Authors: Linda Camp Keith, Jennifer Holmes, and Banks Miller
Publication: Law & Policy 35: 261-289
Original source: lp_final.pdf
Conversion note: this Markdown was generated by text extraction from the hosted file. Review equations, tables, notes, and citations before treating it as a final web copy.
Explaining the Divergence in Asylum Grant Rates among Immigration Judges: An Attitudinal and Cognitive Approach LINDA CAMP KEITH, JENNIFER S. HOLMES, and BANKS P. MILLER In seeking to understand the variation in asylum grant rates by immigration judges (IJs), we apply a variation of the attitudinal model that we modify by incorporating a cognitive model of decision making, arguing that some pieces of information before IJs are treated objectively while others are treated subjectively. This model allows us to account for informational cues that influence decisions while assessing the impact of national interests and human rights conditions. We find that IJ policy predispositions play a dominant role, and that liberal IJs respond to applicant characteristics differently than conservatives, but also that the law constrains decision making. Scholars of the asylum system in the United States have observed that there is a vast discrepancy of grant rates among individual immigration judges (IJs). To date, no one has advanced and tested a theory of why there is such dramatic variation in grant rates. In this article, we introduce and test a theory that incorporates insights from international relations and the judicial decision-making literature. Our major contribution to understanding the decision making of IJs within the U.S. asylum system is to provide a theory for explaining the widely divergent grant rates of IJs, and then to test that theory in a methodologically sophisticated manner. A secondary contribution of our article is to apply an explicitly cognitive model of judicial decision making to the behavior of IJs. We test these theories using U.S. asylum decisions made by IJs between 1997 and 2004. We find that the policy predispositions of the IJs play a dominant role in explaining the observed discrepancies in asylum grant rates and that liberal IJs respond to certain The authors acknowledge the financial support of the Overbrook Foundation for portions of this project. The authors also wish to thank immigration attorney Paul Zoltan and the staff attorneys at HRI for their insights into the asylum process. The authors also wish to thank colleagues Tom Brunell and Clint Peinhardt for comments and suggestions on an earlier version of the manuscript. Address correspondence to Banks P. Miller, University of Texas at Dallas—Political Science, 800 W. Campbell Rd., GR 31, Richardson, Texas 75075, USA. Telephone: (972) 883-6297; E-mail: millerbp@utdallas.edu. bs_bs_banner LAW & POLICY, Vol. 35, No. 4, October 2013 ISSN 0265–8240 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary doi: 10.1111/lapo.12008
applicant characteristics differently than their more conservative colleagues. We also find some limited evidence that the law constrains the decision making of the IJs with respect to applicant characteristics. Recently, observers of both U.S. and Western European asylum policy have noted the hostile asylum environment in which economic and security concerns compete with humanitarian concerns (Levy 2005; Hassan 2000; Schuster 2000). In doing so, they have reintroduced the normative issue of whether foreign policy should influence asylum outcomes and thus reinvigorate the empirical debate among international relations (IR) scholars. The asylum debate reflects the broader IR debate over whether states’ human rights behavior is primarily shaped by strategic and material interests, or by international human rights norms. Most international relations–oriented asylum work assumes a unitary actor, focuses on aggregate levels of asylum granted to persons fleeing various states, and centers on the relationship between the United States and the sending states. Empirical studies of asylum outcomes suggest that both national interests and normative concerns influence asylum outcomes; however, these studies leave unanswered the question of how the individual judges making the actual decisions are influenced by these competing factors (Holmes and Keith 2010; Rottman, Fariss, and Poe 2009; Salehyan and Rosenblum 2008; Rosenblum and Salehyan 2004). At the same time, law professors and the media have scrutinized immigration courts and the process of asylum seeking in the United States. This attention has been fueled by “federal appeals courts around the country complain[ing] of a pattern of biased and incoherent decisions on asylum and rebuk[ing] some immigration judges by name for ‘bullying’ and ‘brow-beating’ people seeking refuge from persecution” (Bernstein 2006; see also Legomsky 2007). The criticism has been compounded by reports of significant variation in grant rates across judges, even among those serving on the same courts (Ramji-Nogales, Schoenholtz, and Schrag 2007, 2009; Legomsky 2007). For example, the U.S. General Accounting Office (GAO) noted that, in 2008, the likelihood of receiving a grant of asylum was 420 times greater than the likelihood of receiving asylum from the IJ least likely to grant asylum to the IJ most likely to grant asylum in the same court (GAO 2008). The U.S. Commission on International Religious Freedom concluded that outcomes of individual asylum claims have come “to depend largely on chance; namely, the IJ who happens to be assigned to hear the case” (2008, 115). These disparities across courts and judges raise significant questions about the quality and consistency of justice in immigration courts, and for many observers and legal practitioners they present a frustrating dilemma. One law professor responded to this frustration by suggesting that law professors should turn to political science and its models of judicial behavior (Taylor 2007). We take up that challenge and apply a variation of the attitudinal model from the judicial behavior literature, which we then modify by taking a cognitive approach to decision making that allows IJs to filter or weight specific legal factors to a greater or lesser extent. In our cognitive model, 262 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
some pieces of information before a decision maker, the IJ, are treated more objectively (i.e., they constrain decision making) than other pieces of information, which are treated subjectively. 1 This model allows us not only to account for the informational cues that influence a judge’s decisions, but also to assess the impact of national interests and human rights interests conditional on a judge’s policy preferences toward asylum, while controlling for a variety of factors that may influence the judges’ decisions. We show that the probability of any asylum seeker being granted asylum is strongly influenced by the random assignment of IJs to cases, which implicates norms of fairness and consistency. NORMS VERSUS INTERESTS IN ASYLUM OUTCOMES Recent debates over U.S. asylum decisions nest within the broader “norms versus interests” debate. The debate typically frames interests to include diplomatic, security, economic, and domestic demands (such as migration control), and defines normative goals to be those that “reflect a truly humanitarian, even altruistic concern with the needs of asylum applicants, regardless of U.S. relations with their country of origin” (Rosenblum and Salehyan 2004, 681). 2 Realists and rational functionalists perceive states as rational actors whose behavior is based primarily upon narrow self-interest (Keohane 1984; Waltz 1979). State commitments to international humanitarian or rights norms are perceived as “cheap talk” (Mearsheimer 1994) that give way to more substantive state interests. Decisions about whether to grant asylum to another state’s citizens will be shaped by the receiving state’s calculation of the benefits and costs of accepting or rejecting the asylum seekers. In regard to asylum policy or outcomes, realists expect that the humanitarian promise of asylum would be trumped by geopolitical concerns such as issues of national security, economic relations, and domestic politics. Conversely, constructivists emphasize the emergence and diffusion of international human rights norms through networks of domestic and transnational actors who shape the discourse of international human rights and rally publics to convince states to formally accept and to adhere to these norms (e.g., Risse, Ropp, and Sikkink 1999; Keck and Sikkink 1998). In addition, states are expected to comply with these norms because they tend to comply with their legal obligations (Henkin 1979) or because they aspire to comply with the norm of pacta sunt servanda (agreements must be kept) (Chayes and Chayes 1993). Thus, constructivists would expect that, given the formal commitments to the international norm of nonrefoulement, legal considerations linked to refugee status should shape asylum outcomes more than national interests. 3 The early asylum literature in political science demonstrated the dominant role of geopolitical and material interests over human rights in U.S. asylum policy (Loescher and Scanlan 1998; Gibney, Dalton, and Vockell 1992; Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 263 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
