Abstract
Does living in a sanctuary city moderate the relationship between fear of deportation and immigrants’ willingness to give out personal information? While fear's impacts on health are well-documented, few studies examine its effects on behavior. Using a survey of approximately 1,500 undocumented Mexican nationals, we find that immigrants with lower fear levels report greater comfort providing personal information in everyday occurrences, including going to restaurants, police encounters, or visiting hospitals, than those who report more fear. This relationship is moderated by sanctuary status: immigrants who believe they live in sanctuary cities are more comfortable sharing information. By linking survey data with sanctuary city information, we offer a novel contribution to the study of domestic immigration enforcement—although we note that these findings may not translate into the Trump 2.0 era. Our findings suggest that policies reducing fear—such as sanctuary city protections—can play a meaningful role in fostering immigrant integration.
Introduction
Immigrants play an important role in cities’ economies and social fabrics (Zhou 2004; Card 2009). Despite this, many undocumented immigrants live in fear of being deported, making it hard for them to be full social and economic participants in the places where they live. While some states and local jurisdictions have worked to mollify the undocumented population's concerns (Walker and Leitner 2011; De Graauw and Bloemraad 2017; Collingwood and O’Brien 2019), nevertheless, many immigrants may be scared to open bank accounts, seek medical care for serious injuries, cooperate with law enforcement, or even go to the doctor when they are sick—thereby compromising their ability to work, feed their children, and acculturate into American society. This is concerning because immigrants contribute massively to the success of the U.S. economy, whether owning small businesses (Singh and Sharma 2013), working in construction (Siniavskaia 2015), working in jobs that native-born Americans do not want to do (Sherman et al. 2019), or helping to protect the future of social security through their contributions (Lee and Miller 2000). In other words, the more immigrants integrate, the better it is for the American economy.
In an effort to reduce the deterrent effects of deportation threats on social participation, local jurisdictions over the past forty years—most notably sanctuary and welcoming cities—have implemented policies intended to incorporate immigrants into local communities, though these efforts have varied in scope and effectiveness (Walker and Leitner 2011; De Graauw 2014; Huang and Liu 2018). Specifically, sanctuary cities place limits on cooperation between local police departments and Immigration and Customs Enforcement (ICE), and more often than not prevent local officials from inquiring into immigrants’ status (O’Brien et al. 2025). Yet, due to the difficulty of surveying the undocumented population, researchers know relatively little about whether immigrant-friendly policies help to reduce individual-level immigrant fear.
This paper addresses two questions. First, does fear of deportation shape undocumented immigrants’ daily interactions and limit their willingness to disclose personal information? Second, does fear of deportation interact with immigrants’ beliefs about residing in a sanctuary city to affect their willingness to disclose personal information? 1 Research shows that undocumented immigrants living in the United States who fear they, or a close family member, could be deported are less willing to cooperate with law enforcement (Becerra et al. 2017). However, we know relatively little about whether immigrant incorporation policies (e.g., sanctuary policy or municipal identification cards) 2 influence immigrants’ willingness to give out personal information. The authors found only one study showing that immigrants who believe they live in a sanctuary jurisdiction are more willing to cooperate with law enforcement (Garrick and Schrank 2024), and a second one that draws a similar conclusion using experimental primes (Wong et al. 2021). However, we do not know whether an immigrant's fear of deportation interacts with sanctuary city residence to impact an immigrant's personal comfort when giving out personal information.
Having answers to these research questions is essential because sanctuary policy justifications often rest on the argument that such policies help incorporate immigrant populations into society. The incorporation of immigrants is vital for a variety of reasons, including stabilizing and improving local economies (Lee and Miller 2000), strengthening labor markets, sustaining the solvency of Social Security, and generating a coherent national ideology (Jones-Correa 2005; Citrin and Sears 2014). Finally, immigration and sanctuary advocates often claim that sanctuary policy is implemented for crime-fighting reasons. In these jurisdictions, presumably immigrant populations are more likely to work with the police—by reporting crimes and even showing up to court proceedings, which entails the production of personal information. Although these explanations are plausible, relatively little research has examined whether undocumented immigrants are more willing to provide information to the police and other entities in sanctuary jurisdictions.
We evaluate our research questions using a novel survey dataset of approximately 1,500 Mexican nationals residing in the United States. The sampling strategy entailed approaching potential respondents waiting in line at Mexican consular offices in several cities across the United States. By using a series of four screening questions—which we outline in the data/methods section below—the survey includes only respondents who are almost certainly undocumented people living in the United States. We find that immigrants who fear they could be deported report that they are less willing to give out personal information to police, schools, banks, hotels, shops/restaurants, and hospitals than other undocumented individuals. This data, even though it was collected during the Biden administration, can help us understand the new immigration reality of heightened enforcement under the second Trump administration.
To our knowledge, no research to date has found these discrete effects of fear of deportation on such a wide range of social settings. We show that people who believe they live in sanctuary cities are more willing to share personal information across a variety of contexts. Our findings indicate that these policies positively affect the immigrant population's incorporation into American society. Finally, we demonstrate that immigrants with modest fear who live in sanctuary cities are significantly more willing to give out personal information compared to immigrants with modest fear who do not live in a sanctuary city. Our results, therefore, are consistent with justifications for sanctuary in the first place: they do help elicit social and economic integration among the undocumented population.
