ABSTRACT Title of Document: CONTEXTUAL VARIATION IN THE PUNISHMENT OF HISPANIC NON- CITIZENS: A MULTILEVEL ANALYSIS FROM AN IMMIGRATION THREAT PERSPECTIVE Byron E. Villagomez M., Master of Arts, 2013 Directed By: Associate Professor Brian D. Johnson, Department of Criminology and Criminal Justice The increasing number of Hispanic non-citizens being punished by the Federal courts has revived the debate about sentencing disparity. This study extends extant research by examining contextual variation in the punishment of Hispanic defendants, with a focus on citizenship status. Using data from the United States Sentencing Commission for FY2008, this paper purports to explain that variation using factors drawn from an immigration threat standpoint. The results indicate that the expected variation exists, with non-citizens receiving shorter or lengthier sentences than citizens, depending on the place of sentencing. Moreover, Hispanic political representation and unauthorized immigrant populations had a slight but significant negative and positive effect, respectively, for all Hispanic defendants. Also, the former aspect had a minor and negative impact on sentence length for Hispanic non- citizens. These findings contribute to a better understanding of how Hispanics are sentenced by the Federal courts, and tracing the route for future research. CONTEXTUAL VARIATION IN THE PUNISHMENT OF HISPANIC NON - CITIZENS: A MULTILEVEL ANALYSIS FROM AN IMMIGRATION THREAT PERSPECTIVE By Byron Ernesto Villagomez Moncayo Thesis submitted to the Faculty of the Graduate School of the University of Maryland, College Park, in partial fulfillment of the requirements for the degree of Master of Arts 2013 Advisory Committee: Associate Professor Brian D. Johnson, Chair Associate Professor Laura Dugan Professor Raymond Paternoster ? Copyright by Byron Ernesto Villagomez Moncayo 2013 ii Dedication This thesis is dedicated to my mother?the person who has always encouraged me to fight for my dreams and who visualized this achievement of mine a long time ago. iii Acknowledgements In the first place, I would like to acknowledge my thesis chair, Dr. Brian Johnson, who guided me through the intense and tough process of drafting and defending my thesis. His patience, advising, and constructive comments were fundamental in the completion of this work. I would also like to thank Dr. Raymond Paternoster for stimulating me to defend the ideas and assumptions of my thesis, and to think critically about them. In addition, I would like to express my endless gratitude to Dr. Laura Dugan. As a wise and demanding teacher, as well as an understanding academic advisor, she led me not only to acquire invaluable academic and research skills, but more important, to defeat my own fears and learn an indelible lesson of life about courage, faith, and humility. Finally, I am especially grateful to my beloved Maria Canela. Her permanent support, love, and companionship were fundamental for overcoming every obstacle, and essential for completing this process successfully. Without her love and support I would not be where I am today. iv Table of Contents Dedication ..................................................................................................................... ii Acknowledgements ...................................................................................................... iii Table of Contents ......................................................................................................... iv List of Tables ................................................................................................................ v List of Figures .............................................................................................................. vi Chapter 1: Introduction ................................................................................................. 1 Chapter 2: Literature Review ...................................................................................... 15 Empirical Research on the Impact of Citizenship Status on Sentencing ................ 15 Empirical Research on Contextual Variation in Punishment by Citizenship Status ................................................................................................................................. 19 Chapter 3: Theoretical Framework ............................................................................. 22 Chapter 4: Data and Methods ..................................................................................... 31 Data ......................................................................................................................... 31 Dependent Variable ................................................................................................ 32 Independent Variables ............................................................................................ 33 Control Variables .................................................................................................... 35 Analytical Approach ............................................................................................... 37 Chapter 5: Findings .................................................................................................... 40 Descriptive Statistics ............................................................................................... 40 Results from Ordinary Least Squares Regressions ................................................. 43 Results from Two-Level Unconditional HLM Model ............................................ 44 Results from the Two-Level Random Intercept Model: Individual Level Fixed Effects ..................................................................................................................... 45 Results from the Two-Level Random Intercept Model: Contextual Level Fixed Effects ..................................................................................................................... 46 Results from the Two-Level Full Random Coefficient Model ............................... 47 Results from Cross-Level Interaction Models ........................................................ 48 Chapter 6: Discussion ................................................................................................ 50 Chapter 7: Conclusions and Limitations .................................................................... 55 Bibliography ............................................................................................................... 72 v List of Tables Table 1. List of States with 287(g) and Fast-Track Programs .................................... 61 Table 2. Summary of Dependent and Independent Variables for USSC Federal Sentencing Data, FY2008 ........................................................................................... 62 Table 3. Descriptive Statistics for USSC Federal Sentencing Data, FY2008 ............ 63 Table 4. Descriptive Statistics by Ctizenship Status for USSC Federal Sentencing Data, FY2008 .............................................................................................................. 64 Table 5. Correlation Matrix for Level 1 Individual-Level Predictors ........................ 65 Table 6. Correlation Matrix for Level 2 Contextual Predictors .................................. 65 Table 7. Models of Case-Level Factors for Sentence Length ..................................... 66 Table 8. Unconditional Two-Level HLM Model of Sentence Length ....................... 66 Table 9. Two-Level HLM Random Intercept Model ? Individual-Level Fixed Effects ..................................................................................................................................... 67 Table 10. Two-Level HLM Random Intercept Model ? Individual-Level and Contextual Fixed Effects............................................................................................. 68 Table 11. Two-Level HLM Random Coefficient Model ? Individual and Contextual- Level Fixed Effects ..................................................................................................... 69 Table 12. HLM Cross-Level Citizenship Status Interaction Models .......................... 70 vi List of Figures Table 1. Sentence Length in Months for Incarcerated Offenders ............................... 71 Table 2. Natural Log of Sentence Length in Months for Incarcerated Offenders ...... 71 1 Chapter 1: Introduction Researchers have traditionally focused on the role of race and?more recently? ethnicity on sentencing disparities (Demuth & Steffensmeier 2004; Mitchell 2005; Ulmer 2012). Even though unwarranted disparities on sentencing have been a major concern for policymakers and researchers for decades, they still continue to be an aspect that has not been definitively resolved (Mitchell & MacKenzie 2004; Tonry 1996). At the federal level, one of the issues that has arisen in the last two decades is the increasing number of non-citizens being sentenced by the district courts. Research on the role of citizenship on federal sentencing has increased in the last decade, but in comparison to other aspects, this is one that still requires more in-depth examination (Ulmer 2012). One topic that warrants more discussion is the ostensible contextual disparity in the punishment of Hispanic non-citizens. This paper attempts to assess this inquiry and explain it by including factors intended to measure the contextually differentiated presence of Hispanic immigrants, as well as the enforcement of the immigration laws across the nation. Ethnicity and Citizenship Status on Sentencing Research on the role of ethnicity on sentencing has increased in the last two decades (Demuth & Steffensmeier 2004; Johnson et al. 2011; Mitchell & MacKenzie 2004; Steffensmeier & Demuth 2000, 2001; Ulmer 2012; Warren et al. 2012). The role of ethnicity on sentencing has largely been studied with reference to individuals of Hispanic origin. The last decades have been characterized by an increasing Hispanic population. According to the 2010 Census (U.S. Census Bureau 2011a), 50.5 million people residing in the United States are of Hispanic origin, which 2 constitutes 16% of the population. Hispanics have increased from 22.4 million in 1990 to 35.3 million in 2000, when this group made up 13% of the population. Between 1990 and 2000, the Hispanic population increased by 57.9% (U.S. Census Bureau 2001). It is estimated that Hispanics will be 30% of the nation?s population by July 1, 2050 (U.S. Census Bureau 2011b).1 Apart from the traditionally higher Hispanic birth rates, immigration has played a major contribution in this regard. This Hispanic population trend has affected the demographics of the U.S. correctional and criminal justice systems. According to the Bureau of Justice Statistics (2011), the estimated male and female Hispanic prison population under state and federal jurisdictions rose from 206,900 and 10,000 in 2000, to 327,200 and 18,700 in 2010, respectively. Hispanics have been the fastest growing group being imprisoned, comprising 20% of the state and federal populations in 2005, a rise of 43% since 1990 (The Sentencing Project 2007). This pattern has been even more notorious at the federal level. In 1992, Hispanics accounted for less than 25% of offenders in the federal courts, but over 40% in 2004 (Light 2010). By March 2012, the Hispanic prison population at the federal level represented a 34.7% of the total (Federal Bureau of Prisons 2012). Ethnicity and citizenship status are closely related. The great majority of non- citizens in the United States are of Hispanic origin. According to the Department of Homeland Security-DHS (2012a, 2012b), both the estimated legal permanent and unauthorized immigrant population residing in the United States in 2011 was 13,070,000 and 11,510,000 people, respectively. The largest single percentage 1 Nonetheless, it should be noted that in the previous four years the influx of Hispanic immigration to the U.S. has noticeably declined (Pew Hispanic Center 2012). 3 corresponded to Mexican immigrants, who represented approximately 3,320,000 (25.4%), and 6,800,000 people (59%), respectively. The percentages of all the other countries of origin were far lower, but the cumulative proportion of the most representative Latin American countries apart from Mexico reached the 17.2% and 16%, respectively. This means that at least 43% of non-citizen legal residents and 75% of the unauthorized immigrant population residing in the United States are of Hispanic origin. Overall, the combination of these estimates reveals at least 58% of non-citizens residing in the United States are Hispanics. The presence of non-citizens in the United States correctional system is noticeable. The Federal Bureau of Prisons (2009) reported that in FY 2009 more than one quarter of the inmate population was non-U.S. citizens (26.7%). The majority of them were Hispanics, with Mexico contributing to the highest proportion. The federal criminal justice system has witnessed a steady increase of non-citizen defendants. An early report of the Bureau of Justice Statistics (1996) revealed that the number of non- citizens processed in the federal criminal justice system increased an average 10% annually from 1984 to 1994. This contrasts with the overall rate of the federal criminal caseload, which increased on average less than 2% annually in the same period of time. More recently, the United States Sentencing Commission (USSC 2008) reported that the proportion of non-citizen offenders rose from 22.7% in FY 1991 to 37.4% in FY 2007. Moreover, since FY 1997, Hispanics have accounted for more than 80% of non-citizen offenders. This means that a large and increasing number of Hispanic non-citizens are being sentenced in the United States. 4 The Pervasive Impact of ?Crimmigration Law? on Hispanic Non-citizens Deportation is a major concern for non-citizens in general and Hispanics in particular (Bibler 2011; Cruz 2010; Guerra 2008; Welch 2004). According to the Pew Hispanic Center?s 2008 National Survey of Latinos (Lopez & Minushkin 2008), approximately 40% of Latinos say they worry a lot about deportation, and an additional 17% say that they worry some that they themselves, a family member or a close friend may be deported. This is a slight increase from 2007, when the magnitude of these concerns reached the 53%. Several qualitative and quantitative studies show that one of the most relevant preoccupations for Hispanic immigrants is the fear of being deported and its consequences (Arbona et al. 2010; Bhuyan 2008; Brabeck & Xu 2010; Brotherton & Barrios 2009; Brotherton & Martin 2009; Bucher et al 2010; Cervantes et al. 2010; Das 2008; DiDenti 2010; Fussell 2011; Gonzales & Chavez 2012; Hagan et al. 2011; Landale et al. 2011; Phillips et al. 2006). Although the dramatic effects of deportation are likely to reach family and community members regardless of their citizenship status, deportability makes an outstanding difference between citizens and non-citizens because U.S. citizens do not face the threat of deportation.2 This is not the case for non-citizens and indeed not for unauthorized residents. Even though unauthorized immigrant residents can be deported just because of their illicit presence in the country, legal authorized as well as unauthorized immigrant residents formally share the same likelihood of deportation when they commit the sort of offenses that trigger the deportation mechanisms (Reyes 2012). Deportation has been traditionally considered an immigration policy instrument. However, nowadays some authors have thoughtful 2 Exceptions include cases of naturalized U.S. citizens that can lose their U.S. citizenship. 