Gibney and Stohl 1988). More recent studies (Holmes and Keith 2010; Rottman, Fariss, and Poe 2009; Salehyan and Rosenblum 2008; Rosenblum and Salehyan 2004), suggest that both normative concerns and national interests influence these outcomes. Rosenblum and Salehyan (2004) shift the focus away from the either-or assumption to focus instead on the relative importance of these two sets of values (presumably to U.S. foreign policy makers), recognizing the potential for them to complement as well as contradict each other. We take this recognition as our point of departure for our analysis. We generate two hypotheses and shift to the micro level of analysis in order to examine the individual decision makers—IJs—and their decision-making processes. Hypothesis 1: IJ decisions are influenced by the judge’s policy predispositions toward immigration issues: judges with liberal policy preferences will be more likely to grant asylum than will judges with conservative policy preferences. In addition, we argue that the nature and context of asylum decision making shapes the processes IJs rely upon to reach decisions. Ultimately, we argue that the approach they employ for objective legal considerations (such as human rights conditions) is different from the one through which they process the more abstract extralegal factors (such as national interests). Thus, “norms” (human rights conditions) will be processed differently than material and security “interests” by these decision makers. Hypothesis 2: The effect of extralegal factors, such as material and security interests, will be more influenced by the policy preferences of the IJs than will be objective legal considerations, such as human rights conditions. We turn next to a discussion of IJs and the institutional context in which they make asylum decisions. U.S. IJs IJs are administrative adjudicators who are formally appointed by the U.S. deputy attorney general; however, the Executive Office of Immigration Review (EOIR) and the chief IJ handle their hiring. 4 In most years, there are over 230 IJs in over 55 immigration courts in 23 states and Puerto Rico. IJs arguably have less formal independence than federal judges and potentially less independence than administrative law judges. Nonetheless, they maintain a high degree of independence. The Immigration and Nationality Act (1952) states that “in deciding the individual cases before them . . . IJs shall exercise their independent judgment and discretion” (§ 240). Despite their lack of life tenure, IJs experience a good deal of autonomy in their decision making because of (1) the large volume of cases they decide (approximately three times the number decided by a typical federal district court judge), 5 264 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
(2) the low probability of reversal by the Board of Immigration Appeals (BIA)6 and the federal circuit courts,7 and (3) the standard of appellate review to which the IJs are subjected: reasonableness. This exceedingly low standard requires that the BIA find that the IJ reached unreasonable conclusions in order to reverse a decision where the facts of a case are reviewed for clear error by the BIA, whereas, application of the law is reviewed de novo. In essence, IJs are judges-as-bureaucrats, with ample discretion and broad civil service protections (see, e.g., Brehm and Gates 1997). The legal strictures in asylum cases are loose because both the facts and the law are vague (Baum 2010; Legomsky 2010). Law (2005) notes “the indeterminacy of the governing legal standards” in asylum cases, which leaves judges “to define vague yet crucial terms-‘political,’ ‘persecution,’ ‘well-founded fear,’ ‘more likely than not’-on a case-by-case basis” with “precedent provid[ing] only limited guidance, given the dependence of asylum claims on case-specific facts” (830). As Martin (2000) observes, the “basic facts in any particular [asylum] case are highly elusive” and “the adjudicator has to decide what happened in a distant country” with only two imperfect sources: general human rights country reports and the personal testimony of the asylum seeker (3). Alexander (2006) notes that immigration courts routinely lack evidence: “the witnesses, objects, and documents that could prove or disprove a fear of persecution, for example, are likely beyond reach overseas” and “indeed, the ability to gather evidence may be blocked by the very government alleged to be the persecutor” (19). In fact, IJs have been criticized by federal court of appeals judges for an exaggeration of the availability of evidence in many countries of origin (for example, see Judge Posner in the Seventh Circuit opinion in Iao v. Gonzales 2005, 533–35). Further compounding factors include the reality that over three-fourths of the applicants do not speak English (any one of 389 languages may be spoken at the hearings), and that most applicants are not represented by counsel, requiring judges to guide lawyerless individuals through the proceedings, advise them of their rights and options for relief of removal, and answer questions (Transactional Records Access Clearinghouse [TRAC] 2009; Alexander 2006). In addition, IJs face serious institutional hurdles such as a heavy workload, 8 a growing case backlog, 9 and little staff assistance. 10 IJs complain about “the constant drumbeat of case completion goals,” the requirement that judges must “rule promptly at the end of the hearing in the form of lengthy, detailed and extemporaneous oral opinion with little or no time to reflect or to deliberate,” that making credibility determinations “is extremely, extremely difficult,” and most importantly, that “there is not enough time to do research and adequately read about country conditions” (Lustig et al. 2008, 65–66). Clearly, these are overburdened and underresourced courts with high stakes for applicants. The president of the judges association recently complained that “for some people, these are the equivalent of death penalty cases, and we are conducting these cases in a traffic court setting” (as cited in Becker and Carbrera 2009). Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 265 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
The most frequent criticism of the IJs has been the disparities in grants and denials among them (e.g., Ramji-Nogales, Schoenholtz, and Schrag 2009; TRAC 2006), which suggest the influence of the judges’ policy preferences or personal biases. Law professors argue that subjects such as asylum, which “inspire ideological or emotional fervor would seem to have the greatest disparate outcomes, since the flesh-and-blood adjudicators who decide the cases will have extra reason to resolve the more indeterminate questions by resort to visceral beliefs and emotional impulses” (Legomsky 2007, 442). Law specifically notes that asylum cases invoke highly charged political issues and is “an area of law that divides the Ninth Circuit along ideological lines, a fact acknowledged by the judges themselves and borne out by the data” (Law 2005, 830). The nature and the context of these decisions lead to reliance on policy preferences in decision making. The lack of certainty in legal standards, the frequent lack of corroborating evidence, and the complexity of credibility determinations invite the use of motivated reasoning in which directional (or policy-motivated) goals overwhelm accuracy goals (Braman and Nelson 2007). The prominence of these directional motivations is important to our cognitive approach to judging because these directional or policy goals will tend to induce what Bartels (2010) has termed top-down decision making. In a top-down approach, the judge brings a theory or predisposition to the case—this is analogous to the ideological proclivities that U.S. Supreme Court Justices bring to cases. This top-down approach is contrasted with a bottom-up approach in which facts are evaluated deliberately, and decision making is constrained by the manner in which the law dictates the treatment of those facts. We develop these concepts further in the following section. An additional contextual factor—the overwhelming workload of IJs— further induces reliance on policy predispositions to decide cases and likely increases IJs’ reliance on informational cues to decide cases (Baum 2010). Thus, rational litigant behavior is difficult because of the unpredictability of the BIA on appeal, since it will be difficult to predict in any given case how the reviewing authority would treat cases with highly ambiguous facts and few legal constraints. This means that institutional control of these adjudicators is diminished, since rational litigant behavior is a necessary precondition for bringing cases deviating from the principal’s preferences that will allow it to control agents lower in the hierarchy (Songer, Cameron, and Segal 1995). We turn to the judicial behavior literature to further explore these issues. JUDICIAL BEHAVIOR AND JUDGING IN ASYLUM CASES Scholars of judicial behavior have tended to focus their attention on three interrelated sets of explanations for judicial choice: attitudinal, strategic, and legal models. In brief, the attitudinal model holds that judges primarily 266 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
pursue their policy preferences (e.g., Segal and Spaeth 2002); the strategic model holds that judges will pursue their policy preferences in light of the preferences of other important actors, such as other judges on the same court, the court above them in the judicial hierarchy, or other relevant institutions (e.g., Epstein and Knight 1998); and the legal model assumes that judges’ choices are constrained by the dictates of precedent, the plain meaning of statutes, and the intent of the framers of statutes (e.g., Gillman 2001). Of these models, we focus on the attitudinal model for reasons explained above and explored further below. To this explanation, however, we add an explicitly cognitive approach. In essence, we posit that the policy preferences of IJs influence their decisions in asylum cases, but that U.S. asylum law also imposes some constraints on the use of the policy proclivities of an IJ. We believe that a comprehensive model of IJ decision making approximates a situation in which an IJ, under tremendous time pressure and unsure of the credibility of an asylum seeker, will use policy predispositions to help process both legally relevant and legally irrelevant facts. The cognitive approach is operationalized in our model as the differential weights placed by an IJ on particular facts, which we believe will vary with the policy predispositions of the judge. A similar approach has been usefully summarized by Bartels as a mix of top-down and bottom-up approaches. After summarizing the attitudinal aspects of asylum decision making, we integrate an attitudinal approach with this cognitive process. As noted above, asylum decision making is likely to activate ideological schema, making the attitudinal model an appropriate choice by predicting that liberal IJs are more likely than conservative IJs to grant asylum. The asylum decision is likely to be dichotomous in nature making a compromise position unavailable: a person is either allowed to stay in the United States or not, and coupled with infrequent appeal and reversal, the decisions have an air of finality (Baum 2010). Therefore, judging these cases with policy proclivities in mind is relatively easy, since asylum cases are likely to activate predetermined schema and the choice is dichotomous. Moreover, as discussed earlier, IJs have much discretion in their decision making, given the voluminous caseload, low likelihood of appellate review, and the high threshold of the reasonableness standard. We believe, given these facets of decision making in the U.S. asylum process, that an IJ’s policy predispositions will be highly determinative of his or her likelihood of granting asylum. The applicability of strategic and legal approaches must be assessed for asylum decision making. Strategic decision making is less likely in asylum cases than it is in other kinds of decision contexts. Reversal by the BIA is unlikely, with only 11 percent of all asylum decisions reversed. Additionally, the difficulty of predicting what the reviewing authority will do with ambiguous facts in any specific case inhibits hierarchical control. Moreover, IJs do not decide cases in panels and, therefore, have little to fear from other IJs in terms of review. The vagueness and ambiguity of asylum cases also lessons applicability of the legal model to the decision making of IJs (Baum 2010; Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 267 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