Background and Theory
Our background section is split into two sub-sections. First, we review the literature on how fear of deportation affects Latinos and undocumented individual's cooperation with officials and participation in civil society. The second section reviews relevant literature on how sanctuary policy might ameliorate some of these negative effects attributed to fear of deportation. These two sub-sections lead into a hypothesis section where we formally lay out our expectations.
Fear of Deportation
Much of what we know about the fear of deportation and undocumented individuals’ willingness to work with authorities is centered around the Latino population in the United States and their experience with or attitudes toward the police. This is because, over the past forty years, Latinos have been widely associated with the immigration experience, and many of them live in mixed-status households (Vargas et al. 2019). Moreover, research often examines attitudes toward the police because, in some locations, this is the most likely scenario where an immigrant could be caught up in detention and deportation proceedings. Given the existing research, our review of the literature begins with an overview of Latino attitudes toward the police, then narrows its focus to immigrants’ perceptions of law enforcement, and later examines how immigration enforcement undermines Latinos’ trust in the police—especially among those who are undocumented. Finally, we include sections on other institutions evaluated in this paper and the general impacts of social and economic isolation on the immigrant experience.
Past research has observed that native-born Latinos are less likely to trust the police compared to foreign-born Latinos. For instance, Barrick (2014) found that native-born Latinos had less trust that police would enforce the law. In addition, Davis and Hendricks (2007) and Correia (2010) found that immigrants rated police more positively than did native-born Americans. Similarly, Martinez and Alamillo (2024) found that immigrants are more likely to favorably evaluate their local police compared to native-born Latinos. However, there are instances where U.S.-born and immigrant Latinos have similar feelings toward police. A key study, Rocha, Knoll and Wrinkle (2015), found that immigration enforcement harmed both native and foreign-born Latinos, as there was a spillover effect, given that native-born Latinos were also stopped and harassed by officials. In sum, foreign-born Latinos are more likely to trust the police, although some conditions (i.e., periods of heightened immigration enforcement) may cause greater hesitancy to report an incident to the authorities.
As we discussed, immigrants, on average, have more trust in the police than native-born Latinos. Although this is not always the case, a critical factor for why some immigrants may not trust the police is related to their immigration status. When encountering police, non-citizen immigrants may feel threatened if questioned or harassed. If immigrants are questioned about their immigration status, they are probably less likely to trust and report to the police due to personal safety concerns about their status. Being questioned and harassed by immigration enforcement has stoked fears of visa revocation and deportation and led to lower rates of crime reporting (Rodriguez et al. 2018; Dhingra, Kilborn and Woldemikael 2022). When studying the experiences of undocumented immigrants with immigration officials, Becerra et al. (2017) found that when immigrants were asked for documents, they reported fear of deportation and had less confidence that police would treat Latino immigrants fairly. Additionally, Theodore and Habans (2016) found that increasing police involvement in immigration enforcement led to negative attitudes about the police among Latinos. Therefore, when Latino immigration status is threatened by police presence and questions about immigration status become more likely, Latino trust in the police is reduced.
We have discussed how fear of deportation among undocumented individuals impacts the provision of information to the police. Research also demonstrates a relationship between fear of deportation and both healthcare hesitancy and reluctance to provide information to school officials and banks. Studies show that undocumented Latino immigrants may avoid health services due to fear of deportation (Rhodes et al. 2015). The threat of immigration enforcement may also have collateral consequences for immigrants who have never had contact with the system, leading immigrants and their U.S.-citizen children to avoid health-promoting institutions to minimize their risk of detection, detention, and deportation (Bean, Telles and Lowell 1987). Another study by Watson (2014) finds that long-term permanent residents respond to the enactment of punitive immigration policies by not enrolling in Medicaid. As harmful policies have been observed to have a negative impact on using health care services, states that have policies that limit federal immigration enforcement have been associated with improved access to child preventative care (Koball, Kirby and Hartig 2022). Therefore, past literature shows that undocumented immigrants are hesitant to seek medical services because of fear of deportation.
This fear of deportation and willingness to give out information is not only limited to police and healthcare interactions but also applies to educational settings. Research shows that undocumented parents fear school officials may disclose personal information to immigration authorities (Gonzales, Heredia and Negrón-Gonzales 2015); therefore, it is reasonable to think that fear of deportation affects all sorts of immigrant interactions with schools. Due to fear that school officials will reveal student and parent information to immigration authorities, undocumented parents may assume that there are no resources for undocumented students (Lad and Braganza 2013) and undocumented college students (Seo 2010).
Also with respect to education, research shows that anti-immigrant state laws and immigration raids increases student absenteeism and worsens educational outcomes (Mussey 2014; Mohl 2016). Punitive immigration laws and raids may make parents less likely to provide information to schools. Bucheli, Rubalcaba and Vargas (2021) find that a greater number of immigration arrests in a particular area corresponds to greater school absenteeism in that same area. In addition to student absenteeism, Kirksey and Sattin-Bajaj (2023) find that reading and math test scores declined as a result of enforcement activities, and there were sharp increases in the number of students leaving their school districts.
Therefore, as research shows, undocumented immigrants have kept their children from going to school because of fear of deportation. It follows that undocumented immigrants with fear of deportation will be less likely to share personal information to schools.
