The unintended consequences of US immigration enforcement policies

Significance

Every day, thousands of immigrants in the United States are held in immigration detention under conditions that are no different than those facing the criminally incarcerated. Thousands of other immigrants are deported to their countries of origin without the opportunity to be heard before an immigration judge. US policymakers hope that these punitive enforcement policies will deter individuals from attempting to enter the United States without authorization. Yet this study of individuals in El Salvador, Guatemala, Honduras, and Mexico finds no evidence that such policies change their intentions to migrate to the United States. On the other hand, the survey analysis indicates that immigration detention may foster beliefs that the US immigration system lacks procedural and outcome fairness.

Abstract

US immigration enforcement policy seeks to change the behaviors and views of not only individuals in the United States but also those of prospective migrants outside the United States. Yet we still know relatively little about the behavioral and attitudinal effects of US enforcement policy on the population abroad. This study uses a randomized experiment embedded in a nationally representative survey that was administered in El Salvador, Guatemala, Honduras, and Mexico to analyze the effects of US deterrence policies on individuals’ migration intentions and their attitudes toward the US immigration system. The two policies that the current study examines are immigration detention and nonjudicial removals. The survey results provide no evidence that a heightened awareness of these US immigration enforcement policies affects individuals’ intentions to migrate to the United States. But heightened awareness about the widespread use of immigration detention in the United States does negatively impact individuals’ assessments about the procedural and outcome fairness of the US immigration system. These findings suggest that immigration detention may foster delegitimating beliefs about the US legal system without producing the intended deterrent effect.

US immigration enforcement policy has grown increasingly punitive over the years (1–3). As underscored in a growing body of research, a key driving force behind this trend is the widespread belief among many policymakers that such punitive policies will have a deterrent effect (4, 5). The pain of punishment is presumed to shift the cost–benefit calculations of individuals who might contemplate unauthorized migration (6). Thus, the basic assumption underlying the expanded use of harsh immigration enforcement policies is that “tougher policies discourage people from migrating in the first place” (7). Empirical research, however, casts doubt on the validity and strength of this assumption (2, 8–10).

Much less examined and appreciated in public discourse are negative legal socialization effects of immigration enforcement policies. By legal socialization, I mean “the process through which individuals develop values, attitudes, and beliefs about laws, the institutions that create laws, and people that enforce laws” (11). Emerging research shows that US immigration enforcement practices are fostering governmental distrust and delegitimating beliefs about US legal authorities among both native and foreign-born populations in the United States (12–17). These negative legal socialization effects likely reach beyond our borders. A cross-border diffusion of such distrust and delegitimating beliefs has important implications for the United States in terms of both domestic and global governance. Studies show that prospective and current migrants’ views about the legitimacy of destination countries’ legal authorities are predictive of their decisions about whether to comply with the destination countries’ immigration laws (18, 19). A widespread cross-border diffusion of distrust and delegitimating views can also undercut the claimed moral authority and leadership of the United States in global politics, especially with respect to issues such as human rights, democracy, and the rule of law (20, 21).

This study explores two critical questions about US immigration enforcement policies aimed at deterring the flow of unauthorized migration. First, do punitive enforcement policies change individuals’ intentions to migrate to the United States? Second, do punitive enforcement policies erode fairness perceptions about the US immigration system among populations abroad? This study focuses on three key dimensions of fairness perceptions that are at the center of a longstanding body of research on justice and legitimacy: procedural justice, distributive justice, and interactional justice (22–26). This study addresses these questions using a large-scale randomized survey experiment in Mexico and the Northern Triangle countries of El Salvador, Guatemala, and Honduras. Migrants from these four countries constitute by far the largest share of unauthorized flow into the United States (27).

In terms of enforcement policies, this study focuses on immigration detention and nonjudicial removals. Just as the United States has been called the “incarceration nation” for having the highest criminal incarceration rate in the world, it has also been described as a “detention nation” for leading the world in its immigration detention rate (28). In fiscal year 2017, a total of 323,591 individuals were initially booked into the custody of the Department of Homeland Security (DHS); that total increased to 510,854 in fiscal year 2019 (29). Over the years, government watchdogs, nongovernmental organizations, journalists, and scholars have documented systematic neglect, abuse, and mistreatment of children and adults in immigration detention (30–33). Yet the use of immigration detention has continued through both Democratic and Republican administrations over the years, and the Trump administration expanded its use by explicitly invoking the deterrence rationale (34).

