as a potential barrier to collective action, this NIMBY effect is modest in size.
To the extent that Americans’ preferences toward immigrants shape public opinion on immigration policy—with immigration being cited as one of the most important issues among American voters in the lead-up to the 2024 presidential election (Brenan Reference Brenan2024)—our findings indicate that voters are less likely to be influenced by where immigration policy is implemented than by the individual motivations and characteristics of the immigrants themselves.
NIMBYISM AND IMMIGRATION PREFERENCES
In its original formulation in urban studies, NIMBYism refers to the tendency of residents to resist placement of important community facilities in their vicinity despite their agreement on the necessity of those projects (Dear Reference Dear1992; Schively Reference Schively2007).Footnote 2 For instance, California homeowners may vote in favor of local ordinances supporting affordable housing developments, acknowledging their regional housing scarcity. When the same ordinances empower these homeowners to decide whether such developments are built within close proximity to their own homes, however, they often oppose their construction. In other words, support for housing initiatives is conditional on spatial considerations—specifically, whether the proposed development is in one’s own “backyard” (Marble and Nall Reference Marble and Nall2021). We adopt this intuition and apply it more broadly: individuals’ preferences for public policies that require land use (Fischel Reference Fischel2001) and attitudes toward the objects of these policies will vary by spatial scale.
Importantly, as NIMBYism is observed from “community groups facing an unwelcome development in their neighborhood” (Dear Reference Dear1992, 288; italics added), we expect to see NIMBYism only at a community or neighborhood level. That is, we consider NIMBYism as mainly an issue of spatial scale or a matter of granularity, in which physical proximity to immigrants only matters at the neighborhood level. The nuanced difference between two related terms merits note. Whereas spatial scale concerns “the appropriate spatiotemporal unit of analysis and level of abstraction for empirical research” (Brenner Reference Brenner1998, 460), physical proximity refers to the physical closeness between objects. Therefore, the lens of spatial scale leads to the expectation that the effect of the expected location of an immigrant will be nonlinear with the distance from or physical proximity to the immigrant. In simple terms, we expect that the considerations one has when evaluating immigrants who may come to their community differ from the considerations one has when the immigrants are considered in more abstract terms. Yet, there should be no difference in considerations between different levels of abstraction outside of one’s own community.
Whereas a number of studies suggest that physical proximity or exposure to immigrants in natives’ daily lives increases anti-immigrant attitudes (Enos Reference Enos2014; Hangartner et al. Reference Hangartner, Dinas, Marbach, Matakos and Xefteris2019), other research suggests that meaningful and positive interaction with immigrants beyond mere exposure can reduce hostility toward immigrants (Clayton, Ferwerda, and Horiuchi Reference Clayton, Ferwerda and Horiuchi2021; Kotzur, Schäfer, and Wagner Reference Kotzur, Schäfer and Wagner2019; see also Lee Reference Lee2024; Liao, Malhotra, and Newman Reference Liao, Malhotra and Newman2020). Unlike this line of research that examines the effect of exposure to immigrants based on contact theory (Allport Reference Allport1954) or intergroup threat theory (Stephan and Stephan Reference Stephan, Stephan and Oskamp2000), NIMBYism deals with a phase before exposure that both theories predict: avoidance (Rothbart and John Reference Rothbart and John1993, 42). That is, natives may avoid being contacted by or engaged in meaningful interaction with immigrants “in their backyard” if they have a choice before the arrival of immigrants.
HYPOTHESES
Unifying research in the study of immigration attitudes has bridged prior divisions (Hainmueller and Hopkins Reference Hainmueller and Hopkins2014) and provided a comprehensive assessment of the attributes that influence natives’ assessment of immigrants (Hainmueller and Hopkins Reference Hainmueller and Hopkins2015). Importantly, however, these studies often omit measures of spatial scale or proximity. It remains unclear whether respondents in these studies are thinking about immigrants simply migrating to the United States in the abstract, or if they are thinking about the impact of immigrants on their own community. Moreover, we do not know if this distinction changes natives’ evaluations of prospective immigrants.