Besides interaction with police, schools, and health care institutions, immigration status and fear of deportation may influence immigrants’ interactions with hotels and restaurants/shops—as undocumented immigrants both shop and work in these local businesses. In terms of shopping or eating at a restaurant, while it is typical not to have to give out personal information like a phone number or use photo identification, it is not out of the question that someone will need to provide their name, telephone number, and address to book a reservation at a restaurant, for example. There may be other scenarios in other shops where a person may want to buy an item but it is on hold, and so the shop would ask for a person's phone number, and so forth, to contact them when the item is back in stock. So it is possible that a person will be asked to give out personal information in these daily interactions.
Further, by their sheer numbers, immigrants play a significant role in hotel and restaurant industries (Kim 2012). Immigrants with various documentation statuses are highly integrated in these sectors (Sarathy and Casanova 2008). Therefore, immigrants working in hotels and restaurants hire family and friends, who may be undocumented immigrants (Varzally 2023). Hotels and restaurants are labor-intensive, and to meet this demand, employers are willing to pay their employees in cash, which does not require them to know their employees’ immigration status (Gleeson 2010). Because immigrants play a key role in these businesses, immigration enforcement may target hotels and restaurants. Moreover, if hotels and restaurants are asking workers who may be undocumented immigrants for their personal information, the immigrants may be less likely to share this information because of the fear of a workplace raid (Corral 2023).
In the context of banking, undocumented immigrants are often unable to open bank accounts because they lack tax identification or Social Security numbers. As a result, interactions with financial institutions may be perceived as risky, discouraging undocumented immigrants from disclosing personal information (Handlin, Krontoft and Testa 2002). This reluctance is particularly consequential given that undocumented immigrants are more likely to be victims of crime (Zatz and Smith 2012) and are less likely than native-born individuals to report victimization to authorities (Khashu 2009). Elevated victimization rates stem from several factors, including limited access to formal legal protections such as the criminal justice system (Zatz and Smith 2012), their reliance on cash incentive employment, and their fear of deportation (Caraballo and Topalli 2023). Taken together, these dynamics suggest that heightened immigration enforcement may produce spillover effects beyond law enforcement, reducing undocumented immigrants’ willingness to disclose personal information to banks and other formal institutions.
More generally, the willingness of immigrants to share personal data with public and private institutions is not just an administrative concern; it is a barometer for feelings of belonging. Scholars of immigrant incorporation have shown that everyday interactions with local and federal authorities are central to how immigrants come to see themselves as part of, or apart from, the community as a whole. As Armenta (2017) argues in Protect, Serve, and Deport, the entanglement of local law enforcement with federal immigration enforcement creates an atmosphere of uncertainty that extends far beyond policing. When any routine encounter can lead to detention or deportation, immigrants learn to avoid contact with authorities altogether. This withdrawal disrupts access to essential services, participation in civic life, and even interpersonal trust, which are key components of incorporation.
At the municipal level, scholars such as De Graauw and Vermeulen (2016) and Provine et al. (2016) have shown how local contexts mediate this trust. De Graauw and Vermeulen argue that cities are critical arenas of immigrant incorporation because immigrants live, work, and interact with government primarily at the local level. Progressive city governments and robust community organizations can translate immigrant presence into sanctuary policies, while more restrictive localities reinforce exclusionary practices. Provine et al. (2016) highlight the moral and institutional contradictions facing local law enforcement—caught between community policing mandates of building trust and federal enforcement expectations. This jurisdictional tug-of-war constructs a patchwork of belonging, where safety depends on one's location and often changes as immigrants move between cities in their daily lives.
Wells (2004) traces this structural inconsistency to the very architecture of U.S. governance: a decentralized system where overlapping agencies pursue conflicting goals and norms regarding immigrant rights. This, in addition to the common law tradition, results in a status quo in which local actors can reinterpret or even resist federal policy. In this context, the sharing of personal data becomes an act conditioned by geography, legality, institutional trust, and federal context. It reflects not only an immigrant's relationship with the state but also the uneven terrain of immigrant incorporation itself. By analyzing the impact of sanctuary policies on immigrants’ willingness to share personal information with various institutions, we are advancing the discussion on whether these policies help immigrants interact with their communities, leading to incorporation.
Taking the above-presented findings into consideration, we expect that Mexican undocumented immigrants who are fearful of deportation will be much more hesitant in giving out personal information not only to police but also in many other scenarios than will Mexican undocumented immigrants who are less fearful.
Sanctuary Cities
Undocumented status in the United States implies uncertainty for people, given that they face prosecution by immigration enforcement in a context where policies are constantly shifting, jurisdictions overlap, and the possibility of deportation is continually looming over their shoulders. This uncertainty is also accompanied by variation in immigration enforcement. Immigration enforcement regimes comprise an array of local, state, and federal policies that often differ depending on the county or even government agency involved (Walker and Leitner 2011; Bruce 2021). In the face of immigration federalism (Gulasekaram and Ramakrishnan 2015), sanctuary cities seek to provide some safety from deportation for their undocumented residents.