Nonjudicial removals have emerged as another key deterrence tool for US immigration enforcement. These removals—sometimes known as “speed deportations”—take the form of summary expulsion without a hearing before an immigration judge (35). Among the key types of nonjudicial removals are expedited removal, administrative removal, and reinstatement of removal orders (36). Nonjudicial removal decisions are made by frontline immigration officials without the safeguard of basic procedural protections afforded to noncitizens in judicial proceedings (37). These removals have become so common in recent years that the majority of individuals ordered removed from the United States do not ever appear before an immigration judge. For example, in fiscal year 2018, the DHS carried out nearly 340,000 removals; 43% of those removals were expedited removals, and 39% were based on the reinstatement of prior removal orders (38). In 2019, DHS issued a notice that it was expanding the use of expedited removal (39).

Materials and Methods

In close collaboration with the Latin American Public Opinion Project (LAPOP) at Vanderbilt University, I developed an experimental module that was placed on the 2018/19 AmericasBarometer. The AmericasBarometer is a longstanding, extensive, and methodologically rigorous survey that covers wide-ranging political and social topics of relevance in the Americas. The survey has been approved by Vanderbilt University’s Institutional Review Board. The interviewers obtain informed consent from individual respondents by implementing the procedures outlined in the detailed training manual published on LAPOP’s website. Using a study information sheet, interviewers describe the nature of the study, explain the potential benefits and costs of voluntary participation in the study, and provide opportunities for respondents to ask questions before deciding whether to participate. Except for the surveys in the United States and Canada, all surveys are conducted face-to-face. Only one respondent per household participates in the survey. Samples are developed using a multistage probabilistic design that results in nationally representative samples of voting-age adults in North America, Latin America, and the Caribbean.

I worked with LAPOP to design, pretest, and integrate the experimental module into the AmericasBarometer survey administered in Mexico and the Northern Triangle countries of Central America. A total of 6,134 individuals in Mexico (N = 1,547), El Salvador (N = 1,508), Guatemala (N = 1,549), and Honduras (N = 1,530) participated in the experimental module of the AmericasBarometer survey between November 2018 and March 2019. During the pretest phase, LAPOP’s fieldwork team based in the respective countries conducted several rounds of field testing to provide valuable input on the wording of the survey items. The items in English are discussed below (SI Appendix, Table S1 presents the same items in Spanish).

Experimental Design.

In each of the four countries, three experimental conditions were randomly assigned to the survey respondents, stratified by the respondents’ level of US social ties (for a further discussion of the research design, see SI Appendix). In each condition, the interviewers read to the respondents prompts containing different short descriptions about the US government’s response to unauthorized migration (see Experimental Conditions below). The respondents then rated their level of agreement or disagreement with three statements that are designed to measure the respondents’ legal attitudes (see Outcome 1 below). Finally, the survey asked the respondents about their intentions to migrate to the United States (see Outcome 2 below).

Experimental Conditions.

In the Control condition, the respondents were given the following prompt: “In the past 12 months, US immigration authorities apprehended thousands of [Mexicans]. Thinking about these [Mexicans], how much do you agree or disagree with the following statements?”^* In contrast, the respondents in the Detention condition were given the following prompt: “In the past 12 months, US immigration authorities apprehended thousands of [Mexicans] and placed them in detention facilities or jails for days, weeks, or months. Thinking about these [Mexicans], how much do you agree or disagree with the following statements?” The respondents in the Court condition were given the following statement: “In the past 12 months, US immigration authorities apprehended thousands of [Mexicans] and did not give them an opportunity to present their cases before an immigration judge. Thinking about these [Mexicans], how much do you agree or disagree with the following statements?” As discussed earlier, each of these prompts accurately describe the US enforcement policy in effect at the time of the survey.

The key differences across the experimental conditions are as follows: The Control condition explained only that many conationals were apprehended by US immigration authorities. By contrast, the Detention condition made salient to the respondents that many apprehended conationals faced immigration detention for an indefinite period of time. Finally, the Court condition made salient the lack of judicial process available to many conationals post apprehension.

Outcome 1.

