Abstract
Excerpted From: Emily Ryo, Ian Peacock, Weston Ley and Christopher Levesque, Racial Disparities in Crime-based Removal Proceedings, 109 Minnesota Law Review 1997 (May, 2025) (184 Footnotes) (Full Document)
“Noncitizens with criminal convictions, their families, and sometimes even their attorneys, who come before the immigration court often feel like they have entered a carnival 'house of mirrors,”’ wrote Dana Leigh Marks and Denise Noonan Slavin, former immigration judges (IJs). This description underscores the frightening and maze-like nature of immigration court process that immigrants with criminal history experience when the government places them in removal proceedings based on their criminal history (crime-based removal proceedings). The experience is aptly described as entering a house of mirrors because at the heart of these removal proceedings are the crimes for which these immigrants have already been punished through the criminal justice system. Yet these crimes continue to follow them, threatening them with severe consequences that are tantamount to double punishment.
This Article offers the first systematic empirical analysis of crime-based removal proceedings. This analysis is important and timely given that current U.S. immigration law and enforcement policy have become heavily focused on detaining and deporting immigrants with criminal records. In fiscal year 2008, immigrants without past criminal convictions accounted for the majority (69%) of those removed after an arrest by U.S. Immigration and Customs Enforcement (ICE). By contrast, in fiscal year 2020, immigrants with past criminal convictions or pending criminal charges accounted for the majority (64%) of all ICE removals; this percentage is even higher (92%) when we consider only those removals resulting from interior (as opposed to border) arrests.
To appreciate the consequential nature of crime-based removal proceedings and their significance in the intersecting world of immigration and criminal law, it is helpful to consider the following three salient facts that define the contemporary U.S. immigration deportation regime. First, deportation is a devastating life event with enduring negative consequences for immigrants, their families, and communities. According to the U.S. Supreme Court, deportation of a long-term resident is tantamount to banishment from their home and family, a “punishment of the most drastic kind.” For those fleeing from violence and persecution in their origin countries, deportation may be a death sentence. Second, contacts with the criminal justice system make an immigrant, including lawful permanent residents, dramatically more likely to be detained and deported from the United States. According to one recent estimate, a criminal conviction makes deportation hundreds of times more likely. Third, there is no right to government-appointed counsel in immigration proceedings, and most immigrants in removal proceedings lack legal representation. Although there is a growing movement in various local jurisdictions to provide low-cost or free legal services to immigrants facing removal, such programs often exclude those with criminal convictions.
The foregoing discussion suggests that immigrants with criminal histories are one of the most stigmatized and vulnerable, yet the least legally protected, groups in our justice system. In addition, many are likely to be long-term U.S. residents with deep social ties to the United States. Different strands of research have started to examine issues affecting this population. For example, scholars from multiple disciplines have studied how immigrants are treated in the criminal justice system. A growing number of scholars have also examined the use of criminal law and the criminal law apparatus as an immigration enforcement tool. Yet, there is a great deal that we still do not know about the immigration consequences of a criminal record in immigration court cases and the access-to-justice challenges that immigrants face when they are placed in removal proceedings due to their triggering criminal history.
This Article is the first to ask--and address through systematic empirical analysis of removal proceedings decided between 1998 and 2023--three key questions that are fundamental to understanding the crime-based removal system in the United States. First, what legal disadvantages do immigrants in crime-based removal proceedings face, and have these disadvantages diminished or grown over time? To address this question, we descriptively examine the differences between proceedings with criminal charges and those without criminal charges in terms of the respective groups' (1) rate of legal representation (“representation rate”), (2) rate at which immigrants are held in immigration detention (“detention rate”), (3) rate of release from detention (“release rate”), and (4) rate at which removal orders are issued (“removal rate”).
Second, are there racial/ethnic disparities in legal outcomes for immigrants in crime-based removal proceedings? To address this question, we undertake a two-step analysis of our originally compiled removal-proceedings data. We first identify the top five most commonly occurring criminal charge categories in the proceedings data. These top five charges are: (1) Controlled Substance, (2) Controlled Substance Trafficking, (3) Domestic Violence, Stalking, or Child Abuse, (4) Firearms, and (5) Violation of Protective Order. Then within each criminal charge category, we analyze racial disparities in release and removal rates. Our analysis focuses on predicted values from regression models that adjust for a variety of characteristics related to immigrant respondents, proceedings, and judges. The two-step analytical approach that we undertake ensures that we are not comparing the legal outcomes of different groups in a pooled sample that aggregates widely disparate criminal charges that are of varying degrees of seriousness.
