The Impact of Interior Immigration Enforcement on the Lives of Undocumented Immigrants

Abstract

How does interior immigration enforcement affect how undocumented immigrants describe their interactions with state and societal institutions? Although there is some evidence that points to a broad range of chilling effects that result when local law enforcement officials work with U.S. Immigration and Customs Enforcement (ICE) on federal immigration enforcement, there is no systematic study that collectively examines the impacts that may result with such enforcement efforts. We situate our paper as evidence of the homeland security state and show how contemporary interior immigration enforcement subjugates undocumented immigrants in nearly all facets of their interactions with state and societal institutions. To illustrate this, we embedded an experiment in a survey (n = 594) drawn from a probability-based sample of undocumented immigrants. When respondents are told that local law enforcement officials are working with ICE on federal immigration enforcement, they report that they would be less likely to report crimes they witness or are victims of to the police, use public services that require them to disclose their personal contact information, do business that requires them to disclose their personal contact information, and participate in public events where the police may be present, among other main findings.

Introduction

Immigration remains a highly salient issue and interior immigration enforcement has become a focal point in the broader debate over comprehensive immigration reform. As Congress remains gridlocked, political contention over interior immigration enforcement is taking place between the federal government and state and local governments. Shortly after taking office, then-President Trump signed Executive Order 13768 making so-called sanctuary jurisdictions ineligible for certain federal funding, including law enforcement grants. ¹ This, alongside other changes to how the former Trump administration carried out the work of interior immigration enforcement, renewed interest in questions examining the impact of interior immigration enforcement policies. The contrasting approach to interior immigration enforcement by the current Biden administration, including immediately rescinding Executive Order 13768 and current attempts to return to Obama-era immigration enforcement priorities, further highlights the salience of interior immigration enforcement in the contemporary landscape of immigration politics and policies. Moreover, localities continue to grapple with the question of whether local law enforcement officials should do the work of federal immigration enforcement via the 287 g program. In this study, we examine the impact that interior immigration enforcement has on how undocumented immigrants describe their interactions with state and societal institutions.

Tightened interior immigration enforcement did not, however, originate with the Trump administration. It is the product of a decades-long process that created the “homeland security state” (De Genova 2007; Gonzales 2014, 2016; Lovato 2008), in which the U.S. government shifted its effort from a focus on the external border to also creating a system of laws and policies centered on interior immigration enforcement actions. The latest development in the growth of the homeland security state is local law enforcement officials increasingly doing the work of federal immigration enforcement through partnerships with U.S. Immigration and Customs Enforcement (ICE). What is unique about this latest iteration is that it complicates the preexisting image of the homeland security state as a unitary body. As such, it has become the case that undocumented immigrants are increasingly wary of any law enforcement figure for fear that they are involved in interior immigration enforcement efforts. By reimagining the homeland security state as an interwoven patchwork of entities dedicated to interior immigration enforcement, we would expect these federal-subnational partnerships to influence all aspects of how undocumented immigrants describe their interactions with state and societal institutions.

We provide evidence of this by embedding an experiment in a survey (n = 594) drawn from a probability-based sample of undocumented immigrants to better understand how undocumented immigrants would change their behaviors when learning that local law enforcement officials are working with ICE on federal immigration enforcement. We find that when local law enforcement officials are working with ICE on federal immigration enforcement, respondents in our sample report that they would be significantly less likely to report crimes to the police, participate in public events where the police may be present, engage in activities, including help-seeking activities, that requires them to disclose their personal contact information, place their children in after-school or day-care programs, and look for a new job. We contribute to a literature that has thus far examined limited outcomes when it comes to the effects of interior immigration enforcement and show how this latest iteration of the homeland security state has far-reaching effects on how undocumented immigrants describe their interactions with state and societal institutions. This is the first study the authors are aware of that simultaneously asks about the many facets of the day-to-day behaviors of undocumented immigrants. This is also one of the first studies to move from qualitative and ethnographic work to a survey experiment of undocumented immigrants.

We begin by illustrating the roots of the homeland security state. We document the literature that examines the impact of interior immigration enforcement policies, paying attention to research that examines undocumented immigrants’ interactions with state and societal institutions. We then derive a set of hypotheses about how the interactions undocumented immigrants have with state and societal institutions might be affected when local law enforcement officials do the work of federal immigration enforcement. After this, we describe the survey vehicle used to test our hypotheses and our survey experiment, wherein respondents are randomly assigned to conditions that vary the interior immigration enforcement context. We then discuss the findings of the survey experiment and the implications of the results.