Legomsky 2010). Nevertheless, we do find some evidence that IJs make decisions that reflect the assessment of legally relevant factors. We know from the literature that asylum grant rates are influenced by extralegal factors such as U.S. military aid or trade relations (Rottman, Fariss, and Poe 2009; Rosenblum and Salehyan 2004), or potential indicators of a bogus asylum seeker (Keith and Holmes 2009). In the European Union, Neumayer (2005) finds divergent recognition rates for applicants of the same nationality and doubts that the differences are due to individual case factors. The work of Bohmer and Shuman (2008) suggests that asylum adjudicators in both the United States and the United Kingdom perceive evidence and facts through limited cultural filters. In addition, the disparities in grant and denial rates among judges, even within the same court (e.g., Ramji-Nogales, Schoenholtz, and Schrag 2007; TRAC 2006), suggest the influence of the individual judge’s policy preferences or personal biases. Our expectation is that these extralegal factors will tend to be emphasized differently depending on an IJ’s policy predisposition. As noted, we adopt a cognitive approach to decision making in asylum cases to explain this variation. Essentially our approach is that IJs use a mixed process—in which their policy predispositions are moderated in certain circumstances but not in others. Bartels (2010) describes this type of process as a continuum ranging from totally top-down decision making, in which the predispositions that a judge brings to a case predominate any new information he or she encounters, to totally bottom-up decision making, in which a decision maker objectively scrutinizes the facts before him or her. We believe that the “norms” factors, such as the level of human rights abuse reported in a country, are likely to be evaluated in a more bottom-up fashion (i.e., somewhat objectively), 11 whereas, material and security concerns, such as whether the applicant is from a country producing high numbers of illegal immigrants, are likely to be evaluated in a more top-down fashion (i.e., somewhat subjectively). Put differently, IJs will assess human rights conditions in a manner that is less contingent on their own policy preferences than their assessment of material and security concerns, which will be more contingent on their policy predispositions. This hypothesis is premised on three important facts about IJ decision making. First, asylum law dictates that IJs should consider the likelihood of persecution should an applicant be returned to a country (Legomsky 2010), but the law does not direct judges to consider the material or foreign policy interests of the United States in making asylum determinations. Second, given the severe time constraints under which IJs make decisions, their search for information about a particular applicant is limited (Lustig et al. 2008). Therefore, the IJ in a case is heavily reliant on U.S. State Department reports about an applicant’s country, and these reports are almost solely concerned with the human rights situation in a country (Bohmer and Shuman 2008). Thus, the fact that the law dictates consideration of human rights conditions, and the fact that this evidence is some of the most readily available evidence 268 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
in a case, means human rights conditions should be evaluated in a more bottom-up fashion than material or security concerns. In addition, factors such as the level of trade between the applicant’s country and the United States—the type of thing that the literature refers to as a material interest— are not presented to IJs and are not supposed to be considered in determining an asylum application. Therefore, these types of factors are likely to be evaluated in a manner that is closer to the top-down end of the decisionmaking continuum—that is, treatment of this type of variable should be dependent on a judge’s policy proclivities. Third, IJs are trained by the EOIR to evaluate and make credibility determinations with respect to the likelihood of persecution should an applicant be returned to their country of origin, but they are not trained to make their determinations on the basis of any other factors (see Immigration Judge Benchbook 2009). Psychological approaches to decision making indicate that two interrelated factors will tend to make decision makers approach particular facts from the bottom up: accountability and fear of invalidity. When decision makers must justify their decisions to someone else (when they are accountable), they tend to be more objective in their decision making (Lerner and Tetlock 1999). In addition, when decision makers must fear invalidity (in this case being overturned) they tend to take a more bottom-up approach to their decision making (Fazio and Townes-Schwen 1999). Importantly, because review of an IJ’s decision is to be premised on asylum law, and because asylum law dictates consideration of certain facts but not others, these two factors, accountability and fear of invalidity, should work to make evaluation of human rights concerns more objective than other factors. To summarize, we theoretically derive two major hypotheses about the decision making of IJs in asylum cases. First, asylum implicates clearly identifiable and emotionally compelling policy choices. Given the fact that review of IJ decisions is unlikely, and that when it does occur it will not be predictable, we believe that IJs will use their policy predispositions to guide their decision making. Specifically, IJs with liberal policy preferences will be more likely to grant relief than will IJs with conservative policy preferences. Second, the assessment of extralegal factors, such as material and security interests, will be more subjective than the treatment of legal factors such as human rights conditions. Specifically, we expect the effect of legal factors will not vary significantly across the liberalism of IJs, whereas the effect of extralegal factors will vary across the IJs’ liberalism. Together, we show that these two hypotheses explain a good deal of the observed variation in IJ decision making. DATA DESCRIPTION We examine 329,101 asylum cases decided by IJs from 1997 to 2004. This period captures cases that were decided between two significant changes in Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 269 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
immigration law: the 1996 Illegal Immigration Reform and Immigrant Responsibility Act (IIRIRA), whose most significant changes include “a new ‘expedited removal’ procedure and ‘credible fear’ screening for asylum seekers arriving at ports of entry with no documents or with documents suspected to be false” (Anker 2011,11), as well as increasing the grounds for mandatory detention, and imposing “a one-year deadline on applying for asylum, delay in work authorization eligibility, and prompt adjudication of asylum applications” (Cianciarulo 2006, 110–11) and the 2005 Real ID Act, which “requires asylum seekers to demonstrate that their race, religion, nationality, membership in a social group, or political opinion represents a ‘central reason’ for the persecution they suffered or fear” (Kerwin 2005, 756–57). 12 Our analysis is limited by data restrictions. First, IJs do not publish their opinions. Second, Congress has placed strict privacy restrictions on asylum cases. The EOIR will only release the form of relief requested, the case outcome, the applicant’s country of origin, language spoken, and whether or not the applicant had a lawyer. DEPENDENT VARIABLE Our unit of analysis is the individual case decision. See the online appendix for a complete description of the data and sources (http://www.utdallas.edu/ ~lck016000/research.htm). Only cases decided on the merits are included. Following the asylum literature, we code outcomes as one in cases in which the judge makes a grant of asylum, a conditional grant, or some other form of relief. Denial of any form of relief is coded as zero. IJ POLICY PREFERENCES The judicial behavior literature portrays judicial votes on rights claims as a liberal-conservative dichotomy, with support of rights claims considered liberal and support of the government interests over the rights claims considered conservative (e.g., Segal and Spaeth 2002). This dichotomy has been applied to antialien/proalien behavior in federal immigration cases (e.g., Williams and Law 2012; Ocepek and Fetzer 2010; Westerland 2009; Law 2005). Following the asylum and judicial behavior literatures, we believe that the ideological divide on asylum issues does parallel that of other immigration issues, with the antiasylum viewpoint reflecting more closely conservative or right-wing perspective, and the proasylum viewpoint reflecting more liberal or left-leaning perspective of both U.S. and European politics. However, we also feel that a more specific measure of ideology may be more apropos in this instance for the reasons we discuss below. To measure judicial policy preferences, studies of judges below the level of the U.S. Supreme Court have turned to surrogates, such as the party affiliation of the judge, or the appointing president’s party affiliation (e.g., Ocepek 270 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