One of this paper's research questions is whether sanctuary cities mitigate the robust negative relationship between fear of deportation and willingness to give out personal information. A sanctuary city is a term that has its origin in movements that sought to provide a reprieve from prosecution to people fleeing civil wars in Central America during the 1980s, led by religious groups (Caminero-Santangelo 2012). One of the main demands of this movement was the opening of legal pathways for people fleeing violence to establish themselves permanently in the United States (Bruce 2021). Today, sanctuary refers to policies implemented by local and state governments as a counterbalance to the generalized prosecution of undocumented people (Bruce 2021). For the most part, sanctuary policies aim to provide undocumented people protections from deportation. Recently, O’Brien et al. (2025) examined nearly 400 sanctuary city policies by coding each city on a 7-point scale—noting that most cities have bans on inquiries into immigration status, and all have non-cooperation clauses vis-à-vis ICE. A smaller number of cities also limit ICE access to jails, ICE notifications about immigrant release dates, and ban 287-G agreements between police and ICE. All told, it makes logical sense that immigrants who think they live in a sanctuary city will perhaps be a bit more comfortable giving out personal information because they know that their information is less likely to result in their deportation. In this way, a sanctuary policy will serve one of its purposes, which is helping to incorporate local immigrant populations.
Other scholars understand these sanctuary policies as a collection of solidary practices rather than a unified agenda. Czajka (2012) defines sanctuary as “heterogeneous, as enacted by an array of actors with differing interests and objectives, and as often fiercely contested.” Sanctuary spaces provide an opportunity for non-state actors (such as civil society, consular offices, and religious organizations) to confront the state's ideas of belonging, residency, and citizenship (Czajka 2012). The New Sanctuary Movement also includes local governments and state governors ordering local or state police to cease cooperation with Immigration Enforcement, as well as implementing programs directed at generating trust between government institutions and undocumented populations—including not asking people about their immigration status. The country has between 300 and 560 sanctuary cities and states, according to academic and journalistic estimates (Dinan 2018; O’Brien, Collingwood and Paarlberg 2024). This complex interplay of policies is a fundamental part of the realities of immigration enforcement in the United States, one that undocumented people must deal with every day.
Definitions aside, researchers have found that sanctuary policies are associated with a variety of incorporating behaviors. For instance, Collingwood and O’Brien (2019) find that Latino political participation, as measured by voter turnout, increased in cities that became sanctuaries, and likewise, the police force became more Latino—albeit the effects are fairly small. Otsu (2021) finds that sanctuary policies may even be associated with lower crime rates, a finding that is generally consistent with research showing there is no relationship between crime and sanctuary policy (Martínez, Martínez-Schuldt and Cantor 2018; O’Brien, Collingwood and El-Khatib 2019). But research also shows the reverse is true: when Texas banned sanctuary policies, the very next day 9/11 calls in the high-density foreign-born neighborhoods of El Paso, TX, dropped significantly—and stayed that way for a month (Collingwood and O’Brien 2019). Using an experiment embedded within a survey of undocumented immigrants, Wong et al. (2021) find that respondents are less trusting of police when exposed to a treatment condition where the local sheriff works with ICE. In addition, Martinez-Schuldt and Martinez (2021) find that Latinos are more likely to report crime victimization in jurisdictions that have passed sanctuary policies. All of this is to say there are very good reasons to expect immigrants will be more likely to give out personal information if they believe they reside in a sanctuary city.
It is important to mention that the heightened federal enforcement and punitive reactions toward sanctuary policies of the second Trump administration interact with local sanctuary efforts. Besides actual enforcement targeting people with no criminal records, the high visibility of immigration operations in the current climate has very likely dampened the soothing effects of sanctuary policy, generating widespread fear among Latino and immigrant communities (Nittle 2025).
Hypotheses
The literature review and theoretical development, along with our original survey data, lead to a few testable hypotheses. First, we should anticipate that undocumented immigrants who are more personally concerned with deportation should, all else equal, be less willing to want to hand out personal information in a variety of venues (e.g., at schools and hospitals). The logic is clear: when a person is very worried about them or their family members being deported, they will want to reduce the amount of personal information they willingly give out so that the possibility of ICE/DHS discovering them is reduced. Therefore, we state the following hypothesis:
However, we believe that the relationship between fear of deportation and comfort in giving out personal information is moderated by whether someone lives in a sanctuary city or not. Sanctuary cities have gained considerable attention since the first Trump administration, with a large increase in sanctuary jurisdictions around the United States. Therefore, it is likely that more and more people are aware of what a sanctuary is and whether they live in one. As Oskooii, Dreier and Collingwood (2018) show, knowledge of what a sanctuary is and whether someone lives in one can be crucial to behavior inside a sanctuary jurisdiction. It makes sense then that living in a sanctuary will make those fearful of deportation a bit more willing to give out personal information because those cities are broadly more immigrant-friendly environments. However, we expect this relationship to manifest primarily among individuals at the lower end of the fear scale, as those experiencing extremely high levels of deportation fear are likely to exhibit generalized or pervasive paranoia.
Methods
To assess our research questions and hypotheses, we rely on a survey conducted among a very hard-to-reach population: undocumented Latino immigrants residing in the United States. 3 In this section, we discuss this survey, our key variables, and our statistical approach.