Shift to Homeland Security State

A central component of a nation-state’s sovereignty is the ability to control and regulate its borders (Krasner 2009). In the U.S., the government has often asserted its sovereignty through its immigration policies (Zolberg 2006). These policies typically revolved around the border, regulating who the country lets in, how many, under what conditions, and who the country expels. As Massey (2007) described, the framing of such policies shifted into crisis mode during the 1970s and 1980s, during which time the U.S.–Mexico border began to be depicted as a “battleground” that was “under attack” from “alien invaders” who threatened American culture and values. The symbolic portrayal of immigration as a crisis and of immigrants as a threat reached new heights in the 1980s during the Cold War (Zolberg 2006). Ronald Reagan, though rather muted compared to more recent Republican presidents such as former President Trump, described immigrants as “swarming” into the U.S., further framing immigration as a question of national security to be dealt with internally (Cannon, Hoffman, and Omang 1983). This rhetoric was the beginning of a punitive turn in immigration that extended beyond the border and permeated within the nation-state, leading to what some have described as the omnipresence of borders (Leerkes, Bachmeier, and Leach 2013). This rhetoric reached a new peak during the debate over H.R. 4437 in 2005 (Wong 2017a), then again during the Obama years when the Tea Party emerged, and reached feverish levels during the Trump years. The U.S. no longer saw its sovereignty and legitimacy as only being threatened along its southern border with Mexico; rather, immigrants within the U.S. became a population to be controlled and, if necessary, removed.

This increasingly expansive government enforcement approach is part of the “homeland security state” (De Genova 2007; Gonzales 2014, 2016; Lovato 2008). Stemming from the rise of neoliberalism and the dismantling of the welfare state, it refers to the “US government apparatus for migration control” (Gonzales 2016, 81) whose goal is the securitization of migration (De Genova 2007). The homeland security state views the nature of the contemporary migration control regime as a “dynamic integral state” (Gonzales 2016, 89) project that encompasses more than external borders and seeks to police and control migrants internally. The homeland security state has expanded its reach by breaking down the boundaries between inside and outside, and views the homeland in a state of constant emergency (Kaplan 2003). This has particularly been true since 9/11. And while the public rhetoric seems to distinguish between “responsible,” “law-abiding” citizens, and “irresponsible” and “deviant” noncitizens (Macías-Rojas 2016), the ultimate goal of the homeland security state is not necessarily to remove all undocumented immigrants from the U.S.; rather, it is to maintain and legitimize the state and its actions in the eyes of the public (Golash-Boza 2016; Harrison and Lloyd 2011) and produce the specter of “guilt” that manifests in immigrants’ feelings of perpetual foreignness and deportability.

In the 1980s, the rhetoric behind the homeland security state shifted. The passage of the 1986 Immigration Reform and Control Act (IRCA) was a significant point in the criminalization of migrants (Medina 1997). It was passed in response to uneasiness about the increasing number of undocumented immigrants in the country, reflecting concerns about the U.S.’ control of its borders, and the U.S.’ response to demonstrate its sovereign power (Arnold 2011). The act criminalized the hiring of undocumented workers and authorized sanctions against employers who knowingly hired undocumented immigrants. It also sped up deportations of some foreigners and built up the U.S. Border Patrol agency. While IRCA did include two legalization programs, it also solidified a distinction between legal and undocumented immigrants.

The homeland security state was further strengthened in 1996, with the passage of the Illegal Immigration Reform and Immigrant Responsibility Act (IIRaIRA) and the Antiterrorism and Effective Death Penalty Act (AEDPA). These laws increased the scope of deportation provisions, narrowed the grounds for appealing a deportation order, and apportioned additional funds to immigration law enforcement (Golash-Boza 2016). The most significant provision of IIRaIRA and AEDPA was the expanded role and authority given to state and local police in immigration enforcement. IIRaIRA included an amendment, codified as section 287(g) in the Immigration and Nationality Act, whereby state and local governments could enter into formal agreements with federal immigration enforcement agencies to identify, apprehend, and detain undocumented immigrants. For the first time, local law enforcement officials could, with training and approval from the federal government, directly enforce federal civil and criminal laws (Armenta 2017; Wong 2012). These programs further expanded the reach of sovereignty by integrating disciplinary techniques of surveillance through the connections between federal and local enforcement (Valdez 2016).

The next major period in the development of the homeland security state followed the 9/11 attacks. What distinguished this period from earlier stages is that new enforcement priorities institutionalized a form of criminalization that differed from “illegalization” (Macías-Rojas 2016). After 9/11, the federal government allocated unprecedented funding toward policing noncitizens in the name of national security and counterterrorism (Armenta 2017). The newly established U.S. Department of Homeland Security (DHS) encouraged local authorities to participate in the 287(g) program while the U.S. Department of Justice also attempted to get local police and sheriffs to participate in immigration enforcement. Unlike the 287(g) program, which was voluntary, the Secure Communities program (established in 2008) required local police to process the fingerprints of all arrested individuals using DHS databases to screen for immigration violations and identify potentially deportable individuals. Under this program, ICE was given a remote presence in all local law enforcement agencies, thus permeating the system of interior enforcement throughout the country.