and Fetzer 2010, specifically in regard to immigration issues; George and Epstein 1992; Tate and Handberg 1991), measures of the appointing president’s liberalism (Segal, Howard, and Hutz 1996), or presidential ADA scores (Zupan 1992), which have been highly successful in predicting the liberalism of judges’ subsequent votes. Information about the party affiliation of IJs is unavailable. We conducted considerable research, looking for potential identifying links, such as contributions to political parties or candidates, and even utilized the exact string searches that Monica Goodling et al. had followed during the Bush administration to vet potential Department of Justice hires (U.S. Department of Justice Office of the Inspector General 2008). However, we found almost no party links to judges. A reasonable alternative approach suggested by the literature is to simply include the ideology of the appointing president. Yet others have argued that a general ideology measure may not be useful in studies of specialized courts or in technical cases (e.g. Miller and Curry 2009; Baum 1994; Jucewicz and Baum 1990). Further, given the self-selecting nature of IJ hiring, presidential ideology is not likely to be a useful surrogate of IJs’ ideological leanings in regard to asylum. Instead, we create a more narrowly focused measure that provides an approximation of a judge’s policy predisposition specifically toward immigration issues, which we believe is more useful for this study than a general indicator of ideology, such as party of the appointing president, given that IJs are not typically visible appointments for a president. 13 Furthermore, our more narrow measure of ideology avoids a problem that some critics argue is endemic to the measure of ideology generally: “judicial ideology may be a multidimensional phenomenon, such that a judge who is liberal in one context may be moderate or conservative in another, or the labels ‘liberal,’ ‘moderate,’ and ‘conservative’ may not seem applicable at all” (Fischman and Law 2009, 136). By creating a measure that is specific to one area of law—here asylum decision making—we lessen the potential for multidimensionality in preferences to obscure our findings. Indeed, previous studies of asylum decision making in Canada have utilized both background characteristics and measures of ideology as determined by experts in asylum and immigration law in attempting to discern judges’ policy predispositions (Gould, Sheppard, and Wheeldon 2010). Other early studies have also turned to the judges’ background characteristics, arguing that they ‘reflect similar socialization processes and life experiences, which in turn produce similar attitudes and ultimately behavior (votes)’ (Gryski, Main, and Dixon 1986, 528). Tate and Handberg posit that “the kinds of personal experiences and exposure the judges have had during their lives” serve as “indicators or clues about life experiences and judges perspectives toward cases” (1991, 460). Moreover, attorneys with certain values and beliefs tend to be drawn to certain careers, such as an immigration-related nongovernmental organization (NGO), a position in corporate law, or in academia. In addition, there is strong evidence that the background characteristics that we consider in this Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 271 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
article have been used by key actors in the asylum decision-making bureaucracy specifically as cues for the likely policy preferences of asylum adjudicators. For instance, when Janet Reno sought to increase the size of the BIA in the 1990s, she was careful to select members who had backgrounds in academia, private practice, and advocacy on behalf of immigrants to balance what was thought to be an overweening dominance by those with a background in government enforcement of immigration laws (see Schoenholtz 2005). Further, when John Ashcroft sought to streamline the BIA, he removed those judges that had previous experience working in NGOs and on behalf of immigrants. To this end, we examine the IJ’s career path to create a tightly focused proxy for a policy predisposition toward immigration rights and asylum. We believe our measure is a strong proxy for immigration liberalism in the IJs’ career paths that likely reflects a socialization process that we discuss further below. Additionally, we believe it reflects that the early career selections of some IJs may indicate an underlying policy proclivity that is subsequently strengthened through additional career socialization. For example, a conservative individual may be more likely to seek out a job as an Immigration and Naturalization Service (INS) agent or a prosecutor. As a result, these career experiences are likely to reinforce those underlying proclivities. Our conceptualization of IJs’ career socialization, or career path as a proxy for immigration liberalism, is informed by (1) the literature on public attitudes toward immigrants, which suggests that social interactions inform these attitudes; (2) the empirical work of Ramji-Nogales, Schoenholtz, and Schrag (2009); and (3) the assumptions of U.S. attorneys general, as discussed above, such as Reno and Ashcroft. The immigration and public opinion literatures suggest that exposure to the academic environment and liberal arts education, which “emphasizes the legitimacy as well as the desirability of cultural and personal differences,” may increase individuals’ tolerance and receptivity to immigrants (Hoskin 1991, 115). Further, meaningful social and occupational interactions with immigrants can “demonstrate the ease of living with such differences” (Hoskin 1991, 115; see also Fetzer 2000; Espenshade and Calhoun 1993). Similarly, we expect that judges with prior work experience in academia, at an immigration-related NGO, and in corporate law will be more receptive and less hostile to immigrant claims for asylum. We also include two other types of experience—private practice experience and generic NGO experience—that may reflect a similar socialization to be disposed toward rights claims. Ramji-Nogales, Schoenholtz, and Schrag (2009) found that judges with prior private experience were more likely to grant than deny, an expectation shared by Ocepek and Fetzer (2010). We also examine prior military experience with the expectation that it will socialize the judge to favor the governments’ interests over the rights claims. However, it is possible that some types of military experience would foster a broader, international perspective that might be less hostile to immigrant claims. 14 Consistent with the judicial behavior literature, we expect that IJs 272 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
with prior federal experience in enforcement or prosecutory roles will be more likely to support the government’s claim over the rights’ claim. 15 Thus, we expect that judges’ with prior career experience in the INS, at the Department of Homeland Security, or experience at the EIOR will be socialized to be more anti-immigrant and tend to deny asylum more often. We follow Songer, Ginn, and Sarver’s (2003) methodology and employ factor analysis to create a factor score for the predisposition of the judge toward immigration rights, which we label “liberal policy preferences.” We use the labels “liberal” and “conservative” for the ease of communication but emphasize here that we do not claim the measure captures the general liberalism or conservatism of the judges’ policy proclivities; rather it narrowly focuses on immigration liberalism. In summary, we use a factor analysis of eleven socializing career experiences to represent the career socialization of IJs to create a proxy for their policy views toward immigration and asylum. 16 We code prior experience in the following capacities: the INS, the Department of Homeland Security (non-INS), the EOIR, an NGO, an immigration-related NGO experience, the military, a law school, private practice, prior judicial experience, corporate law, or time as a prosecutor. We retain the first factor from this analysis, which measures the immigration liberalism of the judge. Although the factor analysis creates a number of factors, it is clear that the first factor measures judicial liberalism, given the factor loadings, and that the first factor is, by far, the most significant factor in the data. Higher scores indicate greater liberalism. The online appendix contains an elaboration of our creation of this factor score for liberal policy preferences. HUMAN RIGHTS CONDITIONS While we do not have access to the individual evidence presented by the applicants, we are able to assess the general human rights in their country of origin, which we believe is an appropriate focus for our analysis. As a leading asylum law expert (Anker 2011) notes, current asylum law “emphasizes the salience of human rights conditions in the applicant’s country in determining whether the applicant has a well-founded fear” and thus the applicant’s circumstances must be evaluated in the context of general conditions in his or her country of origin,” and, moreover, she argues that “evidence related to human rights conditions in the country of origin is relevant, indeed central, to all aspects of an applicant’s claim for protection” (57–58). Anker (2011) reports that the federal courts, the BIA, and Asylum Office recognized the key role of U.S. Department of State country reports of general human rights conditions. For example, the Ninth Circuit writes, “Our case law well establishes that the country report from our Department of State is the ‘most appropriate’ and ‘perhaps best resource,’ for determining country conditions” (Lal v. INS 1999, 1068), and other circuits have criticized IJs and the BIA for not giving enough weight to the country conditions evidence (Anker Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 273 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
2011, 115, citations within). To assess the level of human rights abuse in the sending country, we employ Gibney’s five-point measure of state-level personal integrity rights abuse, based specifically on the State Department human rights reports. Gibney’s measure is one of the standard measures of political repression that has been utilized in the empirical human rights literature for over three decades (for an exhaustive list, see http://www. politicalterrorscale.org/bibliography.php; see also Wood and Gibney 2010). 17 The abuses captured in this scale are arguably the most egregious forms of human rights abuse (Keith 2011; Poe and Tate 1994). The categories within the scale are as follows 18: (0) Countries under a secure rule of law, people not imprisoned for their views, torture is rare or exceptional. . . . Political murders extremely rare. (1) A limited amount of imprisonment for nonviolent political activity. Few persons affected, torture and beating exceptional. . . . Political murders rare. (2) Extensive or recent history of extensive political imprisonment. Execution or other political murders and brutality common. Unlimited detention for political views accepted. (3) Practices of level 2 expanded to larger numbers. Murders, disappearances are common, but terror affects primarily those who interest themselves in politics or ideas. (4) Terrors of level 3 expanded to the whole population. No limits on means or thoroughness with which leaders pursue personal or ideological goals (Gastil 1980). In addition, we follow the asylum literature and include a measure of institutionalized democracy (Polity IV’s eleven-point additive scale, Marshall and Jaggers 2013) on the assumption IJs will tend to associate the threat of persecution with authoritarian regimes and may be more skeptical of asylum claims from highly democratic countries. U.S. MATERIAL AND SECURITY INTERESTS Rosenblum and Salehyan (2004) delineate U.S. instrumental interests as those that seek “to preserve good relations with allies (i.e., by rejecting their refugees), to weaken opponents (i.e. by accepting their refugees), and to limit ‘back-door’ access to the United States through illegitimate asylum claims” (681). Following Rosenblum and Salehyan, we include four indicators: a variable measuring logged total trade between the United States and the home country, a dummy variable for countries that are recipients of U.S. military aid, and a dummy variable that delineates the top ten undocumented immigrant sending countries . We also include an additional indicator to capture the level of economic development in the country of origin— the World Bank Development level —which is a four-point scale with zero 274 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