Drawing on the literature on NIMBYism, we expect that individual natives’ preferences about whom to admit will vary by spatial scale. Unlike a situation in which people merely think of an immigrant migrating somewhere in their country in the abstract (e.g., the United States, their state, or even their city), we expect that ceteris paribus, individuals will be less likely to prefer an immigrant who is proposed to reside in their neighborhood (Hypothesis 1).
Unlike a situation in which people merely think of an immigrant migrating somewhere in their country in the abstract (e.g., the United States, their state, or even their city), we expect that ceteris paribus, individuals will be less likely to prefer an immigrant who is proposed to reside in their neighborhood (Hypothesis 1).
Our second expectation concerns how the NIMBY effect should differ by partisanship. As made clear in the comments by Governor DeSantis, political stunts surrounding migrant transportation programs are justified by the idea that Democrats or liberals hold hypocritical views on immigration, only holding pro-immigration attitudes because they do not have to consider the impact of immigrants on their community. Moreover, given that Democrats or liberals are more likely to hold pro-immigration attitudes than Republicans or conservatives in the United States (Citrin et al. Reference Citrin, Green, Muste and Wong1997; Hawley Reference Hawley2011), if spatial considerations alter preferences, this NIMBY effect should be more pronounced among those who already hold pro-immigrant attitudes.Footnote 3 Republicans, on the contrary, will be more likely to hold anti-immigrant attitudes regardless of the spatial scale compared to Democrats.
In keeping with the predictions of NIMBYism, we expect the effect of spatial scale on the choice of immigrants to partly depend on partisanship. Specifically, we expect that both Democrats and Republicans will be less likely to prefer admitting immigrants posited to move into their neighborhoods, but only Democrats will exhibit a preference for immigrants at more abstract spatial scales. We expect Republicans to be indifferent to all other abstract spatial scales of an immigrant’s expected location (Hypothesis 2).
RESEARCH DESIGN
We examine whether providing survey respondents with information about an immigrant’s expected location—if admitted to the United States—impacts their decision about whom to admit.Footnote 4 To assess our hypotheses, we adopt a “least likely case design,” using two means to make our test more conservative. First, rather than designing a survey experiment focused solely on varying an immigrant’s expected location, we modify Hainmueller and Hopkins’ (Reference Hainmueller and Hopkins2015) seminal conjoint experiment using a nationally representative sample and include all attributes of the immigrant profiles they tested to limit survey experiment omitted variable bias (Dafoe, Zhang, and Caughey Reference Dafoe, Zhang and Caughey2015). This has the additional benefit of ensuring that any effect we detect is due only to the presence of our added attribute. Second, we place our novel attribute toward the middle of the immigrant profile to minimize any recency effect (Greene Reference Greene1986).
In our preregistered survey, we recruited 1,313 American citizens who were 18 years or older from July 16 to July 21, 2024, via the online survey platform Verasight. Each respondent is put in the position of an immigration officer and shown two immigrant profiles with randomly varied information on the following attributes: prior trips to the United States, reason for application, country of origin, language skills, profession, job experience, employment plans, education level, gender, and expected location.Footnote 5 Respondents are then asked, “If you had to choose between them, which of these two immigrants should be given priority to come to the United States to live?” To adequately assess whether immigration preferences are scale-dependent, we randomly vary information about the immigrant’s expected location across the levels “your neighborhood,” “your city,” “your state,” and “the United States.”Footnote 6 Each participant is asked to assess eight pairs of hypothetical immigrants. After removing observations with incomplete responses, we are left with 20,954 immigrant profiles nested within 1,311 respondents (Lee and Muttram Reference Lee and Muttram2025).
The nature of this design allows us to not only attain estimates of the NIMBY effect but also to assess its importance in supporting an immigrant for admission relative to other primary determinants of immigration attitudes. Additionally, employing a forced-choice conjoint design allows us to mimic the real-world considerations Americans are faced with. Unlike other conjoint designs that may allow respondents the ability to choose neither of the proposed immigrants (Vermeulen, Goos, and Vandebroek Reference Vermeulen, Goos and Vandebroek2008), migrants involved in transportation programs are often already granted asylum or legal protection. The only remaining decision is where they will go.