Following the administration of the prompts described above, the respondents answered questions about their legal attitudes toward the US immigration system. I developed these items based on surveys used in the longstanding research on procedural justice, distributive justice, and interactional justice (40–42). Judgments about procedural justice focus on legal authorities’ decision-making process, such as the procedures’ consistency and neutrality (43).^† Distributive justice focuses on outcome fairness, which relates to “the degree that an outcome is consistent with, or can be justified by, a referent standard”; outcome fairness is therefore distinct from outcome favorability, which is about whether one obtains a positive outcome (44). Interactional justice “reflects the degree to which people are treated with politeness, dignity, and respect by authorities or third parties involved in executing procedures or determining outcomes” (22).

Using a seven-point scale (1 = strongly disagree to 7 = strongly agree), the survey asked the respondents to rate their level of agreement with each of the following statements: 1) “Generally speaking, [Mexicans] who are apprehend by U.S. immigration authorities go through a fair process” (Process); 2) “Generally speaking, [Mexicans] who are apprehended by US immigration authorities receive fair outcomes” (Outcome); and 3) “Generally speaking, [Mexicans] who are apprehend by U.S. immigration authorities are treated with respect” (Respect). In my analysis, I reverse coded each item so that higher values correspond to higher levels of disagreement.

Outcome 2.

Next, the respondents answered a question about their US migration intentions (US Migrate). Researchers of migration decision-making have conceptualized migration intentions as the “key proximate determinant of behavior” (45), and studies have shown a close relationship between emigration intentions and actual migration (46). Intentions are different from mere aspirations because, while many people may express a generalized wish or desire to move abroad, they may not be developing plans nor taking steps to actually engage in international migration. Thus, migration intentions constitute the “closest signal related to the ones who are likely to leave in the future” (47). Nonetheless, there may be substantial discrepancies between reported intentions and behavior. People may not ultimately realize their intentions to move, while others who did not have migration intentions may end up moving unexpectedly. These possibilities suggest caution in using migration intention as a proxy for migration behavior.

Using a four-point scale (1 = very likely; 2 = somewhat likely; 3 = a little likely; 4 = not at all likely), the survey asked the respondents the following: “And speaking of the United States, how likely is it that you will go to live or work in the United States in the next 3 years?” (US Migrate). In my analysis, I reverse coded this item so that higher values correspond to higher likelihoods of migrating.

Covariates.

The final regression models analyzing the outcomes described above control for a variety of factors that existing research has shown are significant determinants of migration decisions for individuals in Mexico and Central America (2, 5). These covariates are respondents’ sex (Male), age (Age), education level (Education), household income (Income), US migration history (US Past), the extent of US social ties (US Ties), the frequency of crime victimization in the country of origin during the past 12 mo (Victimization Count), and country of origin (Country). Each of these covariates are discussed in detail in SI Appendix.

Analytic Strategy.

For all regression models, I used ordered logistic regression, which estimates the relationship between an ordinal dependent variable and a set of independent variables. Before conducting the ordered logistic regression analyses, I used a multiple-imputation procedure with chained equations to impute missing values in the following covariates (48): Education, Victimization Count, Income, US Ties, and US Past. I used the mi impute chained command in Stata 16 to multiply impute missing data across ten datasets.

Results

To ensure that the experimental conditions were randomly distributed, I first checked for chance imbalances in covariate distributions between experimental conditions (49). As Table 1 confirms, there are no statistically significant differences (at P < 0.05) in the distribution of individual covariates across the experimental conditions. About half of the respondents in each of the conditions were men. The average age of the respondents was the same across the three conditions: about 40 y old. The average years of schooling was about 8.6 y in each of the conditions. About 7% of the respondents in each of the conditions stated that they “can save” on their current household income, whereas about 23% indicated that they were having a “hard time” on their current income level. About 6 to 7% of the respondents in each condition had lived as an immigrant in the United States. About 35 to 36% of the respondents in each condition had no US social ties. The average number of times that the respondents experienced crime victimization in their countries of origin in the past 12 mo was about the same in each condition: about 0.52 to 0.53 times. Finally, each origin country constituted about a quarter of all respondents in each condition.

Table 1.