representing the lowest level of development and three the highest. While Rosenblum and Salehyan used only the undocumented immigrant sending country as an indicator of a potential bogus asylum seeker, previous work has demonstrated the level of economic development of the sending country may also serve as such a cue (Holmes and Keith 2010; Keith and Holmes 2009). INTERACTIONS One of the most important dimensions we seek to model is the interaction between these six norms and interests indicators and judges’ policy leanings. Thus, we interact each of the six variables with the immigration liberalism scores. We use these interactions to determine whether the assessment of a given factor is contingent on immigration liberalism. Therefore, we use these interactions to assess our second hypothesis concerning the cognitive approach to IJ decision making. CONTROLS We include eight control variables, three of which are measures specifically linked to the applicant. We create dummy variables for whether the applicants are English-speakers or Arabic-speakers. We also include a dummy variable for whether the applicant had legal representation. Additionally, we consider gender of the judge. The judicial behavior literature suggests that female judges vote more liberally than male judges, but the results vary by courts and case area (see Ocepek and Fetzer 2010; Boyd, Epstein, and Martin 2008; Peresie 2005; Songer and Crews-Meyer 2000; Gryski, Main, and Dixon 1986). Songer and Crews-Meyer suggest that “female judges may be quicker to empathize with underdogs in a variety of civil liberties issues since these judges have either experienced or witnessed the problems involved in being a political minority” (759). Ramji-Nogales, Schoenholtz, and Schrag (2009) note that such experiences “might make female judges more sympathetic to stories of persecution, as well as more conscious in eliminating their own biases from the decision making process” (47–48). They also suggest that the gender effect may reflect greater empathy toward the applicant and a less combative setting that may result in a more coherent testimony. We include three measures of U.S. domestic conditions, which may influence the judges’ general probability to grant or deny asylum. The first two are measures of the national unemployment rate and the national change in per capita gross domestic product (GDP) . We also control for whether the immigration court is located in a border state as judges on these courts may face a different mix of asylum seekers. This variable controls for the possibility that the applicant pool in states with border points of entry may vary from that of other states. These courts may hear more defensive cases in which judges Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 275 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
assume post hoc claims are bogus, since the applicants did not make a claim of asylum until facing deportation. On the other hand, border control and expedited removal procedures at ports of entry may actually serve as a filter, eliminating many of the likely bogus claims early in the process. Thus, we are agnostic about the direction of the control variable. Finally, we include a dummy variable for applications initiated in the immigration courts prior to IIRIRA ( pre-IIRIRA) to control for the applicants who will not have gone through the winnowing process of expedited removals at points of entry that IIRIRA put in place. RESULTS We estimate a multilevel model with a random intercept that varies across each of the IJs. The two-level approach accommodates the natural structuring of the data in which applicants nest within judges—that is, applicants appear before one and (usually) only one judge, whereas judges will hear the cases of multiple applicants. Further, allowing the intercept for judges to vary randomly allows us to take into account any clustering due to the tendency for the same IJ to decide cases similarly based on unobserved characteristics of that judge. Finally, this allows us to analyze the extent to which judges with different policy proclivities conditionally treat the factors associated with granting asylum. In short, the two-level random intercept model allows us to ask the substantive question we are interested in while also accounting for the structure of the data. We include six cross-level interaction terms in the model by interacting our immigration liberalism variable with each of the variables related to the norms and interests debate in the IR literature on asylum decision making. First, we estimated a simple model to build upon the standard asylum models in the field. The benefit of this model is that it allows us to demonstrate the importance of immigration liberalism unconditionally. The results are presented in the first two columns of Table 1. We then estimated the model with interactions between immigration liberalism and the material interests, security interests, and humanitarian factors. This second model allows us to evaluate the extent to which our cognitive approach to decision making is applicable. By evaluating the interactions terms, and their associated predicted probabilities, we can determine the extent to which “norms” and “interests” are treated objectively by IJs. The results of this model are reported in the last two columns of Table 1, under the Model 2 heading. We will only briefly discuss Model 1 and discuss more thoroughly the model with interactions. Our models fit the data well. Our decision to fit a random intercept, which varies across each of the 244 IJs, is supported by the data. The Wald c 2 statistic for both models as a whole is highly statistically significant. More interestingly, the likelihood ratio test comparing our models with a random intercept to a simple logit model suggests that our 276 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
Variables Model 1 Model 2 b S.E. b S.E. Judicial Policy Preference Immigration Liberalism ( +) 0.267* 0.061 0.039 0.064 U.S. Material & Security Interests Log of Trade with US ( -) -0.099* 0.019 -0.103* 0.003 US Military Aid ( -) -0.163* 0.012 -0.334* 0.018 Top-Ten Illegal Immigration ( -) -0.079* 0.012 0.099* 0.020 World Bank Development Class ( +) 0.316* 0.008 0.178* 0.013 Human Rights Conditions Democracy (Polity) ( -) -0.054* 0.002 -0.056* 0.003 Human Rights Abuse (PTS-St. Dept.) ( +) 0.210* 0.006 0.197* 0.009 Interaction Terms Log of TradeImmigration Liberalism — — 0.005 0.002 Military AidImmigration Liberalism — — 0.160 0.013 Top-Ten IllegalImmigration Liberalism — — -0.176 0.014 World Bank DevelopmentImmigration Liberalism — — 0.122 0.009 DemocracyImmigration Liberalism — — 0.002 0.002 Human Rights AbuseImmigration Liberalism — — 0.011 0.007 Controls Judge Gender ( +) 0.358* 0.103 0.356* 0.103 English Speaker ( +) 0.488* 0.012 0.485* 0.012 Arabic Speaker ( +/-) 0.265* 0.026 0.255* 0.026 Legal Representation ( +) 2.277* 0.016 2.278* 0.016 Border State ( +/-) 0.106 0.106 0.107 0.106 National Unemployment ( -) -9.813* 0.706 -9.848* 0.707 National GDP Change ( -) -13.876* 0.431 -13.889* 0.431 Pre-IIRIRA Application ( -) -0.280* 0.012 -0.278* 0.012 Constant -1.887* 0.115 -1.635* 0.117 N 329101 329101 N of Immigration Judges 244 244 Wald c2 30664.22 ( p = 0.000) 31003.60 ( p = 0.000) Variance of Random Intercept 0.498* 0.048 0.499* 0.048 LR Test of Random Intercept ( c-bar) 24314.31 ( p = 0.000) 24264.88 ( p = 0.000) * Coefficients are significant at p < 0.05 (two-tailed). Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 277 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
models with random intercepts fit the data significantly better than a model without a random intercept. Another benefit of a random intercept model is that we can account for how much of the variance in the granting of asylum is attributable to each level in the data (Raudenbush and Bryk 2002). If the process of asylum decision making were entirely based on the characteristics of applicants, we would expect that none of the variance in the granting of asylum would be attributable to the IJ. Our estimate of the intraclass correlation coefficient from an empty model, which contains no independent variables, suggests that 14.4 percent of the variance in the decision to grant asylum occurs at the judge level. All of the variables in the model are statistically significant and in the predicted direction with the exception of one control variable: border state. The substantive impact of the judge an asylum seeker faces is significant. Using the coefficient from Model 1, a change in immigration liberalism from the minimum (zero) to the maximum (3.8) (i.e., increasing the liberalism of the judge), increases the predicted probability of a vote to grant asylum by twenty-five percentage points in the unconditional model. Coupled with the fact that over 14 percent of the variance in outcomes occurs at the judge level, the substantive effect of the immigration liberalism of the judge is strong evidence that the decision making of IJs is partially driven by their policy preferences and that variation of outcomes across judicial policy predispositions is partially responsible for the divergence of grant rates across judges. Thus, our first hypothesis that IJ decisions are influenced by their policy predispositions toward immigration issues is strongly supported in our analysis. Furthermore, our results suggest that both norms and interests matter, across a board range of indicators, even while controlling for a significant number of factors related to the judge, the applicant, and domestic economic conditions in Model 1. In Figure 1, we graph the effects for the interacted terms for continuous measures in Model 2 to visualize the influence of IJ liberalism in different contexts. The first major finding is that there is still a considerable difference between IJ’s liberalism scores in the tenth percentile (conservative) and those in the ninetieth percentile (liberal). In the first panel of Figure 1, we plot the differences between conservative judges and liberal judges with respect to trade. We have selected two extralegal/national interest factors that are recognized to be influential on asylum outcomes. First, the effect of trade is as predicted. If the United States trades with the country the individual is fleeing, there is a reduced likelihood that the applicant will be granted asylum. The shifted intercepts here show the effect of the IJ’s policy preferences—no matter the level of trade between the sending nation and the United States, the liberal IJ is considerably more likely than the conservative IJ to admit an applicant. Note that this true for all of the graphs in