5 reasons to believe that it is being used also as a crime control mechanism (Sklansky 2012). Laws passed during the last two decades have increased the threat of deportation, which has even made the distinction between citizens and non-citizens regardless of their residency status more critical (Bibler 2011). In the last three decades, the boundaries that delimit immigration enforcement and criminal justice have blurred (Brody 2011; Frey & Zhao 2011; Hagan & Phillips 2008; Hagan et al. 2008; Kanstroom 2000; Reyes 2012; Sklansky 2012; Stumpf 2011). Immigration has always been a multifaceted phenomenon, but now more than ever it is also immersed in the realm of crime control as gate-keeping, which emphasizes the importance of identity and territory in an era of globalization and massive labor force mobility (Allegro 2010; Bibler 2011; Bloch & Schuster 2005; Bosworth 2012; Ewing 2010; Fassin 2011; Flynn 2005; Robinson 2006; Tambini 2001; Zedner 2010). This paradoxically has given rise to what is called ?crimmigration law? (Bosworth 2012; Franko 2011; Murphy 2012; Reyes 2012; Sklansky 2012; Stumpf 2011; Zedner 2010). Zedner (2010) associates the term crimmigration with the concept of criminalization of immigration. More thoroughly, Stumpf (2011) considers that crimmigration law is a phenomenon by which ?criminal and immigration law intersect both formally and functionally, magnifying the government's exclusionary power? criminal and immigration law combine to expand the circumstances under which the government imposes immigrations consequences for crimes, including expulsion, detention, or incarceration? legislatures have increasingly defined immigration-related conduct as criminal? (p. 10). 6 A number of authors refer to the existence of a separate criminal justice system for ?enemy foreign nationals? (Fekete & Webber 2010; G?mez-Jara 2008). This phenomenon has been magnified due to the recent occurrence of terrorist attacks that have promoted even more tough immigration policies (Bloch & Schuster 2005; Bosworth & Guild 2008; Dubber 2010; Flynn 2005; Hagan & Phillips 2008; K. Johnson & Trujillo 2007; Pauw 2000), a phenomenon often referred as ?securitization? (Bibler 2011). The war against organized crime and terrorism is since then being used to justify the introduction of criminal justice legislation and practices that empower the criminal justice system (Guerra 2008; McLaughlin 2011). Punishment of immigrant crime performs the double task of strengthening values of national identity, and protecting the local and the national population from threatening foreign elements (Diaz 2011; Franko 2011; Mendelson 2010). The first important legal reform that converged immigration and criminal laws took place with the 1986 Immigration Reform and Control Act (IRCA) (Frey & Zhao 2011; Lawston & Escobar 2009). This reform increased funding for U.S. border patrol agents and penalized employers who knowingly hired undocumented workers (Ewing 2010; Hagan & Phillips 2008). The Anti-Drug Abuse Act of 1988 (ADAA) marked the inception of a legislative trend toward increased removal of non-citizens with criminal convictions (Welch 2004). This act set the category of ?aggravated felonies? for justifying deportation, which at that time comprised a narrow class of offenses such as murder, drug trafficking, and firearms trafficking (Davenport 2006). The Immigration Act of 1990 added ?crimes of violence? as a subcategory of aggravated felonies with a term of imprisonment of five years or more, and expanded 7 this category to include the actual or attempted illicit trafficking of any controlled substance and any money laundering offense, and both federal and state convictions (Davenport 2006; Welch 2004). A border enforcement initiative implemented in 1993 ?called ?Prevention through Deterrence??has collaterally led immigrants to cross the border through rural and life-threatening areas where border patrol might have a tactical advantage (Alonso 2003; Hagan & Phillips 2008; Quinn 2011). In 1994, the Immigration and Nationality Technical Corrections Act further expanded the definition of aggravated felony and added even more crimes, for example, theft offenses sanctioned with at least five years of incarceration, child pornography, alien smuggling, and the attempt to commit these offenses (Welch 2004). This legislative approach was substantially exacerbated with two reforms passed in 1996: the Anti-Terrorism and Effective Death Penalty Act (AEDPA), sanctioned in response to the Oklahoma bomb attack of April, 1995 (Welch 2004); and the Illegal Immigration Reform and Immigrant Responsibility Act (IIRIRA), labeled by many critics as the ?Mexican Exclusion Act? (Hagan & Phillips 2008). This legislation transformed the immigration regime, making it even more restrictive (Lawston & Escobar 2009; McDermid 2001; Morawetz 2000). Judicial review for all categories of immigrants eligible for deportation was virtually eliminated (Morrison 1997; Legomsky 2006). Suspended sentences were considered actual sentences for immigration enforcement purposes, and a wide range of crimes became subject to mandatory detention (Morawetz 2000). Sentence length for crimes of theft or violence to be considered aggravated felonies was reduced from five years to one year (Davenport 2006; Welch 2004). 8 In addition, the categories of non-citizens subject to detention and deportation were expanded (Brody 2011; Hagan & Phillips 2008; Montague 2001; Morawetz 2000). For instance, ?expedited removals? were intensified at the expense of voluntary returns. The definition of aggravated felonies was broadened to include 28 offenses (Hagan & Phillips 2008; Morawetz 2000). Likewise, convictions for the broad and subjective category of crimes of moral turpitude remained as a basis for deportation (Dadhania 2011; Moore 2008). These reforms included a retroactive provision that allowed deportations due to aggravated felonies committed before the passage of these laws (Hagan & Phillips 2008; Morawetz 1998). Finally, the PATRIOT Act enlarged even more the net of immigration control by increasing administrative powers to detain and deport immigrants perceived as threats to national security (Hagan & Phillips 2008). Deportation may be seen as an available mechanism for getting rid of threatening aliens (Bloch & Schuster 2005; Gibney 2008; Kanstroom 2000; Pauw 2000). One of the most relevant manifestations of this policy approach is early disposition programs, so called fast-track sentencing. In essence, this is a type of guideline downward departure that allows the judge to reduce the length of the incarceration term in those cases in which the defendant and the prosecutor have signed an agreement under the specific conditions of a program established in the sentencing district. Moreover, the characteristics and history of these programs reveal their intimate relationship with immigration enforcement. Initially, these programs were de facto implemented in some southern border districts during the 1990s. In 2003, the PROTECT Act acknowledged this reality and 9 established a more formalized scheme of fast-track sentencing (Cho 2010; Gorman 2009). Then, the U.S. Sentencing Commission promulgated USSG ? 5K31 (Gorman 2009), authorizing a downward departure of no more than four levels as to whether the Government files a motion for such departure based upon a disposition from a program previously authorized by the Attorney General. This authority issued a memo (Ashcroft 2003) in September 2003 authorizing the implementation of fast- track programs, whether a district confronts a vast number of a specific class of offenses or faces an exceptional circumstance related to a particular type of case that may significantly strain prosecutorial and judicial resources; those cases should also be highly repetitive and present substantially similar fact scenarios, and not involve offenses designated as ?crimes of violence.? Most of fast-track programs have been implemented in southern border districts, and target immigration offenses, with the great majority of them corresponding to illegal reentry after deportation cases (Cho 2010; Gorman 2009; Ogden 2009).3 However, some districts also have programs that target drug-related offenses (Ogden 2009). The specific characteristics and functionality of each of these programs vary by district. On the whole, fast-track programs accelerate the disposal of criminal cases, and serve the purpose of advancing immigrant deportation proceedings (Cho 2010). The use of the criminal justice system as a mechanism for immigration control is also noticeable in the cooperative efforts maintained between federal immigration enforcement and local and state law enforcement. The 287(g) provision is a key and 3 A Memorandum from the Deputy Attorney General dated January 31, 2012, established new and more rigorous requirements for illegal reentry fast-track sentencing programs. 10 controversial feature of IIRIRA that falls within this approach (Coleman 2012; Hagan et al. 2011). Throughout this devolution policy,4 the federal government is authorized to sign a memorandum of agreement with state and local law enforcement in order for them to participate in immigration control, an attribution that corresponded only to federal immigration agents (Parrado 2012; Wishnie 2011). The implementation of 287(g) programs started in Florida only after the Department of Justice issued a memo in 2002 announcing that states have the ?inherent authority? to enforce civil provisions of immigration law (Armenta 2012; Coleman 2012; Parrado 2012; Wishnie 2011). Following Florida, Alabama in 2003 and Arizona in 2005 were the next to implement these kinds of programs, followed by many other states and local governments. In 2007 and 2008, the application of these programs significantly increased with the inclusion of 26 and 34 new jurisdictions, respectively (Parrado 2012). The implementation and effects of these programs vary across jurisdictions (Coleman 2012; Hagan et al. 2011). Not surprisingly, this emphasis on deporting unauthorized and convicted immigrants has been reflected in the deportation rates. The number of deportations has been experiencing a continuous increase with the passing of the 1996 laws, and more clearly since 2002 when the number of deported immigrants more than doubled from 165,000 in that year to 393,000 just seven years later (Parrado 2012). The average daily immigrant detention population rose from 9,011 in FY 1996 to 21,133 in FY 2003 (Phillips et al. 2006). Deportations for criminal and non-criminal reasons 4 In summary, devolution policies define the transferring of competencies that traditionally belonged to the federal government, to the state and local regimes. 11 increased from 33,842 and 17,082 in 1995, to 188,382 and 203,571 in 2011, respectively (Department of Justice 2002; Department of Homeland Security 2012). Deportation has usually been qualified as a collateral consequence of a criminal sanction (Budeiri 1981; Pinard 2006). Nevertheless, many scholars argued that deportation is in fact a punishment and the most important consequence of a conviction for many non-citizen defendants (Chin & Love 2010; Cruz 2010; Das 2008; Dow 2007; Kanstroom 2000; McDermid 2001; Murphy 2012; Ortiz 2011; Pauw 2000; Reyes 2012). For instance, Ortiz (2011) reasoned that today deportation serves incapacitation, deterrent, and retributive purposes, whereas Das (2011) pointed out that deportation has essentially turned into a ?mandatory minimum? penalty for many types of criminal convictions. The recent Supreme Court decision in Padilla v. Kentucky (March, 2010) made it even clearer that deportation is a direct consequence of a criminal conviction for many non-citizen defendants (Kanstroom 2011). In Padilla, the Court held that defense counsel?s failure to advise a non-citizen client of the deportation consequences of a conviction constituted ineffective counsel in violation of the Sixth Amendment (Chin & Love 2010; Ewald 2011). The Court reasoned that since the recent legislative changes expanded the list of deportable offenses and restricted the repeal powers of judges, now deportation is virtually inevitable and automatic for a vast number of non-citizens convicted of crimes. In spite of the fact that this legislative and policy approach on immigration enforcement can be considered a national tendency, it is likely that its impact varies throughout the country. First of all, the magnitude and impact of the recent 12 immigration trends differ across jurisdictions. The immigrant population substantially contrasts between regions and locations. Urban metropolises such as Los Angeles, Houston, and Miami are characterized by vast Hispanic communities, whereas other areas especially in the Midwest have minor Hispanic populations (U.S. Census Bureau 2011a). The proportion of the unauthorized immigrant population also varies across states. The estimated unauthorized immigrant population as of January 2011 was 2,830,000 in California, 1,790,000 in Texas, 740,000 in Florida, and 630,000 in New York (DHS 2012a). The number of Hispanics incarcerated also diverges across territorial circumscriptions. To illustrate, as of 2008, there were 17,010 non-citizens incarcerated in California and 9,940 in Texas, whereas only 16 in Maine and 12 in North Dakota (Bureau of Justice Statistics 2008). Immigrants fare differently through states in a variety of aspects, such as median income, home ownership, and poverty (Pew Hispanic Center 2010). Finally, Latino political representation is also divergent among jurisdictions. For example, Latino legislators in state legislatures during 2009 represented 14% in Arizona, 23% in California, and 44% in New Mexico, whereas less than 5% in Connecticut, Maryland, and Iowa (National Conference of State Legislatures 2009). More consequential are the differentiated statistics regarding deportation and immigration enforcement actions. The number of deportation proceedings initiated among the thirty immigration jurisdictions differs substantially. For instance, according to the Transactional Records Access Clearinghouse Immigration Project (2008), in 2008, Texas took the lead with a total of 42,509 deportation proceedings, 13 which constituted the 18.58%, followed by California with 33,342 (14.57%), Arizona with 19,392 (8.48%), and Florida with 19,075 (8.34%). Conversely, in the rear were Oregon with 729 (.32%), North Carolina with 997 (.44%), Connecticut with 1,447 (.63%), and Tennessee with 2,076 (.91%).5 These figures underscore the fact that the incidence of deportation and immigration enforcement is different across states and jurisdictions. This may not be only a function of the differential presence of Hispanic immigrants across the U.S. territory, but also a consequence of the immigration policy choices made in each location. Even though the federal government is in charge of the homeland security of the country, as mentioned above (p. 10), it has established devolution policies that allow state and local law enforcement authorities and officers to perform immigration enforcement activities, such as the mentioned 287(g) programs. This means that there are jurisdictions that seem more proactive than others at enforcing the immigration law, and as a consequence of that, they seem more interested in the swift removal of immigrants. Sentencing courts are likely to be impacted by these contextually differentiated immigration policy approaches and environments (Chin 2011). This review illustrates how close being a convicted non-citizen is now associated with being deported. This reflects the increasing important role of deportation on non-citizens? sentencing outcomes. It also highlights the contextually differentiated emphasis placed on the removal of immigrants. In this context, the current study is an attempt to advance the quantitative research on the impact of defendant citizenship on sentencing. This paper echoes the argument that immigration 5 It should be noted that these figures does not represent the number of deportations executed during 2008, nor the total number of proceedings concluded in that year. They represent the total number of deportation proceedings initiated in 2008 in each immigration enforcement jurisdiction. 14 enforcement is a fundamental factor for understanding non-citizen sentencing (Chin 2011), but with a particular focus on the contextually differentiated emphasis placed on them. Concretely, this paper will expand on existing sentencing research by 1) examining contextual variations in the punishment of Hispanics, 2) including immigration enforcement contextual pressures as a key element of that analysis, and 3) measuring contextually differentiated approaches and attitudes toward deportation and immigrant population size, and their impact on the punishment of Hispanics. 