Covariate descriptive statistics and balance check

	Proportion/mean (SD)			Total N
Variable	Control condition	Detention condition	Court condition
Male	0.50	0.48	0.50	6,134
Age	39.57 (16.23)	39.50 (16.40)	39.41 (16.39)	6,134
Education	8.59 (4.42)	8.67 (4.56)	8.63 (4.60)	5,922
Income				5,919
Can save	0.07	0.07	0.07	416
Just enough	0.32	0.28	0.30	1,772
Not enough	0.39	0.41	0.40	2,379
Hard time	0.23	0.23	0.23	1,352
US past	0.06	0.07	0.06	6,127
US ties				6,115
No ties	0.36	0.35	0.35	2,158
Talk never	0.09	0.09	0.10	573
Talk rarely	0.25	0.24	0.23	1,477
Talk 1 to 2x per month	0.13	0.12	0.11	736
Talk 1 to 2x per week	0.13	0.12	0.14	796
Talk every day	0.05	0.07	0.06	375
Victimization count	0.52 (1.48)	0.53 (1.39)	0.52 (1.44)	6,097
Country				6,134
Mexico	0.26	0.24	0.26	1,547
Guatemala	0.24	0.26	0.25	1,549
El Salvador	0.26	0.24	0.23	1,508
Honduras	0.24	0.25	0.25	1,530
Total N	2,094	1,995	2,045	6,134

Bivariate tests show that none of the differences in proportions and means across the experimental conditions are significant at P < 0.05. I used one-way ANOVA for the continuous covariates, χ² tests for the nonordinal categorical covariates, and Kruskal–Wallis tests for the ordinal categorical covariates. Total proportions for each of the variables may not add up to 1 due to rounding.

Next, I examined the results of the bivariate tests of the relationship between the experimental conditions and the respondents’ intentions to migrate. Fig. 1 shows that the pattern of expressed intentions to migrate to the United States is almost identical across the three experimental conditions. About 60%, 58%, and 59% of the respondents in the Control, Detention, and Court conditions, respectively, answered that they were “not at all likely” to go to live or work in the United States in the next 3 y. About 21% in each of the three conditions answered that they were “a little likely” to go. Another 10% in each of the three conditions answered that they were “somewhat likely” to go. About 9%, 10%, and 10% in the Control, Detention, and Court conditions, respectively, answered that they were “very likely” to go. In brief, making salient the possibility of immigration detention and lack of judicial process in the removal process did not have a significant effect on the respondents’ intentions to migrate.

Fig. 1.

Intent to migrate by experimental condition.

Deterrent effects may vary for different groups of individuals, particularly depending on their motivations and capabilities for migration. For example, punitive enforcement policies may not have a significant deterrent effect on individuals fleeing crime and violence insofar as their desire to “leave the devil they know” outweighs any risks of hardship in the receiving country (5). Likewise, individuals with family ties in the United States may be less deterrable to the extent they are motivated by family reunification, or their social capital in the United States serves as a buffer against risks of apprehension (50). By contrast, individuals seeking to migrate primarily for economic reasons may be more sensitive to risks presented by punitive enforcement policies. To explore these possibilities, I conducted a set of regression analyses to examine possible interaction effects between the experimental conditions and Victimization Count, US Ties, and Income, respectively, on intentions to migrate. The results of these analyses indicate that none of these interactions are statistically significant (at P < 0.05). SI Appendix, Figs. S2–S4 summarize these findings by showing the conditional marginal effects of the experimental conditions on US Migrate at varying levels of Victimization Count, US Ties, and Income, respectively.

Finally, to check that US Migrate was indeed capturing migration intentions, I conducted regression analyses to examine whether the covariates that prior studies have shown are key predictors of migration intentions are operating in expected ways. Table 2 presents the results (odds ratios) of ordered logistic regression analyses predicting US Migrate. As shown in Table 2, male respondents are more likely to intend to migrate, whereas older respondents are less likely to intend to migrate. The odds of intending to migrate generally increase with increasing levels of economic deprivation. Those with past US migration experience, US social ties, and crime victimization experiences in their origin countries are more likely to intend to migrate. Finally, the respondents in Guatemala, El Salvador, and Honduras, respectively, are more likely to intend to migrate than their counterparts in Mexico. All of these results are consistent with existing research findings on predictors of international migration.

Table 2.

Ordered logistic regression analysis of US Migrate (odds ratios)

	US Migrate
Variable	Model 1	Model 2
Experimental condition†
Detention condition	1.08	1.05
Court condition	1.04	0.99
Male		1.84***
Age		0.95***
Education		0.99
Income‡
Just enough		1.06
Not enough		1.42**
Hard time		1.87***
US past		2.08***
US ties§
Talk never		0.95
Talk rarely		1.75***
Talk 1 to 2x per month		2.79***
Talk 1 to 2x per week		3.65***
Talk every day		4.88***
Victimization count		1.04*
Country¶
Guatemala		1.68***
El Salvador		1.37***
Honduras		2.61***

^†Reference category is the Control condition.