are conservative IJs, consistent with Hypothesis 1. Moving from the lowest level of trade to the highest, a conservative IJ’s likelihood of granting asylum 278 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
0 . 1. 2. 3. 4. 5. 6. 7. 8 Predicted Prob. of Vote to Grant Asylum0123 0 . 1. 2. 3. 4. 5. 6. 7. 8 Predicted Prob. of Vote to Grant Asylum02468 1 0 1 2 0. 1. 2 . 3 . 4. 5 . 6. 7 . 8 Predicted Prob. of Vote to Grant Asylum02468 1 0 0. 1. 2. 3. 4. 5. 6. 7. 8 Predicted Prob. of Vote to Grant Asylum01234 Human Rights Abuse (PTSSD ) Democracy Score (Polity) Conservative IJ Liberal IJ 95% CI 95% CI Conservative IJ Liberal IJ 95% CI 95% CI Conservative IJ World Bank Development Classification Log of Trade with Sending Country Liberal IJ 95% CI 95% CI Conservative IJ Liberal IJ 95% CI 95% CI
Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 279 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
decreases from 0.45 to 0.19, whereas, for a liberal IJ, the decrease is from 0.67 to 0.40. Second, the plot in the second panel displays the effect of the level of development of the sending nation on the likelihood of admission. Again, our expectations are met here as applicants from more developed countries are significantly more likely to be admitted than those from less developed countries, who may be viewed as economic opportunists instead of real asylum seekers. Going from least developed to most developed, a conservative IJ’s probability of granting asylum increases from 0.24 to 0.39, and a liberal IJ’s probability of granting asylum increases from 0.36 to 0.68. Next, we select two factors that judges are trained to consider as legitimate factors in assessing asylum applications. The third panel of Figure 1 displays the effect of the level of human rights abuse in the sending country on the likelihood of a grant of asylum. The effects here are as we would predict, with higher levels of abuse leading to a greater likelihood of a grant. For a conservative judge moving from the lowest level of human rights abuse to the highest level of abuse increases the probability of a grant of asylum from 0.19 to 0.35; similarly, for a liberal judge, the probability of a grant changes from 0.38 for the lowest level of human rights abuse to 0.62 for the highest level. As presented in the graph in the fourth panel of the figure, the more democratic the sending country, the less likely is a grant of asylum. Moving from the lowest level of democracy to the highest, a conservative judge’s probability of a grant decreases from 0.32 to 0.21, and a liberal IJ’s probability of a grant decreases from 0.56 to 0.43. The changes across the dichotomous interactions are presented in
sending military aid to the sending country. Interestingly, judges respond differently to military aid; a conservative IJ (at the tenth percentile in the data) becomes less likely to grant asylum in the presence of military aid with the sending country (moving from a 0.32 probability of a grant to 0.26) while a liberal IJ (at the ninetieth percentile) actually becomes more likely to grant asylum to an applicant from a sending country to which the United States sends military aid (changing from a 0.46 probability when there is no military aid to a 0.50 probability when military aid is present). The last two rows of Table 2 present the effects on an applicant of coming from a top-ten source country of illegal immigrants to the United States. It does not appear
Dichotomous Interactions Conservative Policy Preferences Liberal Policy Preferences Military Aid 0.26 [0.23, 0.30] 0.50 [0.41, 0.58] No Military Aid 0.32 [0.28, 0.36] 0.46 [0.37, 0.54] Top-Ten Illegal Sending Country 0.27 [0.23, 0.30] 0.39 [0.31, 0.47] Not Top-Ten Illegal Sending Country 0.26 [0.23, 0.31] 0.50 [0.41, 0.58] 280 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
that conservative IJs react to whether the applicant comes from a country known for producing illegal immigrants (essentially granting asylum 0.26 of the time in both scenarios), whereas liberal IJs react rather strongly to the fact that the applicant comes from a country that produces many illegal immigrants (decreasing their likelihood of granting asylum from 0.50 to 0.39). Judges with liberal policy proclivities respond to the absence of indicators that the applicant is a bogus asylum more strongly than judges with conservative policy proclivities. It appears that when freed of some doubt that the applicant is an economic migrant, the liberal judge is even more likely to grant asylum, whereas conservatives who have a underlying antiimmigrant proclivity in these cases appear to be unmoved by the absence of abstract indicators that the applicant is not economic migrant. Conservative judges then tend to respond cautiously toward applicants who are fleeing states that receive U.S. military aid, perhaps unconsciously not wanting to offend a military ally by painting it as an abusive regime, and perhaps perceiving individuals fleeing a military ally as a potential threat. It is more difficult to explain why more liberal judges would be slightly more inclined to accept an applicant from a state receiving U.S. military aid. It may reflect a liberal perception of U.S. history of supporting rights-abusive regimes. Our model unveils a complex reality in which judicial policy proclivities do not have a constant influence in every context. The interacted model allows for deeper insight, particularly in that we can assess how the predicted probability of a grant of asylum changes across values of the specific norm or interest variables by the judge’s immigration liberalism. In essence, Model 2 allows us to ask not just whether the policy preferences of IJs enter into the decision-making process but also how those preferences enter the process. Recall, our theory of judging is premised on the notion that the interpretation of the human rights conditions will be constrained somewhat by U.S. asylum law, by the training that IJs receive, and by the evidence that they are most likely to access. This is borne out by examining the extent to which the interpretation of a factor differs between an IJ at the tenth percentile of immigration liberalism and one at the ninetieth percentile. In order to test Hypothesis 2, we present in Table 3 a set of four scenarios to demonstrate the extent to which human rights conditions constrain the use of asylum predispositions in decision making compared to the evaluation of material and security concerns. The first set of scenarios holds the human rights conditions constant at the average values for each variable while varying the material and security interest variables from an alignment unfavorable to a grant (high trade, presence of military aid, applicant from top-ten illegal country and low development) to an alignment favorable to a grant (low trade, no military aid, applicant not from top-ten illegal country and high development). The first two columns show the predicted probabilities and 95 percent confidence intervals for each scenario for a Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 281 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
conservative judge (tenth percentile of immigration liberalism) and a liberal judge (ninetieth percentile of immigration). The last two columns show the absolute difference across the scenarios for a conservative judge and for a liberal judge. It is these values in which we are most interested. For the conservative judge, the change across the high and low material and security interest scenarios is forty-eight percentage points; it is sixty-six percentage points for the liberal judge. Compare these numbers to those below them in the table, where we have held the material and security concerns constant at the average values for each variable and varied the human rights variables. In the third row, the high human rights concerns scenario, the democracy score is set to zero, and the human rights abuse score is set to four; in the fourth row, the low human rights concern scenario, these numbers are set to ten and zero, respectively. Across these two scenarios, both the conservative judge and the liberal judge demonstrate greater constraint, with absolute differences of twenty-seven percentage points and thirty-four percentage points. The increased level of constraint in the human rights scenarios is evidence for the cognitive approach we have proposed. As the material, security, and human rights conditions associated with an applicant’s sending country vary, the assessment of the applicant’s likely material and security characteristics are more dependent on the policy proclivities of the IJ evaluating the applicant than are the human rights conditions. This is indicated by the absolute differences across the four scenarios in Table 3. These differences in the material and security scenarios (first two rows of
Table 3), are greater than are the differences across the human rights scenarios in the bottom two rows of the table. This is important because it
highlights that the consideration of human rights conditions, as specified by U.S. asylum law, is a relative constraint on the use of policy proclivities by IJs—at least when compared to consideration of extralegal material or security concerns. Put differently, IJs will evaluate two applicants that have the same material interests attached to their applications (say coming from a countries with the same World Bank development score) with more variation than they will two applicants with the same humanitarian factors attached to the application (for example, the Polity Democracy score of the sending countries).