Most recently, the former Trump administration imposed, or attempted to impose, interior immigration enforcement measures that would have significantly expanded the power of the homeland security state. The former Trump administration worked aggressively to recruit more states and localities to work with ICE on federal immigration enforcement. In July 2017, the administration held a press conference touting the signing of eighteen new 287(g) agreements with eighteen counties in the State of Texas. This push to expand federal efforts on interior enforcement has also trickled down to the state and local level. According to the National Conference of State Legislatures (NCSL), since 2010, states have enacted 1,274 immigration-related laws and 207 of these have focused on immigration enforcement. These laws and ordinances require state and local police to check the immigration status of a person if a law enforcement officer has a reasonable suspicion that the person is undocumented (e.g., SB 1070 in Arizona, SB 4 in Texas), prevents state and local governments from enacting so-called sanctuary policies (e.g., HB 2315 in Tennessee and SB 168 in Florida), and prohibits the letting, leasing, or renting of apartments or homes to undocumented immigrants (e.g., Ordinance 2006-18 in the City of Hazelton, Pennsylvania), among other provisions.

This latest iteration of the homeland security state—the involvement of states and localities in interior enforcement efforts—has strengthened the technologies of control (McDowell and Wonders 2009-2010). But while the homeland security state has expanded in power, there still exists an underlying assumption that the state operates as a unitary power. To date, studies examining the impact of local immigration enforcement efforts have generally focused on the effects that individual policies have had on specific outcomes, for example, how denying health insurance to undocumented immigrants impacts health outcomes. These are important contributions; however, we argue that the effects of the homeland security state are neither isolated nor discrete, and that to capture the full breadth of outcomes, research must take a wide-angled view on how local law enforcement cooperation with federal immigration enforcement officials can permeate into every facet of how undocumented immigrants describe their interactions with state and societal institutions. Our study is the first to examine the far-reaching effects that result when local law enforcement works with federal authorities on immigration enforcement.

To situate our hypotheses on the wide array of outcomes that result when states or localities tighten interior immigration enforcement efforts by working with ICE on federal immigration enforcement, we present research on the effects of increased immigration enforcement on the public safety, public health, education, and employment of undocumented immigrants, their families, and broader community.

Impact on Public Safety

Some local law enforcement agencies argue that doing the work of federal immigration enforcement is necessary to ensuring public safety, as doing so will decrease crime. This argument is not supported by the available empirical evidence (Gonzalez O’Brien, Collingwood, and El Khatib 2017; Martínez-Schuldt and Martínez 2019; Treyger, Chalfin, and Loeffler 2014; Wong 2017b; Wong et al. 2021). Increased interior immigration enforcement, rather than decreasing crime, can sow distrust and fear in local law enforcement among immigrant communities, which can undermine public safety. For example, using a combination of qualitative interviews, focus groups, and other similar methods, Nguyen and Gill (2016) found that the implementation of the 287(g) program in two jurisdictions in NC created fear and distrust of law enforcement, so much so that the Latin American immigrants they interviewed “became reluctant to leave their houses or drive anywhere due to fear of encountering the police […] because of their new role as “la migra” or immigration agents” (14–15). This distrust is consequential for public safety because, as Comino, Mastrobuoni, and Nicolò (2016) found, undocumented immigrants are already less likely to report crimes to the police, even despite their victimization.

Other studies provide additional evidence that gives cause for these concerns. In analyzing the text of arrest reports before and after the implementation of the 287(g) program in Davidson County, TN, Donato and Rodriguez (2014) found differences in the reasons given for arrests, in particular, that characteristics related to foreignness, “such as country of origin, language use, and legal status […] became more salient after 287(g)” (1696). These effects are problematic, because while some undocumented immigrants view the police as professional and trustworthy (as compared to law enforcement in their home country), others find that contacting the police only affirms their negative perceptions (Armenta and Rosales 2019). One important implication of this distrust and fear is decreased reporting of crimes to law enforcement. Undocumented immigrants who live in inhospitable climates will often decide it is “not worth it” (García 2019, 124) to report small-scale crime. Based on survey data in four metropolitan areas (Los Angeles, Houston, Phoenix, and Chicago), Menjívar et al. (2018) found that “documented” Hispanics/Latinos and some native-born Hispanics/Latinos are less likely to report crimes depending on the interior immigration enforcement context. Such findings are confirmed by Muchow and Amuedo-Dorantes (2020) who, using an index of Google Searches, find that “domestic violence calls per capita [in Los Angeles] decline 3% when immigration enforcement awareness increases by one standard deviation in districts with a higher share of Latino noncitizens, relative to other districts” (13).