15 Chapter 2: Literature Review Empirical Research on the Impact of Citizenship Status on Sentencing Research on the role of defendant citizenship on sentencing has expanded in the last decade. Initially, it was typically included as a control variable in studies, but more recent studies have attempted to conduct more comprehensive analyses on the effect of citizenship directly. Early studies tended to agree that non-citizens are treated more harshly than U.S. citizens, which means that they are more likely to be incarcerated and to receive longer imprisonment sentences (Albonetti 1997; Mustard 2001; Steffensmeir and Demuth 2000). These effects were conditioned on the defendant?s race and ethnicity. For instance, Albonetti (1997) found that in drug cases, for Black and Hispanics, but not for Whites, being a non-citizen significantly increases sentence severity. On the same path, Steffensmeir and Demuth (2000) found that non-citizen Hispanics were sentenced more harshly than non-citizen Blacks and especially Whites, but that the Hispanic-Black difference was trivial. Nevertheless, Mustard (2001) found that these differences could be almost entirely attributed to the differentiated use of downward departures. When the sample was limited only to cases sentenced within the guidelines limits, disparities no longer existed on the basis of citizenship. U.S. citizens were significantly more likely to receive downward departures, and at a larger magnitude. This last finding has been mostly confirmed by more recent studies (Johnson et al. 2008; Johnson & Betsinger 2009; Light 2010; Spohn 2005; Ulmer et al. 2010; Wolfe et al. 2011). For instance, Johnson and colleagues (2008) found that non- 16 citizens were 27% less likely to receive substantial assistance departures, and 15% and 5% shorter reductions for regular and substantial assistance departures, respectively, than citizens. Conversely, Johnson and Betsinger (2009) found that even though Hispanic non-citizens were less likely to receive substantial assistance departures, they were more likely to receive other types of downward departures. Recalling the discussion about the increasing trend of criminalization of immigration, it is not surprising that sentencing outcomes for non-citizens have experienced a noticeable fluctuation. This is confirmed by Light?s (2010) comprehensive assessment of the changes on ethnic disparities in punishment between 1992 and 2004. In the first place, this study found that the odds of being incarcerated have increased over time for Hispanics in comparison to Whites, especially between 2002 and 2004. However, this trend has largely been attributable to citizenship status trends. This means that while ethnic disparity has remained stable, disparities related to citizenship status have noticeably increased over time. For sentence length, this study found that ethnic disparity has reduced over time, especially between 2002 and 2004. In this case, citizenship status only partially explained this trend, with ethnicity itself still being significantly associated with sentence length disparity reduction over time. In the case of regular downward departures, the gap between Hispanics and Whites has decreased over time, but again this change was in part explained to the differential trending by citizenship status. Not surprisingly, citizenship status had no effect on the White or Black trend. Finally, a similar pattern in the opposite direction was found for substantial assistance departures. In this case, the gap between Hispanics and Whites had significantly 17 increased. However, this trend was anew mostly attributable to the differential trending by citizenship status. Light?s (2010) study revealed a subtle pattern on non-citizen?s sentencing, portraying a picture in which Hispanic citizens tend to be more harshly sentenced especially since the beginning of this century. Nonetheless, this effect is more straightforward for incarceration, but far less clear for sentence length. Moreover, more recent studies exhibit a similar pattern (Demuth 2002; Johnson & Betsinger 2009; Wolfe et al. 2011; Wu and DeLone 2012). For instance, Demuth (2002) compared drug cases sentencing outcomes for foreign legal and illegal residents and U.S. citizens, and found that the probability of incarceration was 30% and 44% greater for the former two groups compared to the latter. However, there were not significant disparities in sentence length. Logue (2009) also conducted a study that distinguished between Latino documented and undocumented immigrant drug trafficking offenders, and found that country of origin does have an impact on sentence length among non-citizens. Significant differences between non-citizen Mexicans and non-Mexican Latinos were evidenced, with the formers being more harshly sentenced than other Latinos. Consistent with this tendency, Shermer and Johnson (2010) found that non- citizens were more likely to receive charge reductions, which the authors speculated was due to the fact that illegal aliens are often deported when prosecuted in federal courts. Spohn (2005), and Wu and Spohn (2010) focused only on three districts (Minnesota, Nebraska, and Southern Iowa) and found significant effects for citizenship status only at the aggregate level for sentence length and substantial 18 assistance departures. Non-citizens were more likely to receive shorter sentences, but lower odds of receiving substantial assistance departures. Conversely, it is to note that Wu and DeLone (2012), in their analysis of 17 districts in Arizona, California, Florida, New Mexico, New York, and Texas, found that in the case of immigration offenses, non-citizens tended to receive longer sentences than citizens, and that early disposition departures did not have a particularly strong effect on sentence length reduction for non-citizens. These findings, largely corroborated by Hartley and Tillyer (2012), seem to contradict the main purpose of fast-track sentencing (accelerating non-citizen?s sentencing), which reveals the importance of investigating variations across jurisdictions characterized by heavy immigration caseloads. The studies conducted by Wolfe and colleagues (2011), and Iles (2009) highlight the relevance of differentiating between legal and illegal residents. Concretely, the former found that U.S. citizens were more likely to receive substantial departures (51%), compared to foreign legal residents (39%) and foreign illegal residents (14%). Conversely, in the case of fast-track departures, illegal aliens were more likely to receive this type of departure (54%), compared to legal aliens (20%) and citizens (5%). Interestingly, resident-legal and illegal aliens were 37%, and 9.5 times more likely, respectively, to be imprisoned than U.S. citizens. However, there were no differences for resident-legal aliens compared to citizens in sentence length, but illegal aliens received prison terms 5% shorter than citizens. Illegal aliens were significantly more likely to be sent to prison if detained prior to trial. Even though Iles (2009) exclusively focused on the district court of the U.S. Virgin Islands, it is interesting to note that defendants classified as legal aliens, but 19 citizens of the Dominican Republic, tended to receive longer sentences than those imposed on U.S. citizens. The author suggested that Dominican citizenship can be seen as a proxy for danger to the society, with harsher sentences meted out in the interest of the community. Empirical Research on Contextual Variation in Punishment by Citizenship Status In general, differences across jurisdictions have motivated scholars to investigate contextual variations in punishment and their relationship with race and ethnicity. Researchers have found that the effects of race and ethnicity on sentencing varies across jurisdictions and are conditioned by contextual factors that shape the environment and culture of the judicial decision-making process (Britt 2000; Farrell & Ward 2011; Helms & Jacobs 2002; Jacobs & Carmichael 2002; Johnson 2005; Johnson 2006; Johnson et al. 2008; Kautt 2002; King, et al. 2010; Kramer & Ulmer 1996; Ulmer & Johnson 2004; Ulmer 2005; Ulmer et al. 2010). At the federal level, several studies have found district and circuit variations in the effects of individual variables as well as the significant role of contextual factors on sentencing (Kautt 2002; Ulmer 2005; Ulmer et al. 2010). However, research on contextual variations in the punishment of non-citizens is scarce. Some authors (Johnson et al. 2008; Johnson & Betsinger 2009; Hartley and Tillyer 2012) have suggested that this is an aspect that warrants more research, whereas others (Hartley & Armendariz 2011; Spohn 2005; Wu & Spohn 2010) have examined differences across jurisdictions, but their analyses have been limited to a few districts and have not included contextual variables to explain these differences. 20 Johnson, Ulmer and Kramer (2008) reported that citizenship status was one of the individual-level effects on departure lengths that varied significantly across federal districts. Johnson and Betsinger (2009) considered that Hispanic non-citizens are particularly likely to receive other downward departures because downward departures represent one structural mechanism for managing large immigration caseloads in Southwestern districts, by providing more discretion to prosecutors and offenders to obtain reductions in the severity of the punishment and, consequently, leverage for the early termination of criminal proceedings. This suggests that it is likely that non-citizens may be treated differently across jurisdictions. In their analysis of immigration offenses, Hartley and Tillyer (2012) also reported that cases prosecuted in border jurisdictions received increased sentence lengths compared to non-border districts, and that U.S. citizens were more likely to receive shorter sentences in all districts except West Texas. A comparison between the studies conducted by Spohn (2005) and Wu and Spohn (2010), and Hartley and Armendariz (2011), seems to confirm this assertion. For instance, the two former studies did not find consistently significant effects for citizenship on sentencing across the three jurisdictions that were examined (Minnesota, Nebraska, and Southern Iowa), and only at the aggregate level for sentence length and substantial assistance departures. It is fair to recall that these districts are not characterized by large recent immigration trends. Conversely, Hartley and Armendariz (2011) conducted their study in three districts heavily affected by large immigration influxes, such as Southern California, Arizona, New Mexico, and South and West Texas. Although their analysis was limited to departures and 21 sentence length for drug trafficking offenses, it is interesting to note from this study the numerous significant varied effects that citizenship status had on sentencing outcomes for different types of drugs across jurisdictions. A review of the literature on the role of citizenship on federal sentencing reveals an increasing accumulation in the last ten years. However, the suggested close relationship between immigration pressures and sentencing has still not been completely examined. Moreover, an analysis of the possible contextual variations in the punishment of non-citizens and its relationship with immigration and its enforcement is pending. Some studies have compared sentencing outcome across jurisdictions, but the analyses have been limited to few locations, and they have not explicitly included contextual variables. No study has explicitly addressed the contextual impact of immigrant population concentration and enforcement on sentencing. The current study expands on sentencing research by examining contextual variations in the punishment of Hispanic defendants, with a focus on citizenship status. The inclusion of the idea of immigration threat as a key element of the analysis is an approach that distinguishes this from most previous studies. 22 Chapter 3: Theoretical Framework The blurring of the conceptual and practical boundaries between immigration enforcement and criminal law has led scholars to quote the terms crimmigration law (Stumpf 2011) and immigrationization of criminal law (Frey & Zhao 2011). Although this is a tangible phenomenon, it is not completely clear whether the criminal law is expanding into the realm of immigration law or vice versa (Sklansky 2012). Immigration enforcement and the criminal justice system are interchangeably and conveniently used to target criminal aliens (Stumpf 2011). In this regard, the philosophical boundaries between the traditional categories of law have become more unclear and driven mostly by practical considerations rather than theoretical or thoughtful reasoning (Ortiz 2011). This perspective is deeply marked by skepticism of formal legal categories, their stability and coherence, and by the distrust in the idea that official discretion could and should be controlled (Sklansky 2012). From this, it follows that the reduction of discretion through formal legal techniques, such as sentencing guidelines, is at least a na?ve approach. Savelsberg (1992) has argued that sentencing guidelines cannot overcome substantial societal and political conditions that feed disparities because they ignore the sociology of modern societies, characterized by bargaining rather than the rigid application of formal rules. Law enforcement agents, prosecutors, and immigration officials are encouraged to see criminal and immigration laws as different kinds of tools to be used conveniently according to the circumstances (Sklansky 2012). Moreover, the war on terror has prominently increased the court?s focus on procedural rather than 23 substantive issues not because of their inherent importance, but instead because they are using process as an instrumental tool (Martinez 2008). This instrumental use of criminal and immigration laws and enforcements is conceptualized by Sklanksky (2012) as a prototypical case of what he refers to as ad-hoc instrumentalism. According to this perspective, in some cases, the tools work best in combination, as when an individual suspected of being in the country illegally is arrested for traffic offenses and transferred to immigration authorities. In other cases, immigration enforcement can be used to strengthen criminal justice when a non-citizen is denied bail because of an immigration requirement. Finally, in other cases, criminal prosecutions are used to achieve immigration objectives, by convicting non-citizens of crimes that trigger mandatory deportation, or by promoting plea agreements that include a waiver of claims that might otherwise block deportation. In this latter case, the criminal justice system has been restructured to allow for agency control and promotion of immigration objectives within the criminal prosecution (Eagly 2010). Overall, this theoretical standpoint conceives that the criminal justice system could be used as a mechanism for targeting non-citizens and deporting them. The Federal Sentencing Commission has determined that courts are prohibited from using, among other factors, defendant?s national origin as a determinant for their decisions (USSG ? 5H1.10). Nevertheless, Congress incorporated (8 U.S.C. ? 1228c5) deportation into federal plea bargaining and sentencing by authorizing the stipulation of deportability as part of a plea bargain (Chin 2011). This means that although citizenship status solely by itself is not to be considered as a legal determinant of sentencing, deportation could be. Given that deportation is an outcome 24 only possible for non-citizens, in practice, a differential treatment for non-citizens at sentencing is justified on the grounds of their expedited removal from the country. It should be noted that according to the United States Code (Title, 8, Chapter 12, Subchapter II, ? 