Conservative Policy Preferences Liberal Policy Preferences Conservative Absolute Difference Liberal Absolute Difference High Material & Security Interest 0.15 [0.13, 0.18] 0.20 [0.14, 0.25] 0.48 0.66 Low Material & Security Interest 0.63 [0.58, 0.68] 0.86 [0.83, 0.91] High Human Rights Concerns 0.42 [0.37, 0.46] 0.66 [0.59, 0.74] 0.27 0.34 Low Human Rights Concerns 0.15 [0.13, 0.18] 0.32 [0.24, 0.39] 282 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
CONCLUSION This article identifies reasons for the wide disparity in asylum grant rates by IJs. The previously recognized competing influences of norms versus interests operate differently across judges according to the immigration liberalism of each judge. We have specified an underlying causal mechanism for the operation of these preferences in asylum cases—a cognitive process that accounts for the individual weight that a judge may give national security, economic issues, or human rights cues—while controlling for a variety of factors that may influence the judges’ decisions. We demonstrate that the large effects for immigration liberalism primarily enter the asylum decision-making process through the extralegal consideration of the material and security interests of the United States, and that humanitarian concerns impose some constraint on the use of policy preferences. Thus, we provide insight into both the empirical and normative debate within international relations concerning norms versus interests. The extent to which each matters in asylum outcomes depend upon the judge the applicant draws, the judge’s immigration policy proclivities, and the minimal constraints imposed by U.S. asylum law. Further, the apparent randomness of the variation between IJ grant rates can be explained by applying basic cognitive and political science theories of judicial decision making. We explain, for the first time to our knowledge, the variation in grant rates among IJs in a theoretically driven manner that is empirically sophisticated. Our findings raise significant policy concerns. Is the United States’ asylum policy implemented in a manner that is fair and consistent? Are outcomes driven by the judge one draws? The evidence here suggests that it is does matter which judge hears an applicant’s claims, particularly when material and security concerns may be an issue for an applicant, and the potentially combined effect can substantially alter the applicant’s odds of a grant or denial. Our analysis is premised in part on the assumption that the reliance of judges on their decisional predispositions is a rational response to the institutional constraints under which judges operate, including the vagueness of the law, the lack of concrete evidence, and the difficulty of assessing credibility. Our analysis suggests that one of the side effects of these constraints may be a reliance on cues that may reflect the judges’ preferences or values, which in turn introduces a large element of chance into the process of deciding asylum applications. NOTES 1. We do not take a position on the exact underlying psychological mechanism that underlies the differential weightings that help to explain IJ decision making, because with observational data we are unable to distinguish these mechanisms. Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 283 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
affect the results that we present below. We have chosen to omit it from the final model because we do not believe that presidential ideology has any influence on who is and is not hired as an IJ. Unlike even federal district court judges, presidents and their surrogates are unlikely to pay any attention to who is hired as an IJ during their term. See the online appendix for further explanation. 17. The CIRI measure of physical integrity is the other dominant measure in the empirical human rights literature. We should to use the Gibney measure over the CIRI measure as it is based solely on the State Department country reports and therefore represents assessment of the country information that IJs and the asylum system more generally regard as authoritative. 18. We have rescaled the measure from its original range (1 to 5 range) to 0 to 4, with lower scores still representing less abuse and higher scores more abuse. linda camp keith is an associate professor in political science in the School of Economic, Political and Policy Sciences at the University of Texas at Dallas. Her major area of research is international human rights and the role of law and courts. She also researches U.S. courts including the U.S. Supreme Court and immigration courts. jennifer s. holmes is an associate professor of political economy and political science at the University of Texas at Dallas. Her major area of research is political violence, terrorism, U.S. asylum policy, and political development. banks p . miller is an assistant professor of political science at the University of Texas at Dallas. His major areas of research are judicial decision making, judicial specialization, and U.S. asylum policy.
Alexander, Sydenham B. 2006. “A Political Response to Crisis in the Immigration Courts,” Georgetown Immigration Law Journal 21: 2–60. Anker, Deborah E. 2011. Law of Asylum in the United States , 2011 Edition. Eagan, MN: West. Bartels, Brandon L. 2010. “Top-Down and Bottom-Up Models of Judicial Reasoning.” In The Psychology of Judicial Decision Making , edited by David Klein and Gregory Mitchell. New York: Oxford Univ. Press. Baum, Lawrence. 1994. “What Judges Want: Judges’ Goals and Judicial Behavior,” Political Research Quarterly 47: 749–68. —. 2010. “Judicial Specialization and the Adjudication of Immigration Cases,” Duke Law Journal 59: 1634–721. Becker, Andrew, and Hugo Cabrera. 2009. “Immigration Courts Make Do with Limited Resources Despite Mounting Caseloads,” Center for Investigative Reporting: Investigation , June 30. http://cironline.org/reports/immigration-courts-makedo-limited-resources-despite-mounting-caseloads-2285 (accessed May 14, 2013). Bernstein, Nina. 2006. “Immigration Judges Facing Yearly Performance Reviews,” New York Times 10 August. http://www.nytimes.com/2006/08/10/washington/ 10immig.html?_r=0 (accessed May 16, 2013). Bohmer, Carol, and Amy Shuman. 2008. Rejecting Refugees: Political Asylum in the 21st Century. New York: Routledge Press. Boyd, Christina, Lee Epstein, and Andrew Martin. 2008. “Untangling the Causal Effects of Sex on Judging,” American Journal of Political Science 54: 389–411. Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 285 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
Braman, Eileen, and Thomas Nelson. 2007. “Mechanism of Motivated Reasoning? Analogical Perception in Discrimination Disputes,” American Journal of Political Science 51: 940–56. Brehm, John, and Scott Gates. 1997. Working, Shirking, and Sabotage: Bureaucratic Response to a Democratic Public . Ann Arbor: Univ. of Michigan Press. Brudney, James J., Sara Schiavoni, and Deborah J. Merritt. 1999. “Judicial Hostility toward Labor Unions? Applying the Social Background Model to a Celebrated Concern,” Ohio State Law Review 60: 1675–766. Chayes, Abram, and Antonia Handler Chayes. 1993. “On Compliance,” International Organization 47: 175–205. Cianciarulo, Marisa Silenzi. 2006. “Terrorism and Asylum Seekers: Why the Real ID Act Is a False Promise,” Harvard Journal of Law 43:101–43. Epstein, Lee, and Jack Knight. 1998. The Choices Justices Make . Washington, DC: C.Q. Press. Espenshade, Thomas J., and Charles A. Calhoun. 1993. “An Analysis of Public Opinion toward Undocumented Immigration,” Population Research and Policy Review 12: 189–224. Fazio, Russell H., and Tamara Towles-Schwen. 1999. “The MODE Model of Attitude-behavior Processes.” In Dual Process Theories in Social Psychology, edited by Shelly Chaiken and Yaacov Trope, 97–116. New York: Guilford Press. Fetzer, Joel S. 2000. Public Attitudes Immigration in the United States, France, and Germany. Cambridge: Cambridge Univ. Press. Fischman, Joshua B., and David S. Law. 2009. “What Is Judicial Ideology, and How Should We Measure It?” Washington University Journal of Law and Policy 29: 133–214. Gastil, Raymond. 1980. Freedom in the World: Political Rights and Civil Liberties, 1980. New Brunswick, NJ: Transaction Books. George, Tracey E., and Lee Epstein. 1992. “On the Nature of Supreme Court Decision Making,” American Political Science Review 86: 323–37. Gibney, Mark, and Michael Stohl. 1988. “Human Rights and U.S. Refugee Policy.” In Open Border? Closed Societies? edited by Mark Gibney, 151–73. Westport, CT: Greenwood. Gibney, Mark, Vanessa Dalton, and Marc Vockell. 1992. “USA Refugee Policy: A Human Rights Analysis Update,” Journal of Refugee Studies 5: 33–46. Gillman, Howard. 2001. “What’s Law Got to Do with It? Judicial Behavioralists Test the ‘Legal Model’ of Judicial Decision Making,” Law & Social Inquiry 26: 465–504. Gould, Jon B., Colleen Sheppard, and Johannes Wheeldon. 2010. “A Refugee from Justice? Disparate Treatment in the Federal Court of Canada,” Law and Policy 32: 454–86. Government Accountability Office (GAO). 2008. “Significant Variation Existed in Asylum Outcomes across Immigration Courts and Judges,” GAO-08–940. Washington, DC: GAO. Gryski, Gerard S., and Eleanor C. Main, and William J. Dixon. 1986. “Social Backgrounds as Predictors of Votes on State Courts of Last Resort: The Case of Sex Discrimination,” Western Political Quarterly 39: 528–37. Hassan, Lisa. 2000. “Deterrence Measures and the Preservation of Asylum in the United Kingdom and United States,” Journal of Refugee Studies 13: 184–204. Henkin, Louis. 1979. How Nations Behave: Law and Foreign Policy . New York: Council on Foreign Relations. Holmes, Jennifer S., and Linda Camp Keith. 2010. “Does the Fear of Terrorists Trump the Fear of Persecution in Asylum Outcomes in the Post-September 11 Era?” PS: Political Science 43: 431–36. 286 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