^‡Reference category is Can Save.

^§Reference category is No Ties.

^¶Reference category is Mexico.

^*P < 0.05; **P < 0.01; ***P < 0.001 (two-tailed tests).

So far, I have shown that exposure to materials that are designed to heighten the salience of immigration detention and lack of judicial process in the American immigration enforcement system did not affect the respondents’ intentions to migrate to the United States. Next, I examined whether exposure to these materials had significant effects on the respondents’ legal attitudes. Recall that higher values on Procedure, Outcome, and Respect correspond to higher levels of disagreement with statements that the process is fair, the outcomes are fair, and that apprehended conationals are treated with respect, respectively. In other words, the higher the values on the legal attitude measures, the greater the perceptions that the US immigration system is lacking in procedural, distributive, and interactional justice. Table 3 presents the results (odds ratios) of ordered logistic regression analyses predicting Procedure (models 1 and 2), Outcome (models 3 and 4), and Respect (models 5 and 6). Models 1, 3, and 5 examine the treatment effect on legal attitudes without adjusting for any covariates. Models 2, 4, and 6 examine the treatment effect on legal attitudes adjusting for covariates.^‡

Table 3.

Ordered logistic regression analysis of legal attitudes (odds ratios)

Variable	Procedure		Outcome		Respect
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Experimental condition†
Detention condition	1.18**	1.17**	1.15*	1.14*	1.09	1.08
Court condition	1.08	1.08	1.08	1.08	1.00	0.99
Male		0.82***		0.83***		0.80***
Age		1.00		1.00		1.00*
Education		1.04***		1.05***		1.05***
Income‡
Just enough		1.01		1.05		1.08
Not enough		1.11		1.31**		1.24*
Hard time		1.20		1.41**		1.28*
US past		0.90		0.93		0.70***
US ties§
Talk never		0.97		0.87		1.10
Talk rarely		0.98		0.94		0.98
Talk 1 to 2x per month		0.93		0.84*		0.81**
Talk 1 to 2x per week		0.85*		0.69***		0.80**
Talk every day		0.83		0.69***		0.71**
Victimization count		1.02		1.02		1.06***
Country¶
Guatemala		0.90		0.95		1.04
El Salvador		0.69***		0.69***		0.70***
Honduras		0.72***		0.76***		0.58***

^†Reference category is the Control condition.

^‡Reference category is Can Save.

^§Reference category is No Ties.

^¶Reference category is Mexico.

^*P < 0.05; **P < 0.01; ***P < 0.001 (two-tailed tests).

Given that the differences in the treatment effects between the nested models are minimal, I will focus my discussion on Models 2, 4, and 6. As shown in Model 2, for the respondents in the Detention condition, the odds of expressing more negative assessments about the procedural fairness of the US immigration enforcement system are 1.17 times greater than those of their counterparts in the Control condition (P < 0.01). Likewise, as shown in Model 4, for the respondents in the Detention condition, the odds of expressing more negative assessments about the outcome fairness of the US immigration enforcement system are 1.14 times greater than those of their counterparts in the Control condition (P < 0.05). However, I did not find significant treatment effects on the respondents’ assessments about interactional justice. I also did not find significant treatment effects on any of the legal attitudes for the Court condition.

Discussion

Using a nationally representative survey experiment in El Salvador, Guatemala, Honduras, and Mexico (the top four sending countries of unauthorized migrants to the United States), this study shows that a heightened awareness about the widespread use of immigration detention may foster delegitimating beliefs about the US immigration system without producing the intended deterrent effects. Specifically, I find no evidence that increasing the salience of immigration detention in the minds of prospective migrants affects their intentions to migrate to the United States. This result, which allows us to draw causal inferences given the experimental nature of the study design, is generally consistent with the findings of observational studies on the lack of deterrent effect of punitive immigration enforcement policies (5, 51).

At the same time, the survey experiment indicates that awareness about the widespread use of immigration detention in the United States negatively impacts individuals’ assessments about the procedural and outcome fairness of the US immigration system. This finding extends the growing body of research that shows that immigration detention—both in the United States and in other developed countries such as the UK and the Netherlands—fosters widespread legal cynicism among immigrant detainees (14, 52, 53). Because the subjects in this study are prospective immigrants living in their origin countries (the majority of whom do not have a US migration history, as shown in Table 1), this finding shows that personal experiences with immigration detention is not necessary to generate a “legitimacy deficit” (51).