Although less systematically studied, other interior immigration enforcement policies, such as immigration enforcement actions (i.e., apprehensions) at courthouses, appear to similarly sow distrust and fear. For example, in August 2017, Brooklyn District Attorney Eric Gonzalez and New York Attorney General Eric Schneiderman held a news conference calling for an end to ICE enforcement actions in New York courthouses. Gonzalez stated that immigration enforcement actions at courthouses discourage immigrants from showing up as witnesses and even defendants (New York State Office of the Attorney General 2017). Then, in February 2018, numerous officials in the City of New York penned an open letter calling for a ban on ICE from courthouses, as “Statistics and reports from across New York State confirm that immigrants are declining to appear in court or to initiate legal actions due to fear of ICE” (Council of the City of New York 2018).

We thus hypothesize that when local law enforcement officials work with ICE on federal immigration enforcement, undocumented immigrants will report that they are not only less likely to report crimes that they witness to the police (H₁), but they will also be less likely to report crimes that they are victims of to the police (H₂). Moreover, because distrust can be exacerbated by the interior immigration enforcement context, we also hypothesize that undocumented immigrants will say they are less likely to participate in public events where law enforcement may be present when local law enforcement officials work with ICE on federal immigration enforcement (H₃).

Accessing Public Services

How does increased interior immigration enforcement affect access to public services among undocumented immigrants? On this question, the literature is arguably strongest when it comes to public health services. There is a small, but persuasive literature that has uncovered several adverse public health effects tied to increased interior immigration enforcement. Several studies have examined the impact of restrictive state-level immigration laws. One of the main proposals in this regard was Proposition 187, which would have made undocumented immigrants ineligible for public services, including publicly funded health services, and would have also required public employees to report undocumented immigrants to the then-Immigration and Naturalization Service (INS). Asch, Leake, and Gelberg (1994) found that in the wake of Proposition 187, undocumented tuberculosis patients not only feared that going to a doctor would result in an immigration enforcement action, but were four times more likely to delay seeking care. In analyzing administrative data from the San Francisco County Division of Mental Health and Substance Abuse Services (DMS) system, Fenton, Catalano, and Hargreaves (1996) found that Hispanics/Latinos between the ages of eighteen and forty-five were significantly less likely to use outpatient services and were significantly more likely to use crisis services when comparing the year before and after the passage of Proposition 187. Although it was ultimately enjoined and declared unconstitutional, the deleterious health effects of Proposition 187 were indelible.

Over a decade later, research on similar restrictive state-level immigration laws—e.g., Arizona’s Senate Bill 1070 (SB 1070) and Alabama’s House Bill 56 (HB 56), has found similar effects. In Arizona, Hardy et al. (2012) found that SB 1070 “changed the health-seeking behaviors of residents of a predominately Latino neighborhood by increasing fear, limiting residents’ mobility, and diminishing trust in officials” (1250). As the passage of SB 1070 coincided with their longitudinal study of the health and development of Mexican-origin adolescent mothers and their infants, Toomey et al. (2014) were able to analyze pre- and post-trends, finding that young mothers were less likely to use preventative health care after the passage of SB 1070, among other similar findings. Using county administrative health data, White et al. (2014) found a significant decline in the use of county public health services among Latinos when comparing the year before and after the passage of HB 56, even for public health services that were exempt from the law (i.e., immunizations and treatment for communicable diseases).

Research has also examined the effects of 287(g) and Secure Communities. Wang and Kaushal (2018) used the restricted-use National Health Interview Survey for 2000 to 2012 and exploit time variation (i.e., when 287(g) agreements were adopted and the staggered rollout of Secure Communities) to obtain differences-in-differences estimates for Latino immigrants across several items in the questionnaire. They find that 287(g) and Secure Communities worsened mental health distress scores among Latino immigrants. More generally, research shows how concerns about immigration enforcement not only have negative implications for undocumented immigrants, but can also affect American citizen children who are part of mixed immigration status families. Watson (2010) found that after the passage of IIRaIRA in 1996, Medicaid participation among children of noncitizens decreased significantly, even among the American citizen children of noncitizens.

Chilling effects as they relate to public health are well documented. We argue, however, the effects of increased interior immigration enforcement seep more pervasively into how undocumented immigrants interact with state and societal institutions than has been previously documented. We thus hypothesize that when local law enforcement officials work with ICE on federal immigration enforcement, undocumented immigrants will report that they would be less likely to use public services in general, not just public health services, when doing so requires them to disclose their personal contact information (H₄). We also extend beyond public services by hypothesizing that when local law enforcement officials work with ICE on federal immigration enforcement, undocumented immigrants will be less likely to say they will do business (e.g., open a bank account, get a loan) when doing so requires them to disclose their personal contact information (H₅).