Sanctuary Cities as Emerging Borders Project Survey:
One of the major challenges when assessing the effects of sanctuary city policies on undocumented people's behavior is the difficulty of gathering data from immigrants themselves. Undocumented Mexican immigrants have various characteristics that make them a hard-to-survey (H2S) population. First, identification: they are immigrants and, as such, are difficult to detect and sample, given that they tend to avoid disclosing their status and are hard to distinguish from Mexican Americans and Mexican immigrants with authorization (Massey 2014). Second, given their undocumented status, they are vulnerable to state prosecution and often avoid going outside or talking to strangers (Lyberg et al. 2014). Third, they are Mexicans, a group that is also vulnerable to discrimination regardless of their immigration status because they are often assumed to be undocumented, given that the image of an undocumented immigrant is prototypically Mexican, as will be further explained below (Pearson 2010). According to Massey, who writes in 2014, this has made Mexicans increasingly harder to survey (Massey and Brodmann 2014). Due to these characteristics and the stigma associated with being undocumented, it might be disrespectful or off-putting to ask people directly about their immigration status. Because of this, the survey used employs indirect questions, as detailed below.
Mexicans in the United States constitute a profoundly complex and heterogeneous group, encompassing both long-established communities with centuries-old roots in the American Southwest and more recent arrivals, such as the immigrants examined in this study. Regardless of their actual citizenship status, Mexicans in the United States are burdened by enduring cultural and political tropes that equate Mexican identity with illegality and foreignness. In the American social imagination, the Mexican has become the quintessential immigrant laborer (López 2017), a figure simultaneously necessary and stigmatized. This association has profoundly shaped the experiences of people of Mexican origin, especially those who live in conditions of irregular status. Across U.S. history, these tropes have been strategically mobilized to generate anti-immigrant sentiment; from Eisenhower's Operation Wetback in the 1950s to California Governor Pete Wilson's anti-immigrant propositions in the 1990s, and later, to Donald Trump's three presidential campaigns, where the same imagery and fears were once again invoked and amplified. These challenges are particularly acute for non-Latino research teams, given that undocumented people will be suspicious of anyone who could be working with the government.
The “Ciudades Santuario como fronteras emergentes” survey avoids these obstacles through its design and its quality of being Mexican-run. It uses venue-based sampling, using consulates as locations to find respondents, and filtering questions shown in Table 1 to determine if a person was undocumented or not, methods often used for studying H2S populations (Berry and Gunn 2014). Compared to studies of undocumented people based on approximations derived from general population datasets (which often have small samples and questions designed for a general population), this survey is unique in its large N and approximation to a probability sample through venue-based sampling and indirect filtering.
Questions and Expected Answers for Identifying Documentation Status.
The “Sanctuary Cities Project as Emerging Borders” survey was designed and administered by El Colegio de la Frontera Norte (COLEF) and the Consejo Nacional de Humanidades, Ciencia y Tecnología (CONAHCYT). The research team, including one of this paper's authors, conducted an in-person survey in 2022 and 2023 with immigrants in six important U.S. metropolitan areas: Los Angeles, Chicago, New York, Dallas, Phoenix, and Atlanta. These locations have different contexts for implementing immigration policy and sanctuary efforts, varying across different city jurisdictions in these multi-county metropolitan areas.
This CONAHCYT-supported project focused on interviewing undocumented Mexican immigrants about various issues. After coordinating with consular authorities, the COLEF team approached respondents in and around Mexican Consulates. The survey was administered mainly in Spanish, but English was also available. The target sample consisted of people who had been in the United States for three years or more and were over the age of eighteen years. The survey excluded people with DACA protections because their level of documentation puts them in a different position from entirely undocumented people. The survey is unique in that it is an effort led by a Mexican institution in the context of the United States, taking advantage of the Mexican consular network in six major cities to capture a near-verified sample of undocumented immigrants.
The survey researchers gathered data in two ways. First, interviewers approached potential respondents while waiting for consular services; second, some were contacted independently and surveyed in separate locations. The survey workers were all bilingual Mexican or American-Latino graduate students. When the survey worker arrived at the location, all the people waiting for their documents to be processed at the consulate were approached.
For the reasons stated above, the questionnaire did not ask the immigration status of the respondent directly, but was estimated through a series of indirect questions, as is regular practice with H2S populations (Berry and Gunn 2014). The rationale behind these questions is explained in the following table. Respondents were included in the analytic sample only if they answered a majority of the screening questions in ways that aligned with the research team's expectations for undocumented respondents. 4
Although multiple ways of asking the questions were evaluated, these four items reflect an effort to minimize the risk of alienating potential survey participants by not asking directly about their status but instead allowing the survey workers to make an informed assumption about their inclusion in the sample. The logic for each question, as described above, also reflects this. The analysis presented in this article uses the full sample, made up only of people who were found to be undocumented by their responses to the filter questions detailed in Table 1.
Independent and Dependent Variables
One of the survey's questions asked about the respondent's fear of deportation. The question asked is: “Currently, are you afraid of being detained or deported?” The response options are: a lot (4), some (3), a little (2), or none (1). The distribution of our key independent variable is displayed in Figure 1.
Fear of deportation distribution.