IDENTIFICATION STRATEGY
Our primary estimates of interest are marginal means, or the average probability of selecting an immigrant for admission given an attribute level, averaging over all other attributes.Footnote 7 Because marginal means are not defined relative to an arbitrary reference category, they provide an intuitive description of how an attribute level affects decision making and are the recommended estimates to report in measuring subgroup preferences (Leeper, Hobolt, and Tilley Reference Leeper, Hobolt and Tilley2020). To provide a causal interpretation of our results, we also report the average marginal component effects (AMCEs) of our main attribute levels, which we estimate from a linear probability model specified as:
$$ {\displaystyle \begin{array}{l}\hskip0em \Pr \left({\mathrm{Selected}}_{\mathrm{i}\mathrm{jk}}=1\right)={\unicode{x03B8}}_0+{\unicode{x03B8}}_1\left[{\mathrm{Location}}_{\mathrm{i}\mathrm{jk}}=\mathrm{Neighborhood}\right]\\ {}\hskip13.5em +{\unicode{x03B8}}_2\left[{\mathrm{Location}}_{\mathrm{i}\mathrm{jk}}\hskip-0.2em =\hskip-0.2em \mathrm{City}\right]\hskip-0.2em +{\unicode{x03B8}}_3\left[{\mathrm{Location}}_{\mathrm{i}\mathrm{jk}}\hskip-0.2em =\hskip-0.2em \mathrm{State}\right]\\ {}\hskip13.5em +{{\mathrm{Z}}^{\prime}}_{\mathrm{i}\mathrm{jk}}\Psi +{\mathrm{u}}_{\mathrm{i}}\end{array}} $$
for respondent i, profile j, and choice tasks k, where the dependent variable takes on a value of 1 if an immigrant was preferred for admission and 0 otherwise. The baseline category for our variable of interest is set to “the United States,” Z′ijkΨ denotes all other attributes from the experiment and a vector of their respective coefficients, and ui denotes the individual-level residuals. All standard errors are clustered by respondent ID.Footnote 8
EXISTENCE AND RELATIVE MAGNITUDE OF THE NIMBY EFFECT
Figure 1 presents the marginal means of each attribute level of our expected location variable. If the marginal means are above (below) 0.5, this indicates that the attribute level increases (decreases) immigrant favorability. Put differently, if the 95% confidence intervals overlap with the dotted line at 0.5, we can infer that the information provided led respondents to act no differently than they would if they selected the immigrant profile at random.Footnote 9
Average Probability of Immigrant Admission by Location
As expected, providing respondents with information about the expected location of the immigrant impacts their decision of whether to admit immigrants to the United States. Moreover, this information appears to matter only for decision-making at the neighborhood level, consistent with the nonlinear expectations from NIMBYism. When respondents are forced to think about a potential immigrant in terms of their own community, they are less likely to select this immigrant for admission. Their considerations, however, remain consistent across all other “abstract” spatial scales. Putting this in causal terms, changing an immigrant’s expected location from the United States to the respondent’s neighborhood decreases the probability of selection by 2 percentage points (θ1=−0.02, p<0.05).
Although the data suggest that spatial considerations alter respondents’ propensity to select immigrants for admission, this effect is modest in size. For reference, θ1 represents an effect around 10% of the magnitude of our largest retrievable AMCE—the difference between an immigrant having a contract with a US employer and having no plans to look for work (θ=0.21)—and is comparable in size to the preference respondents exhibit for female immigrants.
Figure 2 presents the marginal means for each attribute level in the design. Two items are noteworthy. First, as the magnitude of other attribute levels indicates, while the effect we recover is statistically significant, it is not nearly the primary determinant in immigrant selection. Rather, an immigrant’s prior visits to the United States, economic and educational status, job experience, work plans, and country of origin appear to matter a great deal more than where the immigrant will locate.