Children and Schools

Research has also shown that increased interior immigration enforcement can negatively impact public education. For example, when using an index of the intensity of interior immigration enforcement at the metropolitan statistical area (MSA) level, Amuedo-Dorantes and Lopez (2015) found that interior immigration enforcement increases the probability that a Hispanic child (6–13 years old) repeats a grade by 6% and increases their likelihood of dropping out of school by 25%. Moreover, Capps et al. (2007) studied the immediate aftermath of three large-scale worksite immigration enforcement actions. They find that on the day of the enforcement actions, schools played an important role “in ensuring that children were not dropped off to empty homes or left at school overnight” (3). But despite these efforts, “after the arrest or disappearance of their parents, children experienced feelings of abandonment and showed symptoms of emotional trauma, psychological duress, and mental health problems” (6). It is thus perhaps intuitive that Chaudry et al. (2010) found that children missed school in the aftermath of the immigration enforcement actions they researched, in addition to experiencing similar traumas to those described above (see also Lopez et al. 2017).

Whereas undocumented parents might very well trust that schools are safe spaces for their children, they may also be concerned about immigration enforcement actions occurring at or around schools. During the Obama administration, the “sensitive locations” memo stated, “This policy is designed to ensure that […] enforcement actions do not occur at nor are focused on sensitive locations such as schools and churches”. However, as headlines such as “Immigrant arrested by ICE after dropping daughter off at school, sending shockwaves through neighborhood” are currently part of conversations about interior immigration enforcement conversation, we hypothesize that when local law enforcement officials work with ICE on federal immigration enforcement, undocumented parents will report that they would be less likely to place their children in an after-school or day-care program (H₆).

Impact on Employment

Increased interior enforcement can affect undocumented immigrants in the workplace in at least two ways: it can increase concerns about worksite immigration enforcement actions and deportation and it can exacerbate power dynamics, wherein undocumented workers are less likely to voice concerns to their employers, advocate for their own rights, and look for new jobs. For example, in surveying Hispanic/Latino immigrant restaurant workers in San Jose, CA and Houston, TX, Gleeson (2010) found that although undocumented workers have knowledge of their rights, and despite employers not having made explicit threats to “call immigration,” workers still chose not to come forward with their concerns (i.e., verbal abuse, no work breaks or meal hours, and wage theft) so as to not attract attention to their immigration status. As Harrison and Lloyd (2011) argued, increased interior immigration enforcement can create a pervasive sense of being surveilled by their employers, which can lock undocumented workers into exploitative conditions.

We thus hypothesize that local law enforcement officials work with ICE on federal immigration enforcement, undocumented workers will report that they would be less likely to look for new jobs (H₇).

Data and Method

To test our hypotheses, we embedded an experiment in a probability-based sample of undocumented immigrants in San Diego County. The survey vehicle used in this study is the Undocumented in America project based out of U.S. Immigration Policy Center (USIPC) at UC San Diego.

Through a partnership between the USIPC and the Mexican Consulate in San Diego County, Wong created a sample frame of undocumented Mexican nationals in San Diego County. The sample frame comprises individuals who receive consular services unique to those living in the United States without authorization. The sample frame, which includes approximately 73,000 people, accounted for 88.6% of undocumented Mexican nationals who were estimated to have lived in San Diego County at the time of the survey (Center for Migration Studies 2016). We note here that because our sample frame comprises undocumented Mexican nationals, we are unable to claim representativeness when it comes to the broader undocumented population. However, as undocumented immigrants, regardless of national origin, are subject to apprehension, detention, and deportation, our results are generalizable insofar that any undocumented immigrant is concerned about the consequences of tightened interior immigration enforcement. Moreover, as undocumented immigrants in the United States similarly live under the threat of the homeland security state, we expect that our results would apply in other contexts where undocumented immigrants make decisions on whether to engage publicly or with official state and societal institutions.

Our survey experiment in San Diego County is situated in a state that has, in recent decades, been a staunch defender of immigrant rights (Colbern and Ramakrishnan 2018). Moreover, San Diego County ranks sixteenth on one index of immigrant integration (New American Economy 2021). The largest city in San Diego County also formally declared itself a welcoming city for immigrants in 2018. ² San Diego County thus provides an important laboratory for examining how interior immigration enforcement affects how undocumented immigrants describe their interactions with state and societal institutions, as it provides a difficult test for our hypotheses. That is, do the hypothesized chilling effects of local law enforcement entanglement with federal immigration enforcement officials obtain even in an otherwise welcoming policy context?

Working with staff at the Consulate, Wong assigned random ID numbers to each record and then cut the sample frame into random draws of approximately 5,000 records for each survey module in the Undocumented in America project. Call sheets with limited information about each respondent—the random ID number assigned to each record, first name, and phone number—are then printed out. Phone numbers are manually dialed by enumerators trained by Wong. Phone numbers are dialed once with no additional follow-up. After each paper call sheet is completed, it is immediately reviewed and then destroyed. All surveys are conducted in Spanish, unless the respondent prefers to speak in English. In this study, 97.9% of surveys were conducted in Spanish. This study received IRB approval from (UC San Diego #180131).