About 47 percent of respondents said they had “no concern” about deportation, whereas 25 percent said they did have “a little concern,” 18 percent “some concern,” and 10 percent “a lot of concern.” To be sure, the highest category of respondents falls into the “none” category, still more than half of the respondents exhibit some concern about deportation. 5 It is likely that these results are modest, given that most people would probably hesitate to tell a stranger that they feel scared. This may be particularly salient among men, due to gendered expectations among Mexican men to not show their emotions. Indeed, the correlation between fear of deportation and sex is 0.15, meaning that women report higher levels of deportation fear. This is in line with research by Castañeda Pérez (2022), which shows that border crossers’ emotional experience with immigration authorities varies according to their positionality. Still, the variable provides a clear measure of deportation fear.
Regarding our moderating variable, whether a respondent thinks they live in a sanctuary city, we use the following question:¿El lugar donde vive es una ciudad santuario? (Is the place where you live a sanctuary city?)
Figure 2 displays the variable's distribution. Forty-eight percent of respondents say they live in a sanctuary city, 37 percent say they do not know, and 15 percent say they do not. In terms of coding, we code anyone who says they live in a sanctuary city a 1, don’t know (.5), and no (0). 6
Live in a sanctuary city distribution.
We use knowledge of whether someone lives in a sanctuary city vs. whether they live in a sanctuary for a few reasons. First, if someone thinks they live in a sanctuary city, regardless of whether that is true, it makes sense they will act as though they live in one—which is likely they will be at least marginally more trusting toward government officials and institutions. And the converse is likely the case. Beyond this, many people report not knowing whether they live in a sanctuary, and so if they do not know they live in one even if they do live in one, they are likely to act more cautiously.
Table 2 reports the frequencies for knowledge of living in a sanctuary city by consulate location. It is important to keep in mind that the respondents do not necessarily live in the consulate city—but possibly in neighboring cities in the corresponding metropolitan area. The table reports the results for each consulate location (e.g., New York, Chicago, and Los Angeles). Strengthening the validity of the data, respondents taking the survey at consulates located in sanctuary cities disproportionately say they live in a sanctuary city. For instance, 74 percent of respondents interviewed in New York say they live in a sanctuary city, whereas just 6.5 percent say they do not know. This makes sense because New York is the largest city in the United States and is often referenced when sanctuary cities are discussed in the media. Nonetheless, this table shows us that we cannot simply assume someone lives in a sanctuary city based on where the interview took place. For this reason, we use the item asking specifically about whether a respondent lives in a sanctuary.
Knowledge of Sanctuary by Mexican Consulate Interview Location.
In terms of control variables, we include measures for the following variables: level of education, economic situation, age, years living in the United States, female, whether the respondent has a local or state ID, whether the respondent has children living in the United States, whether the immigrant has a boss who has asked about their immigration status, and English proficiency. These variables may affect the relationship between fear of deportation and immigrant willingness to give out personal information. For instance, people who speak very fluent English may be more willing to give out personal information, as might someone who has a higher level of education. People with U.S. citizen children may be more comfortable giving out information because they are used to having to provide it or have a greater family safety net. On the other hand, people in greater economic distress may be less willing to give out personal information and have a higher fear of deportation. Importantly, we include a variable capturing whether a person has a local or state (regional) id because some of our outcome measures (opening a bank account) do require someone to have an id. In descriptive statistics, women and people with economic needs tend to be more fearful of deportation. Education and years in the United States are included because they might be related to being better able to handle interactions due to experience and knowledge of U.S. society. Age is included to account for both life cycle and cohort effects, such as having been exposed to different national immigration regimens in the respondents’ lifetime. We list how these variables are coded in online Appendix C and include descriptive statistics about all variables in Table 6.
We include six separate dependent variables, asking respondents how comfortable they are in giving out their personal information in the following places: (1) police, (2) school, (3) bank, (4) hotel, (5) shops/restaurants, and (6) hospital. 7 Figure 3 presents the univariate distribution for each variable, showing some variation. For instance, respondents are much more willing/comfortable to give out personal information in situations/institutions where they are likely to have some ongoing personal relationship, such as in a school, a bank, or a hospital. These are venues where people essentially need to give their personal information regardless of their immigration status if they want access to these institutions. However, individuals feel less comfortable giving out personal information to the police, at hotels, and in shops/restaurants. At face value, these results are on track with our expectations.
Dependent variable distributions.
While these variables do not show the exact same distribution, we do believe they are tapping into a similar dimension. To ensure this is the case, we correlated all six variables against one another. Table 3 shows the results. All variables are fairly strongly correlated with one another, but there is clearly some variation. For example, shop and hotel correlate at 0.77, whereas police and school correlate at 0.55.
Comfortable Giving Out Personal Information Correlation Matrix.
Given that the variables are generally correlated, our primary dependent variable essentially scales these six discrete variables by adding responses together into an index ranging from 6 to 24. Figure 4 presents a histogram depiction of the variable's distribution. The mean of the distribution is 16.37. Given the distribution of this variable, we use linear regression as our main form of statistical analysis.
Dependent variable comfort scale histogram distribution.
Results
We begin our discussion of the results with separate regression analyses of each dependent variable. We then move to a discussion of our scale variable. We focus our interpretation just for Model 1 in Table 4, and note the interpretation follows the same logic for all other models presented in Table 4. Beginning with Model 1 (police) in Table 4, the coefficient is −0.211 and is statistically significant (p < .001). This means that as a respondent moves up the fear of deportation scale from low fear (1) to high fear (4), they are −0.211 points (on a 1–4 scale) less willing to give out personal information to police. Turning to post-estimated predicted values, for instance, if a respondent reports their fear of deportation at 1 (minimum), their willingness to give out information to police is a 2.8/4. If their fear of deportation is 4 (maximum), their willingness to give out personal information to police drops to 2.17 for a net effect of 0.63 points on the 4-point scale.