Average Probability of Immigrant Admission Across All Attribute Levels
Second, the remainder of our study largely replicates the findings of prior iterations of this design (Hainmueller and Hopkins Reference Hainmueller and Hopkins2015). While there are some key differences (e.g., Chinese immigrants face a notable degree of opposition; see Liao Reference Liao2023), this suggests that the determinants of American immigrant preferences remain largely unchanged almost a decade after Hainmueller and Hopkins’ (Reference Hainmueller and Hopkins2015) seminal work. This is particularly notable, as our survey was fielded in July 2024. Thus, even in the lead-up to a presidential election in which immigration was a “lightning-rod” issue, we find remarkable consistency in what impacts Americans’ evaluation of immigrants.
To further illustrate the first point, figure 3 presents predicted probabilities varying only the expected location for what the literature on immigration attitudes would suggest is an “ideal immigrant” and an “unwelcome” immigrant.Footnote 10
Predicted Probability of the Most- and Least-Favored Profile Varying Location
Although the size of the NIMBY effect remains the same, it is far from changing respondents’ decisions to admit these hypothetical immigrants.Footnote 11 The effect of expected location is quite minimal, as evidenced by the small difference in preference for admission as shown in figure 3. In other words, the impact of the NIMBY effect on considerations is negligible in comparison to other factors well documented by the literature (Hainmueller and Hopkins Reference Hainmueller and Hopkins2014). Thus, in line with our expectations, immigration preferences do appear to be scale-dependent, and the average respondent only appears to alter their considerations when their own livelihood is immediately affected. However, this effect is substantively small when compared to the effect of other individual-level characteristics of the immigrant.
DOES THE NIMBY EFFECT VARY BY PARTISANSHIP?
Finally, we assess claims of “liberal hypocrisy” or whether the magnitude or presence of the NIMBY effect varies by respondent partisanship.Footnote 12 If our expectations are correct, we will see both groups of partisans opposing immigrants in their own backyard but only Democrats supporting immigrants in the abstract.
Marginal means, as illustrated in figure 4, reveal remarkably similar assessments of immigrants based on their expected location between Democrats and Republicans. Members of both parties exhibit a small preference against immigrants expected to locate in their own neighborhoods and are indifferent to all other spatial scales. Despite our expectations, we find no evidence that Democrats see immigrants at a more abstract spatial scale as preferable, in line with other work documenting a “hidden immigration consensus” across partisanship (Hainmueller and Hopkins Reference Hainmueller and Hopkins2015). Consistent with the justification for migrant transportation programs then, we find that respondents on average oppose immigrants when they are expected to be located in their neighborhoods. Counter to this justification, however, we fail to uncover evidence that this is simply “liberal hypocrisy.” Across all spatial scales, partisans hold similar preferences.Footnote 13
Average Probability of Immigrant Admission by Location and Respondent’s Party Identification
DISCUSSION
The 2024 presidential election provided the Republican Party with a mandate to pursue its agenda on immigration. In the first week of the 47th president’s term alone, a slew of executive directives reshaped immigration policies, ranging from efforts to end constitutionally guaranteed birthright citizenship to suspending refugee admissions and eliminating the possibility of asylum grants (Montanaro et al. Reference Montanaro, Gatti, Domonoske and Hsu2025). Perhaps more consequentially for those already in the United States, the new administration has increased the number of Immigration and Customs Enforcement (ICE) raids. This has led to mass stops and arrests of both undocumented individuals and American citizens who are targeted based on perceived or ascribed characteristics (Gamboa and Acevedo Reference Gamboa and Acevedo2025). Given the sweeping unilateral actions affecting immigrant communities, it is vital to understand the microfoundations of public attitudes toward immigration.
Our study makes two key contributions to the literature on immigration preferences. First, we provide the first causal evidence of the NIMBY effect on natives’ preferences for immigrants, expanding the application of NIMBYism to a new domain of public opinion research. Unlike previous studies that examine immigration attitudes without specifying where an immigrant is expected to settle (Hainmueller and Hopkins Reference Hainmueller and Hopkins2015), our study directly assesses whether such information alters native citizens’ evaluations of immigrants in line with the expectations of NIMBYism. On average, natives oppose the admission of immigrants who are expected to settle in their own neighborhoods. Whereas previous research suggests that exposure to immigrants may positively shape attitudes, our findings indicate that individuals may seek to avoid such exposure ex-ante (Rothbart and John Reference Rothbart and John1993, 42). Importantly, however, we find that the extent to which NIMBYism changes individuals’ overall stance toward an immigrant is limited compared to other individual-level attributes of an immigrant.