Hoskin, Marilyn. 1991. New Immigrants and Democratic Society . Westport, CT: Praeger. Immigration Judge Benchbook . 2009. U.S. Department of Justice, Executive Office of Immigration Review. http://www.justice.gov/eoir/vll/benchbook/index. html (accessed December 1, 2012). Joseph Jucewicz, and Lawrence Baum. 1990. “Workload Influences on Supreme Court Case Acceptance Rates, 1975–1984,” Western Political Quarterly 43: 123– 35. Keck, M., and K. Sikkink. 1998. Activists beyond Border: Transnational Advocacy Networks in International Politics . New York: Cornell Univ. Press. Keith, Linda Camp. 2011. Political Repression: Courts and Law . Philadelphia: Univ. of Pennsylvania Press. Keith, Linda Camp, and Jennifer S. Holmes. 2009. “Determinants of Asylum Grants: A Rare Examination of Factors Typically Unobservable in U.S. Asylum Decisions,” Journal of Refugee Studies 22: 224–41. Keohane, Robert O. 1984. After Hegemony: Cooperation and Discord in the World Political Economy. Princeton, NJ: Princeton Univ. Press. Kerwin, Donald. 2005. “The Use and Misuse of ‘National Security’ Rationale in Crafting U.S. Refugee and Immigration Policies,” International Journal of Refugee Law 17: 749–63. Law, David S. 2005. “Strategic Judicial Lawmaking: Ideology, Publication, and Asylum in the Ninth Circuit,” University of Cincinnati Law Review 73: 817–66. Legomsky, Stephen H. 2007. “Learning to Live with Unequal Justice: Asylum and the Limits of Consistency,” Stanford Law Review 60: 413–75. —. 2010. “Restructuring Immigration Adjudication,”Duke Law Journal 59: 1634–721. Lerner, Jennifer S., and Philip E. Tetlock. 1999. “Accounting for the Effects of Accountability,” Psychological Bulletin 125: 255–75. Levy, Carl. 2005. “The European Union after 9/11: The Demise of a Liberal Democratic Asylum Regime?” Government and Opposition 40: 26–59. Loescher, Gil, and John A. Scanlan. 1998. Calculated Kindness Refugees and America’s Half-Open Door 1945-Present . New York: Macmillan. Lustig, Stuart L., Niranjan Karnik, Kevin Delucchi, Lakisha Tennakoon, Brent Kaul, Dana Leigh Marks, and Denis Slavin. 2008. “Inside the Judges’ Chambers: Narrative Responses from the National Association of Immigration Judges Stress and Burnout Survey,” Georgetown Immigration Law Journal 23: 57–83. Marshall, Monty G., and Keith Jaggers. 2013. Polity IV Project: Political Regime Characteristics and Transitions, 1800–2010 . http://www.systemicpeace.org/polity/ polity4.htm (accessed December 1, 2012). Martin, David A. 2000. “The 1995 Asylum Reforms: A Historic and Global Perspective,” Washington, DC: Center for Immigration Studies. http://www.cis.org/node/ 172 (accessed May 14, 2013). Mearsheimer, J. J. 1994. “The False Promise of International Institutions,” International Security. 19: 5–26. Miller, Banks, and Brett Curry. 2009. “Expertise, Experience and Ideology on a Specialized Court: The Case of the Court of Appeals for the Federal Circuit,” Law and Society Review 43: 839–64. Neumayer, Eric. 2005. “Asylum Recognition Rates in Western Europe-Their Determinants, Variation and Lack of Convergence.” Journal of Conflict Resolution 49: 43–66. Ocepek, Melissa, and Joel Fetzer. 2010. “The Causes of Liberal Voting in the United States Supreme Court,” International Migration Review 44: 659–96. Peresie, Jennifer L. 2005. “Female Judges Matter: Gender and Collegial Decisionmaking in the Federal Appellate Courts,” Yale Law Journal 114: 1759–90. Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 287 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
Poe, Steven C., and C. Neal Tate. 1994. “Repression of Human Rights to Personal Integrity in the 1980s: A Global Analysis,” American Political Science Review 88: 853–72. Ramji-Nogales, Jaya, Andrew I. Schoenholtz, and Philip G. Schrag. 2007. “Refugee Roulette: Disparities in Asylum Adjudication,” Stanford Law Review 60: 295–411. —. 2009. Refugee Roulette: Disparities in Asylum Adjudication and Proposals for Reform. New York: NYU Press. Raudenbush, S. W., and Bryk, A. S. 2002. Hierarchical Linear Models , 2nd ed. Thousand Oaks, CA: Sage. Wood, Reed M., and Mark A. Gibney. 2010. “The Political Terror Scale (PTS): A Re-introduction and a Comparison to CIRI,” Human Rights Quarterly 33: 367–410. Risse, Thomas, Stephen C. Ropp, and Kathryn Sikkink (eds.). 1999. The Power of Human Rights: International Norms and Domestic Change . New York: Cambridge Univ. Press. Rosenblum, Marc R., and Idean Salehyan. 2004. “Norms and Interests in US Asylum Enforcement,” Journal of Peace Research 41: 677–97. Rottman, Andy J., Christopher J. Fariss, and Steven C. Poe. 2009. “The Path to Asylum in the US and the Determinants for Who Gets In and Why,” International Migration Review 43: 3–34. Salehyan, Idean, and Marc R. Rosenblum. 2008. “International Relations, Domestic Politics, and Asylum Admissions in the United States,” Journal of Politics 61: 104–21. Schoenholtz, Andrew I. 2005. “Refugee Protection in the United States Post- September 11,” Columbia Human Rights Law Review 36: 323–64. Schuster, Lisa. 2000. “A Comparative Analysis of the Asylum Policy of Seven European Governments,” Journal of Refugee Studies 13: 118–32. Segal, Jeffrey A., and Harold J. Spaeth. 2002. The Supreme Court and the Attitudinal Model Revisited. Cambridge: Cambridge Univ. Press. Segal, Jeffrey A., Robert Howard, and Christopher Hutz. 1996. “Presidential Success in Supreme Court Nominations: Testing a Constrained Presidency Model.” Paper presented at the Annual Meeting of the Midwest Political Science Association, April 18–20, Chicago. Songer, Donald, and Kelly Crews-Meyer. 2000. “Does Judge Gender Matter? Decision Making in State Supreme Courts,” Social Science Quarterly 81: 750–62. Songer, Donald, Charles Cameron, and Jeffrey Segal. 1995. “An Empirical Test of the Rational Actor Theory of Litigation,” Journal of Politics 57: 1119–29. Songer, Donald, Martha Humphries Ginn, and Tammy Sarver. 2003. “Do Judges Follow the Law When There Is No Fear of Reversal,” Justice Systems Journal 24: 137–61. Tate, C. Neal, and Roger Handberg. 1991. “Time Building and Theory Building in Personal Attribute Models of Supreme Court Voting Behavior, 1916–88,” American Journal of Political Science 35: 460–80. Taylor, Margaret H. 2007. “ Refugee Roulette in an Administrative Law Context: The Déjà vu of Decisional Disparities in Agency Adjudication,” Stanford Law Review 60: 475–501. Transactional Records Access Clearinghouse (TRAC). 2006. Immigration Judges . http://trac.syr.edu/immigration/reports/160/ (accessed December 1, 2012). —. 2009. Immigration Courts Still a Troubled Institution . http://trac.syr.edu/ immigration/reports/210/ (accessed December 1, 2012). —. 2011. Rising Immigration Backlog At All-time High Yet Criminal, National Security, and Terrorism Cases Fall . http://trac.syr.edu/immigration/reports/261/ (accessed December 1, 2012). 288 LAW & POLICY October 2013 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary
U.S. Commission on International Religious Freedom. 2008. Report on Asylum Seekers in Expedited Removal—Report Card Two Years Later . http://www.uscirf. gov/images/ stories/pdf/scorecard_final.pdf (accessed December 1, 2012). U.S. Department of Justice Office of the Inspector General. 2008. Department of Justice Report: An Investigation of Allegations of Politicized Hiring by Monica Goodling and Other Staff in the Office of the Attorney General , July 28, 2008. Washington, DC. U.S. Department of Justice Office of the Inspector General. Waltz, Kenneth. 1979. Theory of International Politics . Reading, MA: Addison Wesley. Westerland, Chad. 2009. “The Consequences of Immigration Reform for the US Courts of Appeals.” Paper presented at the annual meeting of the Midwest Political Science Association, April 2–5, Chicago. Williams, Margaret S., and Anna O. Law. 2012. “Understanding Judicial Decision Making Immigration Cases at the U.S. Courts of Appeals,” The Justice System Journal 33: 98–119. Zupan, Mark A. 1992. “Measuring the Ideological Preferences of U.S. Presidents: A Proposed (Extremely Simple) Method,” Public Choice 73: 351–61. CASES CITED Iao v. Gonzalez, 400 F.3d 530 (7th Cir. 2005). Lal v. INS, 255 F.3d 998 (9th Cir. 1999). LAWS CITED Geneva Convention relating to the Status of Refugees. 1951. 19 U.S.T. 6259, 189 U. N.T.S. 150. Illegal Immigration Reform and Immigration Responsibility Act (IIRIRA). Pub. L. No. 104–208, 110 Stat. 30009–546 (1996). Real ID Act (Real ID). Pub. L. No. 109–13, 119 Stat. 302 (2005). Refugee Act of 1980. INA § 243(h). Pub. L. No. 96–212, Tit. II, 94 Stat. 107 § 203(e) (March 17, 1980). Keith et al. AN ATTITUDINAL AND COGNITIVE APPROACH 289 © 2013 The Authors Law & Policy © 2013 The University of Denver/Colorado Seminary