1231), an alien who is imprisoned is not to be deported until the completion of his or her sentence. Nevertheless, there are a couple of exceptions established by the same law. An alien may be removed before the completion of the sentence if he or she was convicted for a non-violent offense (excluding any offense related to smuggling or harboring of aliens, or most of those considered aggravated felony), or the removal is appropriate and in the best interests of the United States. However, these exceptions are only applicable in those cases in which the alien is under the custody of the Attorney General. The review conducted in the introductory part of this paper revealed that deportation has become an unavoidable consequence for almost every non-citizen convicted for any federal crime. Therefore, under these conditions, non-citizens do not have enough bargaining leeway to negotiate any concession from prosecutors and judges (Taylor and Wright 2002). In this context, even though the U.S. sentencing guidelines allow courts to mitigate a sentence based on a defendant?s agreement not to contest deportation, it is likely that more frequently they use it as a reason for sentencing within but at the lower end of the guidelines grid (Chin 2011). The fact that judges may rely on the intrinsic deportability status of non- citizens to mete out their sentences can be explained through the lens of Albonetti?s (1991) uncertainty avoidance/causal attribution perspective. According to this perspective, the inability to accurately predict future criminal behavior is a problem 25 that surrounds the work of courtroom workgroup actors. In this sense, judges would rely on stereotypes that link defendant?s individual?s traits and case processing outcomes to assess the defendant?s disposition toward future criminal activity. Therefore, in an attempt to achieve rationality?a bounded rationality?in their decisions, judges make satisfying causal attributions about the crime and the offender based on societal stereotypes. Steffensmeier, Ulmer, and Kramer (1998) elaborated on this standpoint and formulated a three-focal concerns perspective to more specifically explain it. The first, offender?s blameworthiness, is a legally considered factor that indicates that the defendant?s potential punishment increases depending on the offender?s culpability and the degree of injury caused. The second, protection of the community, focuses on the need to incapacitate the offender or to deter would-be offenders. The third, practical implications of sentencing decisions, refers to both organization and individual factors that come into play in sentencing decisions. The first two focal concerns provide a more specific theoretical support for explaining the consideration of deportation on the punishment of Hispanic non- citizens. In relation to offender?s blameworthiness, judges may consider Hispanic non-citizens doubly culpable because they violated the law of a country that welcomed them and in exchange they became a burden for the nation (Demleitner & Sands 2002; Wolfe et al 2011).6 Moreover, Hispanic non-citizens are commonly 6 To illustrate, in United States v. Onwuemene, the district court explicitly said that ?The other thing that I feel that warrants imposition at the high end of the guideline range: You are not a citizen of this country. This country was good enough to allow you to come in here and to confer upon you ... a number of the benefits of this society, form of government, and its opportunities, and you repay that kindness by committing a crime like this. We have got enough criminals in the United States without importing any.? 26 perceived as more delinquent-prone, in spite of the empirical evidence in contrary (Hagan & Palloni 1999; Sampson 2008; Stowell et al. 2009). This may be tied to the augmented perceived flight risk of non-citizens (Schlesinger 2005). Nonetheless, in relation to the second focal concern, judges may consider that the protection of the community is best achieved by accelerating the removal of Hispanic non-citizen offenders from the U.S. territory rather than imposing lengthier incarceration terms. Deportation may be seen as a powerful mechanism for deterrence, retribution and, most notably, incapacitation (Ortiz 2011). In this sense, by imposing shorter imprisonment terms on Hispanic non-citizens, judges may at the same time speed up their banishment from the country by accelerating their deportation proceedings, and give the message that non-citizen would-be offenders will be subjected to expedited removals (Hartley & Tillyer 2012). The practically unavoidable link between deportation and lack of citizenship may be taken by sentencing judges as a source of certainty regarding the future of a Hispanic non-citizen defendant. These considerations lead to the following: Hypothesis 1: To expedite deportation, after controlling for legal considerations, Hispanic non-citizens will receive shorter sentences than Hispanic citizens. Empirical reasons to expect significant contextual variations in the punishment of Hispanic non-citizens in relation to the differentiated contextual emphasis placed on the removal of immigrants from the U.S. territory were provided in the introductory part of this paper. Likewise, this expectation is also supported by thoughtful theoretical reasons. Judges are not isolated from the context in which they 27 work and their preferences and expectations are likely to be substantially conditioned by the surrounding cultural, political, and economic conditions (Myers & Talarico 1987). According to a contextual perspective on sentencing, the sentencing process is influenced by the political, social, and organizational context of the court (Dixon 1995). Moreover, the social worlds perspectives (Ulmer 1997) posits that in practice externally imposed sentencing standards vary in their usage and application by local relationships and informal decision-making criteria (Kautt 2002). The criminal law varies with the societal context in its relation to the political sector and substantive societal conditions influence its use and interpretation (Savelsberg 1992). The role of individual and case processing factors is likely to be conditioned by the court context (Britt 2000; Kautt 2002). The focal concerns perspective also hypothesizes that their meaning, emphasis and interpretation is local (Ulmer & Johnson 2004). Ulmer (2005) explained that the courts work as communities based on participant?s shared workplace and interdependent working relations among key sponsoring agencies. The distinctive organizational cultures and relationships of workgroup courtrooms establish formal and informal case processings and sentencing norms. The influence of external organizations and externally imposed policies is a function of those organizational patterns. In addition, Quinney?s Social Reality of Crime perspective (1975) also supports the differentiated impact of contextual conditions on sentencing. According to this viewpoint, the application of the criminal law varies according to the extent to which the behaviors of the powerless conflict with the interests of those in power. In that sense, the judicial activity will increase when the interests of the dominant class 28 are more threatened. Local conditions influence how the criminal law is applied. Communities vary in their expectations of law enforcement and the administration of justice, and its application is influenced by the visibility of offenses in a community and by the public?s norms about reporting possible violations. This theoretical framework is also coherent with group threat perspectives. In Blumer?s (1958) group prejudice perspective, the dominant group shares a feeling of apprehension that the subordinate group is threatening. Large increases in a dissimilar minority group over a relatively brief period will lead to social and economic conflict with dominant groups and the intensification of social control efforts to maintain the status quo (Caravelis et al. 2011; Jacobs et al. 2005; Skitka et al. 2006; Steffensmeier & Demuth 2001; Wang 2012). Cross-national differences in incarceration rates are significantly explained as a function of internal racial or ethnic threats (Jacobs & Kleban 2003). Similarly, realistic and symbolic threats and inter-group anxiety are significant psychological factors explaining attitudes toward immigrants (Stephan et al. 1999). Indeed, the perceived size more so than the actual size of the undocumented immigrant population is a particularly salient predictor of perceptions of this group as a criminal threat (Wang 2012). Moreover, Johnson and colleagues (2011) found that both perceived and objective ethnic threats are significant predictors of public support for use of ethnicity in punishment. The differentiated magnitude of recent Hispanic immigration trends across jurisdictions may also explain the contextually differentiated impact of their hypothesized threatening role. 29 This theoretical framework allows us to formulate the hypotheses that constitute the core argument of the present paper. First, the review conducted in the introductory part reveals that the nationwide immigration policy approach underwent in the last two decades emphasizes the extended use of deportation as a means of social control for convicted non-citizens. Nonetheless, this emphasis is likely to vary across jurisdictions not only because the influx and presence of Hispanic immigrants is different by location, but also because local policymakers have adopted more or less severe immigration enforcement approaches. These policy choices are also likely to be a function of the local public?s hegemonic opinions and attitudes toward Hispanic immigrants and enforcement, as well as the local political standing of Hispanics. This contextually differentiated emphasis on the deportation of Hispanic immigrants is likely to bring about state variations in the punishment of Hispanic non-citizens. Therefore, these standpoints lead to the following: Hypothesis 2: Sentencing severity for Hispanic non-citizens will vary significantly across states. Hypothesis 3: Hispanic non-citizens will receive shorter sentences in states that place more emphasis on deporting non-citizens than in those states that place less emphasis on it. Hypothesis 4: Hispanic non-citizens will receive shorter sentences in states with greater growth in their Hispanic populations. In summary, this paper is guided by contemporary sentencing and judicial decision-making theory, and proposes an analysis that includes the contextually differentiated magnitude of the Hispanic immigration and the emphasis on 30 immigration enforcement as fundamental factors to explain sentencing outcomes for non-citizen defendants. This research project purports to demonstrate that Hispanic non-citizen defendant?s sentencing is significantly influenced by immigration determinants. This influence is expected to vary across jurisdictions and be conditioned by contextual variables. 31 Chapter 4: Data and Methods Data This study uses secondary data from the Monitoring of Federal Criminal Sentences Series database of the USSC maintained by Inter-University Consortium for Political and Social Research (ICPSR 25424). The subjects in this research are criminal defendants sentenced pursuant to provisions of the Sentencing Reform Act (SRA) of 1984 and reported to the USSC during FY 2008. Only the cases in which a presentence investigation report are included because the USSC recommends analyzing only cases where this report was completed due to missing information on key variables (Hartley & Tillyer 2012). Capital punishment cases are also excluded because of their exceptionality. Given the nature of the proposed study and hypotheses, the sample is limited to Hispanic defendants only. With these constraints, the resulting sample consisted of 26,323 observations.7 However, since this study only considers those defendants sentenced to prison, the final sample consists of 25,927 subjects. These data were restricted to 50 states within the United States, and the District of Columbia.8 This dataset includes data about defendant?s citizenship and ethnicity, as well as court characteristics, sentencing 7 After limiting the data, the final sample consisted of 26,325 observations. Nonetheless, there were two observations with missing information on criminal history only that were listwise deleted. These defendants were two non-Black and male non-citizens, aged 31 and 34, with financial dependents, awarded no departure, without post-high school education, with only one count of conviction, one of them denied bail, not sentenced at trial, with 24 adjusted minimum months of incarceration recommended by sentencing guidelines, and finally sentenced for immigration and drugs offenses, and to 24 and 30 months of incarceration, respectively. 8 The level-two units for this study are states because the proposed contextual variables are measured at the state level. 32 outcomes, and criminal caseloads, making it an appropriate dataset for conducting research on sentencing disparities associated with Hispanic non-citizen offenders. Dependent Variable The dependent variable for the present study is sentence length. This is a continuous variable measured as the natural logarithm of the number of months of incarceration that the offender was sentenced to serve in prison, because according to prior research the distribution of sentence length is highly skewed (Hartley & Tillyer 2012; Iles 2009; Shermer & Johnson 2010). However, since the distribution of the logged dependent variable was still moderately skewed to the left, all the zeros were set to .9. The distributions of the original and logged dependent variable are displayed below in Figures 1 and 2. On the other hand, sentence length models only comprise those cases that received an incarceration sentence, which has led researchers to include a selection bias correction factor (e.g., Ulmer & Johnson 2004; Johnson 2006). For the present study, a Heckman hazard ratio is included in all the purported analyses and models. This new variable was calculated using the Heckman two-step command in Stata 12, and then included in each model.9 [Insert Figure 1 about here] [Insert Figure 2 about here] 9 The accuracy of this method and its likelihood to produce high collinearity has been questioned (Kautt 2002). In this regard, the inclusion of this factor was previously tested by creating the Heckman correction by using the mentioned procedure in Stata 12, and then comparing models with and without it. This comparison was done by examining the condition number calculated using a command in Stata 12 (cndnmb3) as an indicator of severe collinearity issues, with a coefficient of 30 or more considered problematic. In this case, the impact of this correction factor on the condition number of the tested ordinary least squares model was low. In fact, the condition number without the correction factor was 11.2 and with the correction was 11.3. Nonetheless, it is important to note that the correction hazard had a very influential effect on the main individual independent variable, which is citizenship status. In the models that included this correction, citizenship status was always statistically significant, but without it, in all the models this estimate had no statistical significance in many cases. 