This study represents the first in the Undocumented in America series. The survey was fielded between September 2017 and November 2017 and included 594 respondents. In the survey, we embedded an experiment. In the experiment, respondents were randomly assigned to one of two groups. In one group (n = 298 respondents), questions were prefaced with, “If the San Diego Police Department and the San Diego County Sheriff’s Department said they WILL NOT WORK WITH ICE on deportation raids, would you be more or less likely to…?” In the second group (n = 296 respondents), questions were prefaced with, “If the San Diego Police Department and the San Diego County Sheriff’s Department WERE WORKING TOGETHER WITH ICE on deportation raids, would you be more or less likely to…?” Whereas local cooperation with federal immigration enforcement officials can take many forms, including the aforementioned 287(g) and Secure Communities programs, we intentionally use the language “WORKING TOGETHER WITH ICE on deportation raids” to convey to our respondents that local law enforcement is using what is perhaps the bluntest instrument in the interior immigration enforcement toolbox. We note that due to sample size limitations, we were unable to assign respondents to a true control group (i.e., no information about the interior enforcement context). Instead, our survey experiment varies the hypothetical interior immigration enforcement context from permissive to restrictive. We can thus interpret effect sizes as the differences between the high and the low bounds of the reported behavior of undocumented immigrants as the interior immigration enforcement context varies.

Table 1 provides the exact text. An experiment such as this is superior to analyzing observational survey data because asking respondents about one scenario is insufficient for determining how their behavior may or may not change based on the second scenario; asking respondents about one scenario and then the second scenario would likely produce biased results because responses related to the first scenario would likely influence responses to the second scenario. Random assignment to one of the two groups balances the two groups across the broad range of covariates that need to be controlled for in observational analyses, and random assignment to one of the two groups means that differences in responses can be causally attributed to the variation in the two scenarios. Moreover, prior work is limited in evaluating the causal effects of interior immigration enforcement because it is typically qualitative or ethnographic and, if it is based on a large survey, it is typically a sample of the larger Latinx population. Given the general lack of data on outcomes specific to undocumented immigrants—indeed, many studies of undocumented immigrants are, in fact, studies that use Latinx as a proxy for undocumented immigrants without addressing matters of imbalance across important covariates—an experiment such as this is an important step forward for causally identified work on undocumented immigrants. We encourage researchers, however, to continue seeking out data and to think creatively about policy or other discontinuities so that we can push the literature further and better understand how policies causally affect outcomes for undocumented immigrants.

Table 1. Survey experiment questions

It is important to note that our survey does not capture actual changes in behavior; rather, our survey items capture behaviors undocumented immigrants report they would take and we measure behavioral change using a survey experiment that varies the hypothetical interior immigration enforcement context from permissive to restrictive. We can thus interpret effect sizes as the differences between the high and the low bounds of behavior. However, there is long-standing evidence to suggest that reported behaviors are positively correlated with later behavior (Azjen and Fishbein 1980). ³ Moreover, enforcement policies do not even need to be implemented for negative effects to take place. Torche and Sirois (2018) found that the mere knowledge of SB 1070, despite only being signed into law and never implemented, resulted in a decline in birth weight among Latina immigrant women.

The response rate for the survey was 12.2%. Per our agreement with the Consulate, we have a one-call protocol for conducting the survey, meaning each number is dialed only once unless we contact a respondent and the respondent requests that we call back at a later time. If we were able to reach an individual during the first call attempt and the person refused, no social pressure to participate was added. ⁴ If we were unable to contact the respondent, the number was not dialed again. We address potential imbalance in covariates between our respondents and those we were not able to contact in several ways. We used first name to estimate gender. An estimated 48.2% of our random draw was female compared to an estimated 51.9% of our respondents. We also compared the distribution of area codes for the numbers in our random draw to the distribution of area codes for the numbers of our respondents, acknowledging that area codes are imperfect proxies for where in San Diego County respondents live. In our random draw, 49.1% of numbers were in “Area Code 1,” 42.5% were in “Area Code 2,” 6.2% were in “Area Code 3,” and 2.2% were in “Other” area code. Among our respondents, 49.2% of numbers were in “Area Code 1,” 42.7% were in “Area Code 2,” 6.1% were in “Area Code 3,” and 2.0% were in “Other” area code. Moreover, among our respondents, we found no significant differences in response rates by hour. As demographic data on all individuals in the sample frame are unknown to us, it is not possible for us to test for potential imbalance differently. Given the highly sensitive nature of these data, traditional demographic information was purposively excluded in efforts to minimize the possibility of tracing our research records back to any respondent. Thus, it is impossible to know with certainty how representative, and thus generalizable, the sample frame is to the universe of undocumented Mexican nationals in San Diego County. ⁵ As research evolves on surveying undocumented immigrants, we look forward to innovation that builds on our best attempt here.