Comfort Level: Individual Scenario Models Base Models.
*p < .1; ***p < .05; ***p < 0.01.
With respect to living in a sanctuary city, the coefficient in Model 1 is 0.237 and is statistically significant (p < .001). If a respondent thinks they do not live in a sanctuary city (0), the predicted willingness to give out information to police is 2.45 on the 4-point scale, whereas if they think they do live in one this number rises to 2.69. To be sure, this effect is not substantively very large but is consistent with what sanctuary advocates would claim about sanctuary cities providing psychological safety to immigrants and hence enhancing cooperation with governmental authorities. That said, it is important to remember that these results were collected during Joe Biden's administration, so it is quite conceivable that any psychological benefits derived from sanctuary protections have been undermined during Trump 2.0.
Overall, this initial analysis shows that in each scenario, respondents who are more worried about deportation are less willing to provide personal information, as evidenced by negative and statistically significant coefficient estimates. These results provide initial support for Hypothesis 1. However, we also observe that living in a sanctuary city is associated with a willingness to give out personal information in five of the six individual scenarios. These results provide initial support for Hypothesis 2. Only giving out information to a bank is unrelated to whether someone lives in a sanctuary city. This likely has to do with the fact that in order to have a bank account, one is required to give out personal information, so living in a sanctuary makes no difference either way. We include individual-level scenario models with controls in Table A1 in the online Appendix A, and the results are substantively the same.
Table 5 presents the base models augmented with interaction terms. Here, we interact Fear of Deportation with Sanctuary City status, then estimate the combined effect of our two independent variables on our dependent variable. The key focus of this table is the Fear × Sanctuary term. A statistically significant effect is observed in four of the six dependent variables: police, school, bank, and hospital. These are the four variables where more people in general are willing to give out information (likely because these are relatively common scenarios where one would be expected to give out personal information), so it follows that this is where we might observe a differentiation on sanctuary status. Model 1 (police), for instance, shows a fear of deportation coefficient value of −0.128 (p < .001), lives in a sanctuary city coefficient value of 0.472 (p < .001), and an interaction term of −0.128 (p < .01). Translating to predicted values on the outcome measure, this means that respondents who think they live in a sanctuary city and score low on fear of deportation report a score of 2.92 out of 4 on willingness to give out information to police. Respondents who claim to live in a sanctuary city but have a high fear score 2.15 out of 4. Finally, people who have low fear and who do not live in a sanctuary city score 2.58 out of 4, whereas respondents who have high fear and do not live in a sanctuary city score 2.19 out of 4. Overall, this suggests that fear of deportation has a greater net effect on willingness to give out personal information to police, but that perceived sanctuary status has some small but statistically significant effect in encouraging undocumented immigrants to cooperate with police, but this sanctuary effect is concentrated among respondents with low fear of deportation. Overall, these results provide support for Hypothesis 3. 8
Comfort Level: Individual Scenario Models Interaction Models.
*p < .1; **p < .05; ***p < .01.
Finally, we turn to our scaled dependent variable. The results are consistent with the main findings from the earlier models: our main independent variables (fear of deportation and sanctuary city) and their interaction are statistically significant. For instance, in Model 1, Table 6, the fear of deportation coefficient is −1.027, which means that as fear of deportation increases by one point on the 4-point scale, a respondent reduces willingness to give out personal information by 1.027 points on the comfort scale (which ranges from 6 to 24). Likewise, respondents reportedly living in sanctuary cities are 1.145 scaled points more comfortable in giving out information compared to respondents who claim to not live in a sanctuary city. The interaction term is also statistically significant in Model 2. Overall, these models provide strong support for H1–H3.
Comfort Giving Personal Information Out, Scaled Dependent Variable.
*p < .1; **p < .05; ***p < .01.
To better evaluate Model 3 substantively, we present the effects graphically in Figure 5. The graph plots out predicted values for different simulated values of our key independent variables and a respondent's expected comfort level at that particular value of the two independent variables and their interaction. We hold all other variables at their central tendency. For example, on the right side of the plot—where respondents are very fearful of being deported—respondents fall below the mean on the comfort scale—around 14.7 points (out of 24). Notably, there is no difference between people who believe they reside in sanctuary cities, don’t know, or do not think they live in a sanctuary city.
Predicted effect of fear of deportation, sanctuary city status, and comfort in giving out personal information.
This begins to change though as one moves from great fear to less fear, and among people who report a little fear, a statistically significant gap emerges between people who live in sanctuaries vs. those that do not (or don’t know). Essentially, people who think they live in a sanctuary city and are a little fearful of deportation score nearly 16.8 (ranging between 6 and 24) on the comfort scale, whereas people who think they do not live in a sanctuary city score about 15.7 on the comfort scale. Finally, the gap is the largest among people who really aren’t worried about deportation. People who think they live in a sanctuary city are fully 2 points higher on the comfort scale. This translates into about an 11 percentage point difference between respondents with low levels of fear living in different safety environments (sanctuary vs. non-sanctuary). While we cannot say for sure these differences translate into substantive real-world differences among people on the ground (since we are observing attitudinal, not behavioral, data), these effects are large enough to make a subset of the immigrant population appear to feel a bit more comfortable participating in social and economic activity. These findings could help to explain the psychological underpinnings behind some economic research showing that sanctuary status may reduce unemployment rates and increase housing prices. That said, it should be once again noted that the effects of ICE targeting some sanctuary cities in 2025–2026 may undercut any mollifying effects of sanctuary cities.