Two additional analysis results are worth noting. First, the study did not find a significant treatment effect of the Detention condition on the respondents’ assessments of interactional justice. One possible explanation for this null finding may relate to the interpersonal, rather than structural, aspect of interactional justice. People typically recognize that the nature of interpersonal treatment that they may receive in any given legal process can easily vary from one government agent or legal authority to another. For example, as Ryo (54) found in her study of unauthorized immigrants, “a common refrain in describing the U.S. Border Patrol was that there were ‘good’ people and ‘bad’ people.” By contrast, assessments about the fairness of decision-making procedures and outcomes in a legal process implicate concerns that are more structural in nature and thus less likely to be dependent on behaviors of individual government agents or legal authorities. This may explain why awareness of the prevalent use of immigration detention in the United States may trigger global judgments about procedural and distributive justice—but not interactional justice—of the US immigration enforcement system.

Second, the Court condition, unlike the Detention condition, did not produce a significant effect on legal attitudes. This result may be unsurprising given the high level of distrust and skepticism that Mexicans and Central Americans may have about courts in their own countries. Past surveys have found that individuals in Latin America exhibit a great deal of distrust toward, and skepticism about, their court systems. Research also shows that this public distrust has been growing over the years. For example, only about 41% of Latin Americans in the 2018/2019 AmericasBarometer thought the courts guaranteed fair trials, compared to 46% in 2004 (55). A high level of distrust toward courts may have led the respondents in this study to devalue or discount the importance of US immigration courts in the removal process. In other words, if the respondents’ assumption is that courts are irrelevant or even detrimental to the functioning of a legitimate legal process, then awareness about the lack of judicial oversight in the removal process would not affect their assessments about the system fairness. This possibility calls for a better understanding of whether and how legal attitudes may operate through a comparative transnational lens for prospective and current immigrants (56, 57).

It bears emphasis that this study’s findings should be interpreted with caution given its limitations. First, the lack of a significant treatment effect on intentions to migrate suggests that “deterrence hurdles” may be substantial for immigration enforcement policies. According to research on deterrence, three major types of hurdles must be overcome before deterrence can take effect: the legal knowledge hurdle, the rational choice hurdle, and the perceived net cost hurdle (58). The basic idea is that people must know the law, they must be able to conduct the relevant cost–benefit analysis required by deterrence theory, and they must conclude that the perceived costs of legal sanctions outweigh the perceived benefits of legal noncompliance. These hurdles likely operate in even more complex and unexpected ways for individuals abroad to make deterrence unlikely in the immigration enforcement context (34). Although the experimental manipulations were intended to lower the knowledge hurdle and potentially generate downstream effects on the perceived net cost hurdle, it is possible that the manipulations were too weak. For example, the succinct nature of the experimental manipulations raises the question whether a lengthier statement that includes a detailed description of US immigration detention (such as the average length and conditions of detention) could have produced a significant deterrent effect.

Second, the experimental manipulations consisted of varying statements about what has happened to many of the respondents’ conationals who were apprehended by US immigration officials. That even this limited manipulation generated significant treatment effects on the legal attitudes of the respondents in the Detention condition suggests that immigration detention has the potential to trigger strong value judgments related to justice assessments. But the study does not allow us to parse precisely what aspects of immigration detention foster negative perceptions about the fairness of the US immigration system. In addition, while this study has focused on three key antecedents (fairness evaluations about procedural, distributive, and interactional justice) of legitimacy perceptions of legal authority (25), it did not directly measure or investigate legitimacy perceptions. Under what conditions people’s fairness evaluations might shape their perceptions about the legitimacy of legal authority and decisions to engage in legal noncompliance remains an important issue for future inquiry in immigration law.

Finally, the study did not—and was not designed to—capture information about what the respondents might already know about the US immigration enforcement system and how such knowledge might spread through origin communities to shape the community members’ migration intentions and legal attitudes. Addressing these and related limitations in future research will bring us closer to understanding how US immigration policy might be operating as foreign policy that bolsters or undermines our democratic institutions in the eyes of prospective migrants.

Data Availability

Survey data and code used in this study are posted at openICPSR, https://doi.org/10.3886/E137261V1 (59).