33 Independent Variables The main independent variable at the individual level is citizenship status, which is measured with a dummy variable indicating whether a defendant is a non- U.S. citizen. The third hypothesis is referred to the magnitude of the state?s contextual emphasis on deportation and is tested with two independent variables. First, a dummy variable measures whether a state has adopted a fast-track sentencing program and/or signed a 287(g) agreement. The list of states that have adopted these programs is included below in Table 1. This information comes from the Attorney Generals? memo of early disposition programs re-authorization (Department of Justice 2008), and the U.S. Immigration and Customs Enforcement website.10 [Insert Table 1 about here] Second, public?s attitudes toward deportation of illegal immigrants are measured at the state level with a proportion score elaborated using the CBS News/New York Times Monthly Poll, May 2007 (ICPSR 23444). This dataset consists of a random digit-dialing telephone interview survey conducted between May 18, and May 23, 2007, of a stratified representative sample of 1,132 respondents, with African Americans being oversampled (17%). After limiting the sample to those cases in which the information was complete for the variable of interest, the final simple consisted of 1,037 subjects. This variable was constructed from the following question asked in that survey: ?Should illegal immigrants be prosecuted and deported for being in the U.S. illegally, or shouldn?t they?? Then, a percentage coefficient was 10 U.S. Immigration and Customs Enforcement: http://www.ice.gov/news/library/factsheets/287g.htm #moa. 34 assigned to each state by calculating the proportion of respondents that favored deportation compared to those in each state that did not.11 Three variables are used to test the group threat theory (hypothesis four): immigration caseload pressure, Hispanic population, and Hispanic political representation. First, immigration offenses caseload pressure is measured as the average number of immigration cases sentenced in a state12 in a year divided by the number of authorized judgeships, and then divided by 100 to make a more meaningful interpretation of the results. Each one unit change in immigration caseload therefore represents an increase of 100 immigration cases per judge in the state. Data on authorized judgeships come from the United States Courts website,13 and the number of cases comes from the main sentencing dataset. Second, the estimated proportion of unauthorized immigrants as of 2007 is included. The Pew Hispanic Center (2011) provides state-level data of the estimated unauthorized immigrant population, and the U.S. Census Bureau provides data of the 2007 total population estimates (2007). This variable is measured as the proportion of unauthorized immigrants compared to the total population in each state. The estimates were calculated by the Pew Hispanic Center using the ?residual method,? by which a demographic estimate of the legal foreign-born-population is subtracted from the total foreign born population, and the remainder is the source of population estimates of unauthorized immigrants. The official source of data used to calculate these estimates 11 Even though the representativeness of the responses for most states was very low, it was possible to assign a coefficient to each one since all of them had at least two respondents, with an average of 20 respondents per state. States with five or less respondents were the following: Alaska, Delaware, District of Columbia, Hawaii, Nebraska, North Dakota, South Dakota, Vermont, and Wyoming. 12 This variable is also measured at the state level, given that all the other contextual variables are measured at this same level. 13 United States Courts:http://www.uscourts.gov/JudgesAndJudgeships/AuthorizedJudgeships/Chrono- logicalHistoryOfAuthorizedJudgeshipsIndex.aspx. 35 was the Current Population Survey. All 2007 estimates are based on the average share of the national unauthorized immigration population for 2006-2008. Finally, the Hispanic political representation is measured as the proportion of Hispanic legislators in each state Congress. This variable corresponds also to hypothesis four because a larger number of Hispanic legislators is partially indicative of a larger presence of Hispanics in the state. Though, it may be considered also a moderating factor because a larger Hispanic political representation may also be interpreted as indicative of more cohesive and settled Hispanic communities in that location. This information is obtained from the National Conference of State Legislatures web page for year 2009.14 Control Variables Based on prior research on sentencing disparities and contextual variation in punishment at the federal level (Johnson et al. 2008; Kautt 2002; Shermer & Johnson 2010; Ulmer et al. 2010), this study includes individual legal and extra-legal control variables. Gender is a dummy variable that codes males with one and females with zero. Age is measured as a continuous variable for defendant?s age in years at the time of sentencing. Education is measured as a dummy variable indicating whether a defendant has some post-high school education or not, with high school education or less, coded as the reference category with a zero. A dummy variable, coded one if the offender had any financial dependents and zero if otherwise, is also included. Race is 14 http://www.ncsl.org/legislatures-elections/legisdata/latino-legislators-overview.aspx. The last available and completely reported data on the proportion of Hispanic legislators per state congress is of year 2009. However, these data fairly represents the Hispanic political representation for year 2008 because they refer to legislators elected in that year and on duty throughout 2009. Moreover, only the states of New Jersey and Virginia held legislative elections in 2009, which took place in November of that year and were effective the following legislative period that began in 2010. 36 measured with a dummy variable that indicates whether a defendant is African American or not, with the former coded one and the latter, zero. The type of crime is measured with a series of dummy variables for immigration,15 others,16 and drug offenses,17 with the latter as the reference category.18 This categorization is justified because 51% of the data correspond to immigration offenses, and 35% to drug offenses, and in the case of non-citizens these categories comprise 63% and 28% of the sample, respectively. Downward and upward departures are also included as control variables, and are measured with four dummy indicators, with no departures as the reference category. Upward departures include all the categories that fall above the guidelines range. Downward departures are classified into regular downward departures,19 substantial assistance departures,20 and early disposition departures.21 15 Immigration offenses comprise two broad categories of crimes: smuggling, transporting, or harboring an unlawful alien, and unlawfully entering or remaining in the United States. Unfortunately, federal sentencing data do not provide this information disaggregated into these two categories. 16 Even though the type of offenses included within the category of others may not be theoretically coherent, the very few cases corresponding to each one as well as the main focus of the present study makes it preferable to proceed in this way. This category includes the following offenses: murder, manslaughter, kidnaping, sexual abuse, assault, bank robbery, arson, burglary, auto theft, larceny, fraud, embezzlement, forgery, bribery, tax offenses, money laundering, racketeering, gambling, civil rights offenses, pornography prostitution, offenses in prisons, crimes against the administration of justice, environmental offenses, national defense offenses, antitrust violations, food and drug offenses, traffic violations and other offenses. 17 Drug offenses include, according to the federal sentencing dataset codebook, trafficking, communication facilities, and simple possession. 18 This decision was mainly due to the high negative collinearity (-.77) between immigration and drug offenses, which increased the likelihood of committing a Type II error in the case of the immigration offenses estimate. 19 Regular downward departures are regulated under the Federal rule 5K2. In these cases, the judge retains all the discretion to decide whether or not a downward departure could be granted in the presence of some extenuating circumstances. This category also included for this study all the other types of downward departures not included in the remaining categories. 20 Substantial assistance departures are regulated under Federal Rule 5K1.1 and is granted to defendants considered to have provided substantial assistance to law enforcement. In the first place, these departures require a formal motion from the prosecutor, but in the end is the judge who determines whether or not they should be conceded. 21 Early disposition departures are explained in detail previously on page eight of this thesis. 37 A dummy variable is also included to measure whether the case involved multiple counts of conviction or not, with the latter as the reference category. Pre- sentence detention is also measured with a dummy variable coded one if the defendant was detained and zero otherwise. Mode of conviction is measured with a dummy variable indicating whether the offender was convicted through trial or guilty plea, with the latter as the reference category. Regarding criminal history, according to Engen and Gainey (2000), sentencing research in jurisdictions where a guidelines system has been implemented should include the presumptive guideline sentence because the relation between crime seriousness, criminal history, and sentence length is not linear. The presumptive sentence is equal to the minimum months of incarceration recommended by the sentencing guidelines capped at 470,22 although the defendant?s criminal history is also included because prior research suggests that it carries additional weight beyond the presumptive sentence (Johnson et al. 2008). This variable is measured as the guideline criminal history score, ranging from 0 to 6 (6, the most serious). There is not a high correlation between these two variables (r = .12). A thorough coding scheme summary of the dependent and independent variables is included in Table 2. [Insert Table 2 about here] Analytical Approach The analytical approach to be developed relies on the procedures employed in similar previous studies (e.g., Hartley & Tillyer 2012; Johnson et al. 2008; Johnson & 22 It is to be noted that the presumptive sentence was also logged in order to ensure normality in the error terms and keep consistency with the dependent variable. As in the case of sentence length, this variable was also slightly skewed to the left, so the zeros were also set to .9. 38 Betsinger 2009; Kautt 2002). The approach that best fits the theoretical basis and predictions of the present study relies on the use of two-level hierarchical linear models. The main focus of this paper is not exclusively on the impact of individual- level variables on sentencing within nested data. In these cases, the recommended approach is to control for inter-district variation by including a set of dummy variables for each district (e.g., Johnson & Betsinger 2009). The theoretical predictions that guide the present study, however, mostly refer to cross-level interactions that require not only removing interclass correlation due to group dependency, but including variables at the state level to explain between-group variations. In this case, the appropriate and recommended technique is hierarchical linear models (Johnson 2010; Kautt 2002; Ulmer & Johnson 2004). Unlike ordinary least square models, multilevel models allow not only to account for statistical dependencies between data clusters, but also to include information from different levels of analysis (Johnson 2010). This avoids the violation of essential assumptions of linear regression, such as homoscedasticity and independence of error terms, and allows for the interaction between contextual and individual variables. The analysis takes the following steps. First, hypothesis number one is tested using an ordinary least squares model. For this particular part of the analysis, district variation is controlled for by including a series of dummy variables for each district. This analysis comprehends three progressive models.23 The first includes only citizenship status, the second adds legal control variables, and finally, the third includes all the remaining variables. 23 All of these models include the selection hazard rate and the district controls. 39 Afterwards, the basic assumption of hierarchical linear models that exist there correlated errors between clusters is tested, in this case between states. Although several studies have shown that this is the case for federal sentencing data (e.g., Johnson & Betsinger 2009; Kautt 2002), the second hypothesis is tested by estimating a two-level unconditional model and calculating the interclass correlation coefficient to investigate the amount of variance explained by between-group differences (Johnson 2010). This procedure will allow for estimating the relative amount of sentencing variation that occurs both at the individual and state levels of analysis. After doing this, all the variables are grand-mean centered and added to the following models. Subsequently, a random intercept model that includes only individual-level variables is run, then a random intercept model that also includes the state-level contextual variables, and then a random coefficient model with specific random effects. Finally, this study relies on cross-model interaction models to test the main hypotheses of the present study. Nonetheless, preliminary analyses are conducted in order to test for misspecification, determine possible collinearity problems, and pursue the most robust and stable estimators (Johnson 2010).24 24 Effect sizes were calculated for each model. However, none of these effects were different than zero except for the presumptive sentence. Thus, given that they did not provide any meaningful additional information, they are not included in the finally reported results. 40 Chapter 5: Findings Descriptive Statistics From observing Tables 3 and 4 included below, it can be seen that among the full sample, 77% are Hispanic non-citizens. Interestingly, 96% of them were detained prior to trial, in contrast to only 73% among Hispanic citizens. This seems to provide descriptive evidence that the perceived flight risk is more heavily weighted by federal judges in the case of non-citizens. In addition, both Hispanic citizens and non-citizens are at first sight similarly situated in respect to guidelines departures because in both cases approximately 60% of them were sentenced within the guidelines. However, a closer examination reveals noteworthy differences. First, regarding regular downward departures, 19% of citizens received them compared to 15% of non-citizens. Concerning substantial assistance and early disposition departures the differences are more germane. For instance, 18% of citizens received substantial assistance departures compared to only 6% of non-citizens. In the case of early disposition departures, 19% of non-citizens were awarded this type of departure compared to only 4% of citizens. These differences may not only be attributable to sentencing features linked to citizenship status, but also due to contextual variations.25 [Insert Table 3 about here] [Insert Table 4 about here] A very significant aspect that emerges from the descriptive statistics is the prevalence of the type of offense by citizenship status. For instance, Hispanic citizens comprise 13% of the immigration offenses, compared to 64% in the case of Hispanic 25 It is to be noted that early disposition departures are an outcome that corresponds mostly to the state and sentencing district of Arizona (44%), followed by Texas South (26%). 