Results

Regarding interactions with the police, 60.8% of respondents say they would be less likely to report a crime that they witnessed to the police when local law enforcement officials work with ICE on federal immigration enforcement (p < .001). Moreover, 42.9% of respondents say they would be less likely to report a crime that they were a victim of to the police when local law enforcement officials work with ICE on federal immigration enforcement (p < .001).

The data show similar chilling effects when it comes to undocumented immigrants’ reported interactions with institutions that require them to disclose their personal contact information. 69.6% of respondents say they would be less likely to use public services (e.g., go to City Hall) that require them to disclose their personal contact information when local law enforcement officials work with ICE on federal immigration enforcement (p < .001). Moreover, 63.9% say they would be less likely to do business (e.g., open a bank account, get a loan) that requires them to disclose their personal contact information when local law enforcement officials work with ICE on federal immigration enforcement (p < .001).

Moreover, 68.3% of respondents say they would be less likely to participate in public events where police may be present if local law enforcement officials work with ICE on federal immigration enforcement; among respondents with children, 42.9% report they would be less likely to place their children in an after-school or day-care program if local law enforcement officials work with ICE on federal immigration enforcement; and 52.1% of respondents report they would be less likely to look for a new job if local law enforcement officials work with ICE on federal immigration enforcement.

Table 2 summarizes the results. Figure 1 graphically depicts the results. Two sample t-tests are used to calculate average treatment effects and 95% confidence intervals. ⁶ In the figure, behavioral items are sorted along the x-axis by effect size.

Table 2. Results summary

Figure 1. Average treatment effects.

Multivariate Analysis

The differences-in-means between the two experimental conditions are striking, but do the results in our sample hold when accounting for other factors? In particular, do the results hold when accounting for factors that may increase concerns about being detained or deported independent of whether local law enforcement officials work with ICE on federal immigration enforcement? Here, we estimate a series of logistic regression models that estimate the effect of the “working together with ICE” condition while also accounting for whether respondents have children, which may exacerbate concerns about family separation (children), whether respondents have immediate family members, meaning a parent, spouse, or sibling, who are U.S. citizens, which may also exacerbate concerns about family separation, but also increases the stakes of being detained or deported given the possibility that a respondent might be able to file an immigration petition (citizen relative), whether respondents have family members or close friends who have been deported, which makes even more vivid the consequences of immigration enforcement (deport), and whether respondents “strongly agree” or “agree” that they know what their rights are if ICE shows up at their door, as not knowing one’s rights can potentially result in more generalized concerns about immigration enforcement (know rights). We note here that we are limited in the time, and thus the number of items we can include in each questionnaire. ⁷

Table 3 reports the results based on our sample. As the table shows, the effects of the “working together with ICE” condition remain significant for all the day-to-day interactions analyzed (see Models 1 to 7). Moreover, the control variables are mostly statistically insignificant. This should not be interpreted to mean that the control variables do not matter. More to the point, the results show that the chilling effects that result when local law enforcement officials work with ICE on federal immigration enforcement are generalized across respondents.

Table 3. Multivariate analysis

***p < .01, **p < .05, *p < .1 (two-tailed tests). Model 1 examines the likelihood of reporting a crime that respondents witness to the police. Model 2 examines the likelihood of reporting a crime that respondents are victims of to the police. Model 3 examines the likelihood of attending public events where police may be present. Model 4 examines the likelihood of using public services (e.g., go to City Hall) that requires respondents to disclose their personal contact information. Model 5 examines the likelihood of doing business (e.g., open a bank account, get a loan) that requires respondents to disclose their personal contact information. Model 6 examines the likelihood of placing children in an after-school or day-care program (among respondents with children). Model 7 examines the likelihood of looking for a new job.

Discussion and Conclusion

How does interior immigration enforcement affect how undocumented immigrants describe their interactions with state and societal institutions? In this study, we used a survey experiment to show that when local law enforcement officials work with ICE on federal immigration enforcement, undocumented immigrants, based on our sample, are 60.8% less likely to say they would report crimes they witness to the police, 42.9% less likely to say they would report crimes they are victims of to the police, 69.6% less likely to say they would use public services (e.g., go to City Hall) that requires them to disclose their personal contact information, 63.9% less likely to do business (e.g., open a bank account, get a loan) that requires them to disclose their personal contact information, 68.3% less likely to say they would participate in public events where the police may be present, 42.9% less likely to say they would place their children in an after-school or day-care program, and 52.1% less likely to say they would look for a new job. These results are robust to whether respondents have children, whether they have immediate family members who are U.S. citizens, whether they have family members or close friends who have been deported, and whether they know what their rights are if ICE shows up at their door. Ours is the only survey experiment that we are aware of that does this using a probability-based sample of undocumented immigrants.