Although we are primarily concerned with the fear of deportation and sanctuary city variables, we now discuss our control variables. First, variables measuring education, age, years in the United States, and gender are not statistically significant in the full model. However, some of our control variables are statistically significant as well. It appears that immigrants who are less secure in their economic situation are less willing to give out personal information, whereas respondents with some sort of local identification and who speak good English are more willing to give out personal information—which makes sense. Finally, respondents who have children living in the United States and whose boss has asked them about their immigration status appear to be more comfortable giving out personal information. While this could be seen as a possible logical inconsistency, these variables could be measuring signs of stability in these peoples’ lives. By having kids in the United States, these kids may be U.S. citizens, and so this respondent may be more acculturated to U.S. institutions like schools and police. In addition, it is feasible that a boss might ask about an immigrant's status because they are looking out for them. In any event, future research can further investigate these relationships, but the point remains that adding these variables to the model does not substantively change our findings.
Discussion and Conclusion
This paper examined two overarching research questions: (1) Are undocumented immigrants who fear deportation less willing to give out personal information, and (2) Does sanctuary policy moderate the relationship between fear of deportation and willingness to give out personal information? While the answer to these research questions may seem somewhat obvious, to date, scholars have had a difficult time answering them because gathering individual-level survey responses from H2S populations is so difficult. Some research has addressed the second question (Wong et al. 2021; Garrick and Schrank 2024), but its findings are experimental and not based on individual respondents’ understanding that they do or do not reside in a sanctuary jurisdiction, rather they are told to respond to fictitious scenarios, or the analysis is based on a relatively limited sample. The data this paper analyzes marks a departure from much of the extant research and enables us to provide a strong test of our hypotheses regarding immigrants’ willingness to provide personal information. More importantly, our data gathers the actual views of undocumented immigrants, something often lacking in the sanctuary cities literature. It is important to acknowledge the limitations of this study. Although venue-based sampling represents a strong alternative for studying hard-to-reach populations, it does not constitute a probability sample and is constrained by the specific cities included in the survey. Additionally, respondents may have been primed by the wording of the dependent variable, as questions beginning with “considering your immigration status” may have prompted respondents to imagine worst-case scenarios. While we recognize these limitations and encourage the development of probability-based samples of undocumented populations, such approaches remain difficult to implement for a population characterized by heightened vulnerability and institutional avoidance.
At the same time, this study benefits from a unique temporal context. The data were collected after the first Trump administration but prior to the subsequent expansion and normalization of highly visible immigration enforcement against non-criminal undocumented immigrants. As a result, this sample captures undocumented immigrants’ attitudes and behaviors during a narrowing window in which participation in research was still possible before fear and distrust of institutions intensified further. In this sense, the data represent a rare opportunity to observe undocumented immigrants’ willingness to engage and disclose information before enforcement practices rendered such engagement even more difficult. This change since the second Trump administration also highlights the vulnerability of sanctuary policies toward unilateral enforcement on the part of the federal government.
In this context, we find strong and consistent results across several dependent variables, showing that both fear of deportation and residing in a sanctuary city are associated with immigrants’ willingness to give out personal information. This means that the claim that sanctuary policy and sanctuary cities help increase immigrant trust in police and immigrant trust in local institutions is likely rooted in fact. Otherwise, we would not see such consistent responses in the relationship between fear of deportation and giving out personal information or sanctuary city status and giving out personal information.
Future research could investigate whether sanctuary cities continue to moderate immigrants’ fear of deportation and their willingness to hand out personal information. The second Trump Administration has implemented a very aggressive deportation agenda involving expanding immigrant detention, as well as high-level and widely covered stories of immigrant deportation. Therefore, the relationship we observed between our key independent and dependent variables may be so strong now that a sanctuary moderator would not bring us additional explanatory power. In other words, fear of deportation during the second Trump administration may be so strong as to negate any effect of sanctuary status. The federal enforcement context is now so hostile that friendly policies at the local level may fail to bring immigrants out of the shadows. Finally, given the linkages between fear, sanctuary status, and cooperation with formal and information institutions, future research may investigate whether undocumented immigrants disproportionately live in sanctuary jurisdictions as a means toward enjoying greater protections.
Supplemental Material
sj-docx-1-uar-10.1177_10780874261449679 - Supplemental material for Ameliorating the Fear: Sanctuary Cities, Fear of Deportation, and Comfort Giving Out Personal Information
Supplemental material, sj-docx-1-uar-10.1177_10780874261449679 for Ameliorating the Fear: Sanctuary Cities, Fear of Deportation, and Comfort Giving Out Personal Information by José E. Múzquiz, Gabriel Elías Martínez and Loren Collingwood in Urban Affairs Review
Footnotes
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