41 non-citizens. Conversely, drug offenses corresponded to 60% of the former group, while only 28% of the latter. This may to a great extent explain why citizens received on average lengthier sentences (61 months) than non-citizens (35 months), since immigration offenses receive shorter sentence lengths than drug crimes. In reference to the contextual variables, it is to be noted that more than half of the states (53%) have signed either one or both of fast-track sentencing or 287(g) programs. In the case of attitudes toward deportation, there is a moderate prevalence favoring it (.65 in a 0 to 1 scale). Regarding immigration caseload pressure, the most significant datum is that there is relatively large variation and skewedness to the right, which means that a few number of states share the huge majority of immigration offenses. For instance, only one standard deviation below the mean falls below zero, whereas adding three standard deviations to the mean (1.77) is still far below the maximum value (2.91). In the case of unauthorized immigrants, it is noteworthy that the percentage of unauthorized immigrants is relatively low, with an average of 3% and a maximum of just 9%. The variability in percentage terms is also not that large, with a standard deviation of 2%. Finally, the presence of Hispanic legislators in each state Congress remains relatively low in the current days with a mean of just 4%, although it seems again that a few states comprise a huge presence of Hispanic legislators, given that there is a maximum of 44% and a standard deviation larger than the mean (8%). As part of the descriptive analyses, correlations among the independent variables were also examined. The purpose of this procedure was to determine whether there is high collinearity between one or more variables. The main problem 42 with high collinearity is the occurrence of a Type II error due to excessively large standard errors. Table 5 displays a correlation table for the individual level variables. None of them seems to be indicative of multicollinearity issues, with moderate correlations for immigration offenses and citizenship status (.43) and the presumptive sentence (.44). After running preliminary models, no issues regarding multicollinearity were evident from the results.26 [Insert Table 5 about here] Furthermore, Table 6 displays a correlation table only for the level two contextual variables. In this case, there are noticeable cases of severe multicollinearity issues. There is a huge correlation between Hispanic political representation and immigration caseload pressure (.86), and estimated unauthorized immigrants (.53), as well as between fast-track/287(g) programs and estimated unauthorized immigrants (.57). As mentioned above, the main problem of a high collinearity is the likelihood of committing a Type II error. Several preliminary models were run with each of these variables separately, and then with a progressive combination of all of them. These diagnostics also included likelihood-ratio tests to determine the specific contribution of each variable to the model. The only variable that was significant in all the models and tests was Hispanic political representation. Immigration caseload pressure was the other statistically significant variable in the likelihood-ratio tests, but not in the models in which Hispanic political representation was included. Moreover, the inclusion of that variable in the more complex models produced instability and model fitting problems. On the other hand, including all the 26 Additional multicollinearity diagnostics using variance inflation factors were conducted and none of the variables showed problematic issues of multicollinearity. 43 other variables did not lead to substantial modifications in the effect and significance of these and the other individual-level variables, and did not affect the accuracy and stability of the models. Therefore, all of contextual variables except immigration caseload pressure are included in the definitive reported models. [Insert Table 6 about here] Results from Ordinary Least Squares Regressions The ordinary least square regressions in this study are used to test hypothesis one.27 As mentioned above, these analyses were conducted progressively. Results from these models are shown in Table 7, and they seem to provide apparent support for that hypothesis. From the first model that only includes citizenship status without any controls, it emerges that Hispanic non-citizens were likely to receive on average sentences that were 88% shorter than Hispanic citizens, with this result being statistically significant. It should be noted that this model only explained 29% of the variation in the dependent variable, which in comparison to the following more complete models is notoriously low.28 [Insert Table 7 about here] After including legal control variables, the effect of citizenship status dramatically decreased, even though it retained statistical significance at the .001 level. According to this model, being a Hispanic non-citizen was associated with sentence lengths that were on average 6% shorter compared to Hispanic citizens. It is 27 District 81 (Oklahoma East) was used as the reference category. Moreover, to comply with the homoscedasticity assumption, robust standard errors were specified. Finally, although the relation between sentence length and age was significantly curvilinear, in favor of parsimony and consistency across models, and given that this is only a control variable, the squared term for age is omitted from the finally reported models. 28 A model that did not include the selection hazard rate and the district controls explained only 5% of the variance in the dependent variable. 44 noteworthy from this model that individual legal variables had a significant effect on sentence length. Particularly, a 1% increase in the presumptive sentence was associated with a .74% increase in sentence length. All of the downward and upward departures had also noticeable and significant effects on sentence length, with defendants who received regular, substantial assistance, and early disposition downward departures being sentenced on average to 41%, 58% and 58%, respectively, shorter sentences than those who did not receive these benefits. Those who received upward departures were sentenced to 75% lengthier sentences, on average. This model explained 86% of the variance in the dependent variable. Finally, the comparison between the second and third models reveals that the apparently large effect of citizenship status on sentencing was mostly accounted for by the legal control variables because its effect did not experience any modification once the extra-legal variables were included. Still, that variable retained statistical significance at the .001 level. None of the other control variables suffered any important modification in their magnitude, directionality, and statistical significance. The last model explained 86% of the variation in the dependent variable. 29 Results from Two-Level Unconditional HLM Model Table 8 presents results from the two-level unconditional model of sentence length. The results, in particular the interclass correlation coefficient, indicate that about 9% of the total variance was attributable to between state variations. This means that the majority of the variation on sentence length was mainly due to 29 Model diagnostics revealed that there were three influential observations with high leverage over the results. Nonetheless, none of the coefficients suffered any substantial variation in their directionality, magnitude, and significance once these observations were removed. Thus, they were kept in all the ultimately reported models. 45 individual level factors, which is also reflected in the magnitude of the variance coefficients for each level of analysis. However, the size of this variation is similar and even larger than what has been shown in previous studies of federal sentencing. For instance, Kautt (2002) found that level two (district) and level three (circuit) factors explained 7% and 3% of the variance, respectively. Also, Ulmer and colleagues (2010) found that 6% of the variance in sentence length existed between district courts. This means that the amount of variance attributable to the second level of analysis in this study is consistent with what has been found in prior research and constitutes an adequate baseline for the subsequent stages of analysis. [Insert Table 8 about here] Results from the Two-Level Random Intercept Model: Individual Level Fixed Effects This model is the logical next step in the analysis because it essentially signifies the inclusion of individual level variables. Results from this model are depicted in Table 9. The inclusion of level-one variables explained approximately 85% of the total variance at the individual level, and also explained 92% of the variance between states.30 This means that most of the variation at the state level was explained by the impact of individual level variables. [Insert Table 9 about here] Looking at the fixed effects, the results are practically similar to those of the ordinary least square model, with the legal individual level variables having the more meaningful effects on sentence length. The presumptive recommended sentence, criminal history, and the guidelines departures played the major roles in explaining 30 These are Bryk/Raudenbush R-squared values calculated by Stata. 46 sentence length for Hispanic defendants. Regarding citizenship status, Hispanic non- citizens received on average shorter (6%) sentences than Hispanic citizens. Importantly, this effect was statistically significant at the .01 level. Results from the Two-Level Random Intercept Model: Contextual Level Fixed Effects The next step in the analysis requires the inclusion of level-two variables. Results from this model are presented in Table 10. The inclusion of these four new variables explained an additional 2% in the variation at the state level (94%),31 although including these variables did not produce any modification in the direction, magnitude and significance of any of the individual level variables, counting citizenship status. Among the four contextual variables, only Hispanic political representation showed a statistically significant negative impact on sentence length. Even though the magnitude of the coefficient may be considered low, it exerted a meaningful effect on the dependent variable, with a one unit (%) increase in the percentage of Hispanic legislators in a state congress associated with a 1% reduction in sentence length.32 [Insert Table 10 about here] 31 This is a Bryk/Raudenbush R-squared value calculated by Stata. 32 In order to test the existence of influential level two outliers, model diagnostics were conducted. From these analyses it emerged that the states of Texas, Arizona, and New York seemed to exert great influence on the results. Alternative analyses were conducted with and without these states and the only somewhat noteworthy modification was that the statistical significance of citizenship status was downgraded from a .001 to a .05 level in the model without Texas, which seems to be a minor change. Moreover, the directionality of the effect was the same, and the magnitude experienced a small reduction (from -.06 to -.04). Therefore, the finally reported models include all the level-2 units. However, a caveat that should be mentioned is the relative influence of the selection bias hazard rate in the results of Texas. Ordinary least squares regressions with and without the correction factor were run only for this state and in the first case being a non-citizen had a statistical significant (p = .000) negative effect on sentence length, whereas in the latter case this significance was completely lost (p = .841). This is likely to be linked to the fact that in the study sample the great majority (40%) of those placed on probation corresponded to Texas. All of the remaining states had far lower percentages. 47 Results from the Two-Level Full Random Coefficient Model This can be considered the final full model with random effects, prior to the examination of cross-level interactions. Results from this model are depicted in Table 11. Even though the main focus of this paper is on the fixed effects of the model, a brief commentary about the selection of the random effects could be relevant. In the specification of a multilevel model it is recommended to use an unstructured covariance matrix because it places no constrains in the covariances between the effects themselves and the constant. First, the hypotheses of this paper merited the inclusion of citizenship status as a random effects. Likewise, this effect was significant in the likelihood-ratio tests. Second, the tests revealed that most of the individual random effects were statistically significant. In this case, the only random effects of theoretical importance that in conjunction with citizenship status produced a stable, feasible and more parsimonious model were the presumptive sentence, substantial assistance departure, and criminal history. Therefore, these were the random effects that were ultimately included in the following models. [Insert Table 11 about here] Examining more in-depth on these variations, it is interesting to consider both the fixed coefficients and the random effect variances. In particular reference to the Hispanic defendant being non-citizen, the interpretation of the random effects shows that for approximately two thirds of the states, non-citizens were sentenced to incarceration terms that were 2% lengthier and 13% shorter than citizens.33 This reveals two noteworthy facts about how Hispanic non-citizens are sentenced across 33 These values were obtained by adding and subtracting from the corresponding fixed effect coefficient the standard deviation of the respective random effect, and then obtaining the antilog of the calculated value. 48 states. First, the significance and magnitude of the random effects reveal that the effect of being non-citizen does vary across states. Second, the directionality of this effect also varies, with some states sentencing non-citizens to lengthier sentences, and others meting out shorter incarceration terms. Regarding the fixed effects, allowing for the slopes to vary across states in the mentioned variables led to very slight or imperceptible changes in all the estimates. In relation to citizenship status, being a non-citizen remained statistically significant and with the same directionality and magnitude in its coefficient effect. Concerning the contextual variables, it is noteworthy how the estimated percentage of unauthorized immigrants reached statistical significance once the random effects were included in the model. Interpreting the results for this estimate, a 1% increase in the number of unauthorized immigrants in a state was associated with approximately a .01% increase in sentence length. On the other hand, the estimate of Hispanic political representation retained statistical significance in this model and produced essentially the same results as in the previous one. None of the other contextual variables experienced any substantial modification. Results from Cross-Level Interaction Models Results from these models are reported in Table 12.34 In the first place, it is noteworthy that among the cross-level interactions the only one that was marginally significant was the interaction between citizenship status and Hispanic political representation. A cautious interpretation is that the data do not support the existence of a non-citizen specific effect in interaction with most of the selected contextual 34 These interactive models were performed separately for each interaction effect and included all the control variables and random effects specified in the previous models. 