These findings complement the existing literature that describes how undocumented immigrants embody various performative aspects of being “American” in order to evade detection from authorities. Whether it is dressing in fashionable U.S. styles, driving in newer model cars (Rodríguez and Hagan 2004), or walking with an inconspicuous gait and speaking English outside the home (García 2014), there is a prevailing scholarship that shows lacking legal status does not necessarily translate to life in the shadows. At the same time, while undocumented immigrants may not retreat from public life, increased immigration enforcement has a demonstrated effect on undocumented immigrants’ interactions with institutions, such as medical systems (Berk and Schur 2001), law enforcement (Nguyen and Gill 2014), and schools (Capps et al. 2007). Our findings corroborate this work and move the scholarship forward by demonstrating the broad range of institutions undocumented immigrants say they would avoid to prevent detection from authorities. We find that when local law enforcement becomes entangled with federal immigration enforcement, undocumented immigrants, based on our sample, would not only avoid law enforcement, but also avoid interacting with a broader array of state and societal institutions, including banks, schools, and hospitals. This further demonstrates how the homeland security state, as it entrenches itself across multiple levels of government in a dense patchwork of policies and practices, causes undocumented immigrants to feel deep uncertainty about engaging in public-facing behaviors. As such, they retreat almost completely from interactions with a wide array of public institutions. Described by Brayne (2014) as “system avoidance,” undocumented immigrants see these vital community and social service institutions as an extension of the homeland security state. Even though these institutions provide core functions of social provision, they are an extension of the carceral state (Wacquant 2009), prompting undocumented immigrants to avoid contact so as to not make their presence known to authorities. Therefore, the prevailing image of the homeland security state as a unitary entity is insufficient in grasping the full scope of the impact of enforcement actions. Only by acknowledging that the homeland security state is in fact multiple entities working together on enforcement can we begin to capture how this interwoven relationship permeates deeply into undocumented immigrants’ lives.

Our findings add to a growing body of literature that demonstrates the adverse effects that result when local law enforcement officials do the work of federal immigration enforcement in several ways. First, we provide evidence of the magnitude (i.e., effect size) of the chilling effects that result when local law enforcement officials work with ICE on federal immigration enforcement. Our findings thus move the literature forward by showing how significantly undocumented immigrants are affected by tightened interior immigration enforcement. While our quantitative findings cannot explain why the magnitude of effects vary across day-to-day behaviors, we encourage future research to attempt to explain why.

Previously, the scholarship focused on singular outcomes of individual policies. Previous research has been useful for those studying health, education, and employment outcomes, among others. This paper moves this literature forward by examining the range of behavioral effects of enhanced interior enforcement in a single survey experiment. Doing so provides a more accurate depiction of the full scope of undocumented immigrants’ lives, and how the latest version of the homeland security state touches upon virtually every facet of their daily behaviors. Although we analyzed seven different behavioral indicators, we likely have only scratched the surface of the pervasiveness of interior immigration enforcement.

Our findings, based on our sample, also have implications for states and localities that are already, or are considering, working with ICE on federal immigration enforcement. To the extent that people, undocumented or not, are less likely to report crimes that they witness, let alone are victims of, to the police, serious public safety implications can result. Being less likely to participate in public events where the police may be present impedes the societal incorporation of undocumented immigrants. Moreover, to the extent that undocumented immigrants are less likely to engage with public institutions, whether this means going to City Hall or going to the bank, they are less likely to obtain services that they may be eligible for and need. When undocumented parents are less likely to enroll their children in an after-school or day-care programs the consequences of tightened interior immigration enforcement accrue to children and may have broader implications for their education and development. And being less likely to look for a new job may lock undocumented workers in exploitative workplaces or working conditions.

Of course, the purpose of “attrition-through-enforcement” interior immigration enforcement policies is to make conditions of living so harsh that “self-deportation” is preferred to being undocumented. We thus have no illusions that what may be viewed as adverse policy effects to some, may be viewed as desired policy effects to others. These differences notwithstanding, our results, based on our sample, underscore the importance and need for more systematic empirical research on the effects of interior immigration enforcement policies. In our own work, we further explore one of the mechanisms that links tightened interior immigration enforcement with decreased interaction with local law enforcement officials, which is decreased trust (see also Wong et al. 2021). We also explore how undocumented immigrants navigate ambiguities in interior immigration enforcement policies at state and local levels (i.e., restrictive state-level laws, but welcoming city-level ordinances, and vice versa). Moreover, whereas we show that the chilling effects of tightened interior immigration enforcement are systematic and pervasive, we also look forward to research about the resistance strategies that undocumented immigrants adopt, particularly against the backdrop of an increasingly restrictive federal immigration policy context.