Third, does an IJ's race matter in crime-based removal proceedings? More specifically, do immigrants who share the same racial identity as the IJs who preside over their crime-based removal proceeding experience more favorable legal outcomes than those whose racial identities differ? We address this question by examining the release and removal rates of crime-based removal proceedings for which there is racial concordance between immigrant respondents and their judges (as compared to those proceedings that lack such racial concordance). For the reasons discussed earlier, we also conduct this analysis separately for each of the top five most commonly occurring criminal charges in the proceedings data.
To preview our results, we find that immigrants in crime-based removal proceedings (as compared to those in non-crime-based removal proceedings) are substantially less likely to obtain legal representation, more likely to be detained, less likely to be released from detention, and more likely to be ordered removed. We refer to these disadvantages as “double penalties,” following scholars who have argued that harsh immigration consequences of criminal conviction amount to punishing an individual twice for the same crime. Our analysis shows that these double penalties are large and have grown over time, especially in more recent years, particularly towards the end of the first Trump administration and into the Biden administration.
We also find evidence of significant racial disparities in the rates at which immigrants are released from detention, and the rates at which they are ordered removed. Among those with controlled-substance and controlled-substance trafficking charges, Hispanic immigrants have the lowest rate of release and the highest rate of removal. Among those with domestic-violence and firearm charges, Black immigrants have the lowest rate of release from detention and the highest rate of removal. Finally, we find significant positive effects of racial concordance on release and removal rates for non-white immigrants. Specifically, Asian, Black, and Hispanic immigrants generally experience more favorable outcomes if their presiding judges are of the same race. On the other hand, white immigrants fare better before non-white judges, in part because while all judges decide more favorably when the immigrant respondent is white rather than non-white, some non-white judges appear to evidence a relatively higher degree of white favoritism than white judges.
This Article makes three major contributions. First, by shining a light on what happens to immigrants who become ensnared in the immigration enforcement system as a result of their criminal history, we explore an important set of collateral consequences of contacts with the criminal justice system that has escaped systematic empirical scrutiny. A longstanding scholarship has examined the extent and nature of a variety of deleterious downstream effects of contacts with the criminal justice system on outcomes as varied as employment, occupational licensing, voting, and civic engagement. However, empirical studies that examine immigration-related collateral consequences on noncitizens of their contacts with the criminal justice system are scarce. We argue that filling this knowledge gap is critical for a fuller and more nuanced understanding of expanding powers of the carceral state and the disproportionate impact that such an expansion has on marginalized communities of color.
Second, no study to date has analyzed the centrality of immigrant respondents' race in understanding disparities in legal outcomes in removal proceedings. The influence of extralegal factors on IJs' decision-making has been of enduring interest to scholars. Prior research has found that in immigration court adjudication, demographic profiles and work experiences of IJs matter, and so do state and local contexts in which immigration courts are located, as well as outside influences and pressures placed on IJs. But this is the first study that investigates whether and to what extent racial identity of immigrant respondents might matter in removal proceedings. Our empirical findings uncovering racial disparities in immigration court outcomes underscore the importance of examining race as a “master status” that might perpetuate inequalities in immigration adjudication. In this way, our empirical findings make a significant new contribution to the growing body of scholarship on immigration courts and access to justice for immigrants.
Third, the current study is the first to provide empirical evidence of racial-concordance effects in immigration proceedings and to propose possible explanations for the phenomenon that requires future investigation. Across many different social contexts, voluminous research documents the role that ingroup bias or ingroup favoritism--the tendency to favor members of one's own group over those in other groups--plays in shaping human behavior. However, the mechanisms driving this behavior are not well understood. One possible explanation for the racial-concordance effect that we find in this study might be judicial empathy. Prior research on judicial behavior suggests that judicial empathy--the ability to understand and identify with others' perspectives and situations--may be an important determinant of how judges decide cases. Empathy, in turn, has been conceptualized as a product of personal experiences and interactions that typically vary along racial and gender lines. We draw attention to the importance of studying these kinds of dynamics. More broadly, we call for a better understanding of how judge characteristics interact with respondent characteristics to shape legal outcomes in removal proceedings.