49 variables, and that a very modest effect may have existed in the case of Hispanic political representation. In addition, only in the case of attitudes toward deportation and Hispanic political representation the interaction estimates had the expected negative directionality. The finding for Hispanic political representation is probably the most noteworthy in the present study. It is interesting to note how it had a negative and significant effect for Hispanic citizens, but also negative?although only marginally significant?effect for Hispanic non-citizens. Nonetheless, the magnitude of the effect for the interactive effect by citizenship status is extremely low. Given these results, it is reasonable to say that it is likely that the effect of Hispanic political representation is negative and significant for the full sample, but that the evidence of a specific citizenship status effect is modest at best. It is apparent from these and the previous models that two of the proposed variables had an effect for both Hispanic citizens and non-citizens. Although the magnitude of the effects are slight, there is some evidence that an increase in the percent of unauthorized immigrants in a state is associated with small but significant increments in sentence length for Hispanic defendants. In addition, an increase in the percentage of Hispanic legislators in a state congress is associated with a minor but significant reduction in sentence length for both types of Hispanic defendants. [Insert Table 12 about here] 50 Chapter 6: Discussion Among the broader issues that surround the demand for immigration reform is the struggle of Hispanics for obtaining the U.S. citizenship. Probably seen as an avenue for fulfilling legitimate ambitions of a better future for them and their families, becoming a citizen is likely to be a life goal for a large number of Hispanic residents. In this context, the impact of the citizenship status on sentencing could have become a significant factor for explaining sentencing disparities among Hispanic defendants. One of the possible explanations to support this argument is that under the current conditions established by restrictive legislation and policy practices, deportation is almost an unavoidable consequence for non-citizens convicted of a federal crime. That is the core theoretical component of the so-called crimmigration law phenomenon. In essence, it is argued that the criminal and immigration laws are interchangeably and instrumentally used to formally legitimate and practically operationalize the prosecution and removal of non-citizens. This standpoint relies in the idea that immigration is perceived as threatening by the mainstream society. Under this reasoning, the role of the federal criminal justice system is consequential. As a republican institution, it works as a formal and legitimate mechanism of social control, although contemporary theoretical perspectives purport that substantial societal realities and bargaining practices play a decisive role in the decision-making processes of the criminal justice system (Savelsberg 1992; Sklansksy 2012). In this regard, the federal criminal justice system may be working as a mechanism for targeting, prosecuting and subsequently facilitating the deportation of non-citizens. Given the large share of Hispanics in the population of 51 the United States and their increasing presence in the federal criminal justice system during the last two decades, this phenomenon is likely to have a greater impact on individuals and communities of this ethnic group. The focus of this paper is precisely on how the implications of the so-called crimmigration law affect Hispanic defendants. For that purpose, this study proposed the innovative methodology of examining contextual variations in the punishment of Hispanic non-citizens across states, hypothesizing that those disparities may be related to the contextually differentiated impact of the more recent Hispanic immigration and its enforcement. This study started by examining whether or not Hispanic non-citizens receive shorter sentences than Hispanic citizens, with the idea that the criminal justice system may expedite their removal by imposing shorter incarceration terms. After controlling for legal and extralegal factors that previous research has shown to have some impact on sentencing outcomes, it emerged that Hispanic non-citizens do receive sentences that are on average approximately 6% shorter than those imposed on Hispanic citizens. This is consistent with what has been shown by the more recent research (Shermer & Johnson 2010; Spohn 2005; Wu & Spohn 2010), but goes beyond them because it reveals that the difference between citizens and non-citizens seems to be more consequential among Hispanics. In summary, this finding supports the notion that among Hispanics, non-citizens share one or more characteristics that influence judges to mete out shorter sentences. As already mentioned, this study purports that the more relevant factor for explaining that disparity is the practically unavoidable deportation that non-citizen offenders face. Unfortunately, as more thoroughly discussed below, the available data 52 do not provide the kind of information necessary to directly test this assumption. For that reason, the present work proposed to examine whether this link could be to some extent explained by the differentiated magnitude of the Hispanic immigration across states, their political positioning, and the enforcement of the immigration laws. The first logical step in the mentioned strategy was to determine whether or not there is a significant variation in punishment across states. After running an unconditional model, this assumption was supported by the data and concordant with what has been found in previous research. Although most of the variation in the dependent variable was explained by individual level legal factors, the theoretical and practical importance of the context should not be neglected. The random effects included in the analyses supported the assertion that there is variability across states in how Hispanic non-citizens are judged. Approximately two thirds of the states imposed on non-citizens sentences that were between 13% shorter and 2% lengthier than citizens. The next step was to explain that variation using contextual variables. The underlying theoretical argument of this work relies on the perception that Hispanic immigrants are threatening. The traditional conflict theories and the contemporary focal concerns perspective support this standpoint. Though, the expected behavior of the criminal justice system regarding Hispanic non-citizens was not harsher punishments but the opposite. This apparent counterintuitive proposition is on the contrary well-grounded on the theoretical presuppositions of this thesis. Since deportation may be considered not just a collateral consequence of punishment but a direct one for non-citizens, it is hypothesized that non-citizens receive shorter sentences in those locations in which they could be perceived as more threatening. 53 To test this assumption, this paper proposed five contextual variables to measure these realities. From the side of the increasing presence of Hispanics in the U.S. territory, this study used the Hispanic political representation, the Hispanic population, and the immigration caseload pressure per state. From the side of the enforcement, the analyses included whether or not a state signed a program intended to accelerate the removal of non-citizens, and the attitudes of their habitants in favoring or not the deportation of unauthorized immigrants. Preliminary analyses demonstrated that immigration caseload pressure was a variable that caused problems in the specification and stability of the models. For that reason, only the other four contextual variables were included in the ultimately reported models. The findings regarding these variables only partially supported some of the main notions and propositions of this paper. First, only Hispanic population and Hispanic political representation had a statistically significant impact on sentence length for the entire sample. Surprisingly, the directionality of the estimates of each of these two variables were contradictory. Indeed, a 1% people increase in the unauthorized immigrant population was associated with somewhat lengthier sentences, while 1% increase of Hispanic legislators in a state congress was associated with slightly shorter sentences. Even so, no evidence emerged that these effects were different for Hispanic citizens compared to Hispanic non-citizens in the case of the estimates of unauthorized immigrants, and not conclusive in the case of Hispanic political representation. In fact, the inclusion of interaction terms intended to find a specific citizenship status effect only produced a marginally significant effect for Hispanic political representation, although extremely low in magnitude. 54 Therefore, these contextual variables only partially contributed to explaining the disparity in punishment between citizens and non-citizens among Hispanic defendants across locations. However, they did contribute to explaining the contextual variation that existed across states in the punishment of Hispanic defendants. These are relevant findings that lends credence to the hypothesis that the immigration threat perspective as measured in this paper has significant effects on how the federal criminal justice system is treating Hispanic defendants. 55 Chapter 7: Conclusions and Limitations In spite of the innovations and relevant findings presented in this study, there is a number of limitations. First of all, the ideal approach to determine whether deportation plays a relevant role on sentencing would be to directly use deportability status restricted to non-citizens as a key independent variable. Nonetheless, the dataset does not provide the kind of information necessary to conduct such an analysis. Only in very few downward departure decisions do judges explicitly address deportation as a main or secondary reason for departing. This seems to confirm Chin?s (2011) assertion that judges are more likely to sentence non-citizens within but at the lower end of the guidelines grid. Consequently, whether or not deportation influences the decision of federal sentencing judges is something not entirely possible to disentangle. However, this limitation was taken as a challenge to be overcome by offering a novel approach to examining more thoroughly the contextual differences that exist across states in sentencing outcomes for Hispanic citizens and non-citizens. Another limitation that merits discussion relies on the choice and validity of the variables that were used to measure the contextually differentiated presence of Hispanic immigrants and its enforcement across states. No major problems are apparent regarding the Hispanic population and the percentage of Hispanic congress legislators in each state. However, concerning immigration enforcement and deportation, it would have been ideal to include data on deportation proceedings by district or state. This could have included data about rates of deported immigrants, efficiency in immigration proceedings, and waiver decisions. Unfortunately, the organizational structure of the immigration enforcement system precluded such an 56 analysis. The problem refers to the fact that the immigration courts are located in a limited number of states and they have competence to judge non-citizens coming from different locations. Thus, it is not really viable to determine from which state a non-citizen comes. Using the available data from the 30 immigration jurisdictions could not only reduce statistical power, but also bias the estimates because they would not account for the true state emphasis on immigration enforcement. It is also necessary to point out that the measure of attitudes toward deportation suffers from some specific limitations that may affect the reliability of the variable estimate. First of all, it was based only on one question that assessed public opinion about whether unauthorized immigrants should be prosecuted and deported. The number of valid responses is low (n = 1,037) if the representativeness of the opinions are considered. There are states in which the number of respondents is less than five, with many of them having all the respondents favoring one of the answers. Therefore, this variable should not be seen as a true estimate of the intended parameter. Future research may include a more consistently measured estimate given that public?s attitudes toward deportation and immigration is a relevant and interesting factor to be considered by scholars on the sentencing research arena. Finally, one issue with the dataset is that the number of non-citizens is very small in many states. Though, this aspect did not represent a major obstacle for the present study as proposed. Nonetheless, prior research has shown that differences in terms of residency status (legal or illegal foreign resident) and country of origin have some significant effects on sentencing outcomes under certain circumstances. Thus, an analysis that considers these elements could shed more light on the core issues 57 discussed and analyzed in this work. On the whole, taking into account that the main focus of this paper was on non-citizens in general, this was not a truly concern here. This paper attempted to demonstrate that there is contextual variation in the punishment of non-citizens among Hispanic defendants, and explain it using variables developed from an immigrant threat perspective. In that purpose, this study provides evidence in favor of the existence of that variation. Nonetheless, the proposed contextual variables did not significantly explained the contextual disparity in sentence length between Hispanic citizens and non-citizens. However, the percentage of Hispanic legislators in each state congress and the unauthorized immigrant population in each state had a significant effect on the punishment of Hispanic defendants regardless of their citizenship status. These are interesting findings that contribute to support the idea that broader immigration aspects have something do to with sentencing decisions for Hispanics. The context proves again to be a meaningful aspect to be considered by researchers. It is very likely that there are relevant differences in how Hispanic are sentenced across states, and this study sheds some light on explaining them. It is interesting to see how the political representation of Hispanics in state congresses is associated with the sentencing outcomes of this ethnic group. There could be several interpretations of the substance of this variable. On the one hand, the representation of Hispanics in state congresses may be a measure of the presence of Hispanics in that location. It is reasonable to assert that the number of congressmen is associated with the number of residents that supported and voted for them. In this regard, the larger the number of Hispanic voters in a state, the larger the proportion of 58 Hispanic legislators in that congress. However, the population component by itself is not enough to explaining the magnitude of political representation. Those residents should be vested the right to vote and have enough political influence to promote politicians to elective positions. This may explain why 44% of legislators in the congress of New Mexico were Hispanic, whereas only 23% in California, the state more largely populated by Hispanics. Therefore, the number of Hispanic legislators in a state congress is the visible aspect of a possible series of factors, which in this particular case are also associated with reduced sentenced for Hispanics. By the same line of reasoning, an increase in the Hispanic population was instead associated with lengthier sentences. This finding gives credence to the threat argument in the sense that larger Hispanic populations feeds stereotypes that are to some extent reflected in the procedures and outcomes of the criminal justice system. Even though the magnitude of the effect is low, the fact that an increase in the population of a particular ethnic group is associated with an effect on sentencing is a remarkable aspect that indicates how meaningful is to consider aspects that may be tied to more particular concerns of the individuals of a particular ethnic group. From this discussion it emerges that the political representation of Hispanics may play an attenuating role regarding the threat perceived by their increasing presence. This paper traces several avenues for future research. First, the variation that exists in the punishment of Hispanics is a topic that warrants more examination. This paper proposed to explain it from an immigration threat standpoint and found interesting results. Future research may attempt to explain that variation from other perspectives, which will enrich the discussion about the contextual variation in the 59 punishment of Hispanics. Nonetheless, the inquiries raised by the propositions and findings of this thesis also call for more research based on immigration threat and enforcement