Second, we directly test the logic underpinning the rhetoric used to justify migrant transportation programs in several Republican-led states. Contrary to expectations and the language employed by Republican elites, we find no partisan differences in the NIMBY effect. Although spatial considerations matter, native support for immigrants appears to be more strongly driven by the socioeconomic and cultural background of immigrants than by Americans’ partisanship or neighborhood-specific concerns. Justifications for these programs appear ill-founded.
Although spatial considerations matter, native support for immigrants appears to be more strongly driven by the socioeconomic and cultural background of immigrants than by Americans’ partisanship or neighborhood-specific concerns.
Although we document the existence of the NIMBY effect, one limitation of our study is its inability to directly explain why NIMBYism is occurring. Prior research suggests that NIMBYism may stem from concerns over property values, prejudice, personal security, or neighborhood quality and aesthetics (Dear Reference Dear1992). While we cannot adjudicate between these explanations, we note that we document a NIMBY effect even in the presence of individual-level attributes that address both the cultural (i.e., country of origin and English proficiency) and economic (i.e., profession, job experience, and work plans) dimensions. This points to the possibility that even if observed NIMBYism is a combination of these concerns, it may also merely reflect status quo bias (Samuelson and Zeckhauser Reference Samuelson and Zeckhauser1988) or risk aversion (Fischel Reference Fischel2001). Regardless of who may move into the neighborhood, people may simply have a small preference to keep their community the way it is.Footnote 14
We do not expect NIMBYism toward immigrants to be exhibited only among the American public, yet we have limited theoretical expectations about the extent to which it is prevalent in other developed democracies. The findings of our study are also insufficient to make any causal claims outside of the U.S. context. However, it is likely that NIMBYism in immigration attitudes is not specific to the United States because its core idea—that individuals’ support for policies can vary by spatial scale—is applicable more broadly. Observations from existing literature that the underpinnings of immigration attitudes show similar patterns in North America (Harell et al. Reference Harell, Soroka, Iyengar and Valentino2012) and Western Europe (Citrin and Sides Reference Citrin and Sides2008) provide another reason to believe that NIMBYism in immigration attitudes may be observable outside of the United States. Conversely, we also recognize that specific political contexts may limit the generalizability of the findings of our study in contexts outside of the United States. It may be that the political environment, in which political parties take distinct policy positions and mobilize voters on immigration, affects the applicability of NIMBY theory in explaining immigration attitudes. This is likely where future research can better specify the scope conditions of this theory.
Immigration is one area in which the NIMBY phenomenon is a concern for proponents of new public policy, but it is surely not the only one. The NIMBY effect is commonly invoked in discussions of redistribution and attitudes toward the unhoused, for example, with the intuition that people may support policies that aid those facing housing insecurity so long as they do not impact one’s own well-being. Extrapolating our results to this setting would suggest it is the individual-level attributes of the unhoused person (e.g., prior job experience, work ethic, or education level) that explain more of the variation in attitude formation toward the individual than the potential for whether and how much one interacts with this person in their community. Despite NIMBYism being viewed as a barrier to collective action (Foster and Warren Reference Foster and Warren2022), our results suggest that opposition to land-use policies may be better explained by factors beyond physical proximity.
Supplementary material
To view supplementary material for this article, please visit http://doi.org/10.1017/S1049096525101467.
ACKNOWLEDGMENTS
The authors are especially grateful to Jenn Merolla and the UCR Identity and Politics Lab for support. We also thank Dan Biggers, Ben Bishin, Miguel Carreras, Kim Yi Dionne, Steven Liao, Ben Newman, Nick Weller, the participants of the 2025 Southern California Political Behavior Conference, the UCR Mass Behavior Workshop, and the UCR Political Science Quantitative Methods Group for comments and constructive feedback.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the PS: Political Science & Politics Harvard Dataverse at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/6TSZXR&version=DRAFT.
CONFLICTS OF INTEREST
The authors declare that there are no ethical issues or conflicts of interest in this research.
Open access