The rest of this Article proceeds in four major parts. Part I provides the historical and legal context necessary for understanding how crime-based removal proceedings gained prominence and how they operate. Part II explains what motivates the current study's focus on racial disparities and why examining this dimension of crime-based removal proceedings is important for both research and policy. Part III describes the data on removal proceedings that we originally compiled for the purposes of this study. This Part also explains the analytical strategy that we adopted in analyzing these proceedings data. Part IV presents the results of our empirical analyses and the key findings that address the three questions that we presented earlier. We conclude with a discussion about the policy implications of our findings and directions for future research.
[. . .]
(a) The main results in our study rely on linear probability models (LPMs). While LPMs offer distinct advantages, they also suffer from limitations documented elsewhere. We re-estimated our models using logistic regression to test for the sensitivity of results to this choice of functional form.
(b) Geography represents a potential confounder of our results. Criminal statutes and binding caselaw vary across states and federal courts of appeal. This is potentially an issue for our analysis insofar as racial groups in removal proceedings are neither randomly nor equally distributed across these jurisdictions. Our main analysis uses judge fixed effects and judges are largely nested in the same jurisdictions. This means that our models essentially control for the effect of jurisdiction-specific factors that do not vary over time. However, to address more directly the issue of variations in applicable laws across jurisdictions, we also re-estimated the models in our main analysis using fixed effects for the federal court of appeals under whose precedent a given proceeding was decided.
(c) Removal proceedings can last many months or even years. Sometimes this means that more than one judge can end up presiding over the hearings for a given proceeding. This means that the race of a judge at the completion of a given proceeding may not have been the same throughout the entire proceeding. To address the possibility that this potentially biases our results, we re-estimated the models in our main analysis by including only those proceedings for which the judge never changed across all hearings.
(d) Judges in our data vary in the total number of proceedings over which they have presided, partly as a function of their time on the bench and the speed with which they decide proceedings. This introduces the possibility that judges with high volumes of completed proceedings may have an outsized influence on our results. To address this possibility, we re-estimated the models in our main analysis by including weights to reweigh proceedings to be inversely proportional to the total number of proceedings completed for a given judge.
(e) In conducting our main analysis, we did not consider whether criminal charges had been sustained. Although there might be good reasons to analyze only those proceedings that have sustained charges, there are significant drawbacks to doing so. First, focusing on sustained charges would significantly reduce the size of our analytic sample. As the relationships we explore here involve the intersection of many variables, having an adequate sample is key to our analysis. Second, the EOIR Data do not indicate when any given charge might have been sustained. This means that we cannot know whether the charge was sustained prior to or following an immigrant's release from detention, which complicates our analysis of release from detention as an outcome of interest. Nonetheless, to address the possibility that our main results are biased by not considering the status of the charges, we re-estimated our models using a sample consisting only of proceedings in which charges had been sustained.
(f) For our main analysis, we treated proceedings that lacked records in both the Reps Assigned Table and the Hearings Table as lacking legal representation (i.e., Represented = 0). To consider the possibility that some of these proceedings did in fact have legal representation, we re-estimated the models in our main analysis after excluding these proceedings.
(g) Some of our main analysis examines the effect of IJ Race on release and removal rates. For the vast majority of judges (and proceedings), IJ Race captures judges' own self-reported race. For a very small number of judges without self-reported race, we predicted their race using surnames. Given the possibility that our use of the surname method could bias our results, we re-estimated the models in our main analysis by including only judges with self-reported values for IJ Race.
Emily Ryo, JD, PhD; Charles L. B. Lowndes Distinguished Professor of Law & Professor of Sociology, Duke Law School.
Ian Peacock, PhD; JD Candidate, University of Chicago Law School.
Weston Ley, Research Associate, Duke Law School.
Christopher Levesque, PhD; Assistant Professor of Sociology and Law & Society, Kenyon College.