viewpoints. These discussion and analysis could be improved by including more detailed data about how sentencing judges consider the deportability status of a defendant. In this regard, the United States Sentencing Commission may play a major role by requesting the explicit reporting of these considerations. The Supreme Court resolution in Padilla v. Kentucky may be taken as an opportunity to introducing more transparency in the judicial proceedings of Hispanic non-citizens. On the same path, reasonable credence has been given to the idea that the consideration of the deportability status of a Hispanic defendant is more heavily weighted and discussed at the earlier stages of the criminal proceedings. It is hypothesized that the prosecutor charge defendants for milder offenses and pleads for shorter incarceration terms in order to accelerate the removal of non-citizen offenders. Consequently, incorporating data from the earlier stages of the criminal proceedings could be a fruitful source of information for more neatly and accurately measure the specific weight of deportability statuses on sentencing. Prospective research may collect this type of data, which will probably contribute to better understanding the effect of immigration enforcement on sentencing. Another valuable way to investigate more in depth the contextual variations in the punishment of Hispanics is conducting longitudinal studies. The effect of immigration considerations on sentencing and its probable contextually differentiated emphasis is likely to have changed over time. The progressively increasing presence of non-citizens in the criminal justice system is probably modifying the imaginary of 60 the federal judicial officers about their duties and expectations. The federal criminal justices system is each time more saturated with non-citizen offenders, most of whom are sentenced for immigration offenses. This noticeable change may be affecting not only the way in which non-citizens are judged, but also the way in which other types of offenses and offenders are sentenced. For instance, prior research has found that the magnitude of the caseload pressure, as well as its content has significant effects on sentencing (Johnson 2005). Prospective research may examine the sentencing disparities analyzed in this paper over time and also investigate the extent to which the increasing presence of non-citizens and the magnitude of the immigration offenses caseload has modified sentencing proceedings and outcomes. The main findings of this thesis may be interpreted in two ways. In the first place, it confirms what recent previous research has shown, in the sense that non- citizens receive on average milder sentences than citizens. In this regard, the specific contribution of this study refers to the comparison between citizens and non-citizens among Hispanics. The conclusion is that the citizenship status does have an effect on sentencing among Hispanics. On the other hand, this study laid the ground for a discussion about how the political representation of Hispanics plays a role on sentencing. Future research should also consider the relatively important impact that increasing the political representation of Hispanics may have on the fate of Hispanic offenders. The idea that a more active and consequential political involvement may contribute to reducing disparities in the treatment of Hispanics by the formal social control institutions is striking and may call the attention of policymakers and grassroots organizations. 61 Table 1. List of States with 287(g) and Fast-Track Programs States 287(g) Program Fast-track Program Alabama YES NO Alaska NO NO Arizona YES YES Arkansas YES NO California YES YES Colorado YES NO Connecticut YES NO Delaware NO NO Florida YES YES Georgia YES YES Hawaii NO NO Idaho NO YES Illinois NO NO Indiana NO NO Iowa NO NO Kansas NO YES Kentucky NO NO Louisiana NO NO Maine NO NO Maryland YES NO Massachusetts YES NO Michigan NO NO Minnesota NO NO Mississippi NO NO Missouri NO NO Montana NO NO Nebraska NO YES Nevada YES NO New Hampshire NO NO New Jersey YES NO New Mexico YES YES New York NO YES North Carolina YES NO North Dakota NO NO Ohio YES NO Oklahoma YES NO Oregon NO YES Pennsylvania NO NO Rhode Island NO NO South Carolina YES NO South Dakota NO NO Tennessee YES NO Texas YES YES Utah YES YES Vermont NO NO Virginia YES NO Washington NO YES West Virginia NO NO Wisconsin NO NO Wyoming NO NO 62 Table 2. Summary of Dependent and Independent Variables for USSC Federal Sentencing Data, FY2008 Variables Coding Scheme Description Dependent Variable Ln Sentence Length Log (months) Natural log of the total number of months of imprisonment (capped at 470). Individual-Level Independent Variables Non-citizen 1 = yes Dummy indicator for offenders who are not U.S. citizens. Ln Presumptive Sentence Log (months) Natural log of the total adjusted minimum months of incarceration recommended by sentencing guidelines (capped at 470). Criminal history USSC Scale United States Sentencing Commission?s scale rating prior criminal history from 1 to 6. Downward Departure 1 = yes Offender received 5K2 downward departure or any other downward departure. Substantial Assistance Departure 1 = yes Offender received 5K1.1 downward departure for substantial assistance to government. Early disposition departures 1 = yes Offender received 5K3.1 downward departure. Upward departures 1 = yes Offender received an upward departure. Multiple counts 1 = yes A dummy indicator for offenders convicted of multiple offenses. Age at sentencing Years Age of offender at time of sentencing. Male 1 = yes A dummy indicator for male offenders. Black 1 = Black Dummy indicator for black, with non-black (white) the omitted category. Financial dependent 1 = yes A dummy indicator for offenders with financial dependents. Post-high school education 1 = yes A dummy indicator for offenders with any post-high-school education. Presentence detainment 1 = yes A dummy indicator for offenders detained or released prior to trial. Trial conviction 1 = yes A dummy indicator for offenders convicted at bench or jury trial. Type of offense 2 dummy variables Dummy indicators for immigration and drug offenses, with other offenses as the reference category. State-Level Independent Variables Fast-track/287(g) programs 1 = yes Dummy indicator for states that have adopted fast-track programs and/or 287(g) programs. Attitudes toward deportation Score A proportion index score (0 to 1) that measures the attitudes toward deportation of immigrants in each state. Immigration caseload pressure Rate The number of immigration offenses cases divided by 100 and by the number of authorized judgeships per state. Estimated unauthorized immigrants Proportion Percentage of unauthorized immigrants per state. Hispanic political representation Proportion Percentage of Hispanic legislators in each state Congress. ABBREVIATION: Ln = Natural logarithm 63 Table 3. Descriptive Statistics for USSC Federal Sentencing Data, FY 2008 Mean SD Min. Max. Dependent Variable (N = 25,927) Sentence Length 41.10 48.13 0.90 470.00 LnSentence Length 3.17 1.10 -0.11 6.15 Individual-Level Variables (N = 25,927) Defendant is non-citizen 0.77 0.42 0.00 1.00 Presumptive Sentence 47.32 55.40 0.90 447.00 LnPresumptive Sentence 3.18 1.36 -0.11 6.10 Criminal history 2.41 1.58 1.00 6.00 No Departure 0.59 0.49 0.00 1.00 Downward Departures 0.16 0.36 0.00 1.00 Substantial Assistance Departure 0.09 0.28 0.00 1.00 Early Disposition Departure 0.15 0.36 0.00 1.00 Upward Departures 0.02 0.12 0.00 1.00 Multiple counts 0.12 0.32 0.00 1.00 Defendant's age 33.37 9.47 17.00 83.00 Defendant is male 0.92 0.27 0.00 1.00 Non-black 0.98 0.14 0.00 1.00 Black 0.02 0.14 0.00 1.00 Financial dependent 0.72 0.45 0.00 1.00 Post-high school education 0.08 0.27 0.00 1.00 Presentence detainment 0.91 0.29 0.00 1.00 Trial conviction 0.02 0.14 0.00 1.00 Immigration 0.36 0.48 0.00 1.00 Drug 0.13 0.33 0.00 1.00 Other crimes 0.77 0.42 0.00 1.00 Hazard rate 0.03 0.10 0.00 1.33 State-Level Variables (N = 51) Fast-track/287(g) programs 0.53 0.50 0.00 1.00 Favor deportation 0.65 0.21 0.00 1.00 Immigration caseload pressure 0.24 0.51 0.01 2.91 Estimated unauthorized immigrants 3.01 2.05 0.60 9.40 Hispanic political representation 3.55 7.51 0.00 44.00 ABBREVIATIONS: Ln = natural logarithm; SD = standard deviation. 64 Table 4. Descriptive Statistics by Citizenship Status for USSC Federal Sentencing Data, FY 2008 Citizens (N = 6,063) Non-citizens (N = 19,864) Mean SD Mean SD Dependent Variable Sentence Length 60.91 62.04 35.06 41.13 LnSentence Length 3.59 1.15 3.04 1.04 Individual-Level Variables Presumptive Sentence 72.81 70.78 39.53 47.09 LnPresumptive Sentence 3.81 1.06 2.98 1.38 Criminal history 2.43 1.70 2.40 1.55 No Departure 0.58 0.49 0.59 0.49 Downward Departures 0.19 0.39 0.15 0.36 Substantial Assistance Departure 0.18 0.38 0.06 0.23 Early Disposition Departure 0.04 0.19 0.19 0.39 Upward Departures 0.01 0.12 0.02 0.13 Multiple counts 0.18 0.39 0.09 0.29 Defendant's age 32.86 10.22 33.53 9.23 Defendant is male 0.83 0.38 0.95 0.23 Non-black 0.97 0.18 0.98 0.13 Black 0.03 0.18 0.02 0.13 Financial dependent 0.68 0.46 0.73 0.44 Post-high school education 0.15 0.36 0.06 0.24 Presentence detainment 0.73 0.44 0.96 0.19 Trial conviction 0.03 0.18 0.01 0.12 Immigration 0.13 0.33 0.64 0.48 Drug 0.60 0.49 0.28 0.45 Other crimes 0.27 0.44 0.08 0.28 ABBREVIATIONS: Ln = natural logarithm; SD = standard deviation. 65 Table 5. Correlation Matrix for Level 1 Individual-Level Predictors (ID) A B C D E F G H I J K L M N O P Q Non-citizen A 1.00 Presumptive Sentence B -0.25 1.00 Criminal history C -0.01 0.11 1.00 Upward Departures D 0.01 -0.04 0.01 1.00 Downward Departure E -0.05 0.12 0.02 -0.05 1.00 Substantial Assistance F -0.18 0.30 -0.10 -0.04 -0.13 1.00 Early Disposition G 0.18 -0.11 0.15 -0.05 -0.19 -0.13 1.00 Multiple counts H -0.12 0.29 -0.11 0.00 0.02 0.07 -0.13 1.00 Age I 0.03 0.09 0.06 0.01 0.06 0.01 -0.02 0.03 1.00 Male J 0.18 0.04 0.16 0.00 -0.04 -0.06 0.06 -0.02 -0.02 1.00 Black K -0.05 0.12 -0.01 -0.01 0.03 0.06 -0.06 0.07 0.03 0.02 1.00 Financial dependent L 0.04 0.01 -0.01 -0.01 -0.01 0.03 0.02 -0.02 0.09 0.00 0.01 1.00 Education M -0.14 0.07 -0.13 -0.01 0.04 0.09 -0.07 0.09 0.08 -0.07 0.03 -0.03 1.00 Presentence detainment N 0.34 0.02 0.16 0.02 -0.05 -0.13 0.10 -0.05 -0.05 0.17 -0.01 0.02 -0.14 1.00 Trial conviction O -0.07 0.21 -0.01 0.01 0.04 -0.04 -0.06 0.19 0.06 -0.01 0.02 -0.01 0.04 -0.02 1.00 Immigration P 0.43 -0.44 0.34 0.04 -0.05 -0.26 0.27 -0.26 0.02 0.11 -0.11 0.04 -0.18 0.19 -0.10 1.00 Others Q -0.24 0.02 -0.04 0.03 0.05 0.02 -0.15 0.18 0.02 -0.06 0.03 -0.04 0.12 -0.18 0.07 -0.39 1.00 N=25,927. Table 6. Correlation Matrix for Level 2 Contextual Predictors Fast-track/287(g) programs Attitudes toward deportation Immigration caseload pressure Est. unauthorized immigrants Hispanic political representation Fast-track/287(g) programs 1.00 - - - - Attitudes toward deportation 0.16 1.00 - - - Immigration caseload pressure 0.28 0.08 1.00 - - Estimated unauthorized immigrants 0.57 -0.18 0.36 1.00 - Hispanic political representation 0.34 -0.02 0.86 0.53 1.00 N=51. 66 Table 7. Models of Case-Level Factors for Sentence Length Individual-level factors Ln Sentence Length Model 1 Model 2 Model 3 b SE b SE b SE Constant 4.25 0.59 *** 1.40 0.24 *** 1.26 0.24 *** Defendant is non-citizen -0.88 0.01 *** -0.06 0.01 *** -0.06 0.01 *** Presumptive Sentence - - 0.74 0.00 *** 0.72 0.00 *** Criminal history - - 0.01 0.00 *** 0.02 0.00 *** Downward Departures - - -0.41 0.01 *** -0.40 0.01 *** Substantial Assistance Departure - - -0.58 0.01 *** -0.58 0.01 *** Early Disposition Departure - - -0.58 0.01 *** -0.57 0.01 *** Upward Departures - - 0.75 0.03 *** 0.75 0.03 *** Multiple counts - - - - 0.05 0.01 *** Defendant's age - - - - 0.00 0.00 *** Defendant is male - - - - 0.06 0.01 *** Black - - - - 0.12 0.02 *** Financial dependent - - - - 0.00 0.01 Post-high school education - - - - 0.00 0.01 Presentence detainment - - - - 0.08 0.02 *** Trial conviction - - - - 0.18 0.02 *** Immigration crimes - - - - -0.06 0.01 *** Other crimes - - - - -0.06 0.01 *** Hazard rate -5.61 0.12 *** -1.83 0.07 *** -1.60 0.08 *** Block of district dummies - - - - - - R2 0.29 0.86 0.86 N 25,927 25,927 25,927 ? p = .1; * p = .05; ** p = .01; *** p = .001. References: Other Offense, District 81. 67 Table 8. Unconditional Two-Level HLM Model of Sentence Length Ln Sentence Length Fixed effects b SE Intercept 0.25*** 0.05 Random effects Variance SD Level 1 1.11*** 1.05 Level 2 0.11*** 0.34 Between State proportion of variance 0.09 ? p = .1; * p = .05; ** p = .01; *** p = .001. Table 9. Two-Level HLM Random Intercept Model - Individual- Level Fixed Effects Individual-level factors Ln Sentence Length b SE Constant 0.22 0.02 *** Defendant is non-citizen -0.06 0.02 ** Presumptive Sentence 0.72 0.02 *** Criminal history 0.02 0.00 *** Downward Departures -0.40 0.02 *** Substantial Assistance Departure -0.57 0.09 *** Early Disposition Departure -0.57 0.05 *** Upward Departures 0.76 0.07 *** Multiple counts 0.05 0.02 * Defendant's age 0.00 0.00 *** Defendant is male 0.06 0.01 *** Black 0.11 0.03 *** Financial dependent 0.00 0.00 Post-high school education 0.00 0.01 Presentence detainment 0.08 0.07 Trial conviction 0.18 0.03 *** Immigration crimes -0.05 0.02 ? Other crimes -0.04 0.02 ? Level 1 R2 0.85 Level 2 R2 0.92 ? p = .1; * p = .05; ** p = .01; *** p = .001. 68 Table 10. Two-Level HLM Random Intercept Model - Individual- Level and Contextual Fixed Effects Individual-level factors Ln Sentence Length b SE Constant 0.12 0.02 *** Defendant is non-citizen -0.06 0.02 *** Presumptive Sentence 0.72 0.02 *** Criminal history 0.02 0.00 *** Downward Departures -0.40 0.02 *** Substantial Assistance Departure -0.57 0.09 *** Early Disposition Departure -0.56 0.05 *** Upward Departures 0.76 0.07 *** Multiple counts 0.05 0.02 * Defendant's age 0.00 0.00 *** Defendant is male 0.06 0.01 *** Black 0.11 0.03 *** Financial dependent 0.00 0.00 Post-high school education 0.00 0.01 Presentence detainment 0.08 0.07 Trial conviction 0.18 0.03 *** Immigration crimes -0.05 0.02 ? Other crimes -0.04 0.02 ? Fast-track/287(g) programs 0.00 0.03 Favor deportation -0.06 0.10 Estimated unauthorized immigrants 0.00 0.01 Hispanic political representation -0.01 0.00 *** Level 1 R2 0.85 Level 2 R2 0.94 ? p = .1; * p = .05; ** p = .01; *** p = .001. 69 Table 11. Two-Level HLM Random Coefficient Model - Individual and Contextual- Level Fixed Effects Part A Individual-level factors Ln Sentence Length b SE Constant 0.12 0.02 *** Defendant is non-citizen -0.06 0.02 *** Presumptive Sentence 0.75 0.01 *** Criminal history 0.01 0.00 *** Downward Departures -0.42 0.03 *** Substantial Assistance Departure -0.57 0.04 *** Early Disposition Departure -0.58 0.04 *** Upward Departures 0.73 0.05 *** Multiple counts 0.05 0.02 ** Defendant's age 0.00 0.00 *** Defendant is male 0.06 0.02 *** Black 0.09 0.03 *** Financial dependent 0.00 0.00 Post-high school education 0.00 0.01 Presentence detainment 0.07 0.07 Trial conviction 0.16 0.03 *** Immigration crimes -0.04 0.02 ? Other crimes -0.03 0.03 Fast-track/287(g) programs -0.01 0.02 Favor deportation 0.03 0.05 Estimated unauthorized immigrants 0.01 0.00 *** Hispanic political representation -0.01 0.00 *** ? p = .1; * p = .05; ** p = .01; *** p = .001. Part B Random effects Variance component SD of variance component Defendant is non-citizen 0.0067 0.0819 *** Presumptive Sentence 0.0078 0.0886 *** Criminal history 0.0004 0.0193 *** Substantial Assistance Departure 0.0515 0.2269 *** ? p = .1; * p = .05; ** p = .01; *** p = .001. 70 Table 12. HLM Cross-Level Citizenship Status Interaction Models Ln Sentence Length b SE Fast-track/287(g) programs Interaction Intercept 0.123 0.018 *** Defendant is non-citizen -0.059 0.017 *** Fast-track/287(g) programs -0.007 0.017 Fast-track/287(g) programs*non-citizen 0.006 0.025 Attitudes toward deportation Interaction Intercept 0.123 0.018 *** Defendant is non-citizen -0.061 0.016 *** Attitudes toward deportation 0.026 0.053 Attitudes toward deportation*non-citizen -0.059 0.078 Unauthorized immigrant Interaction Intercept 0.123 0.018 *** Defendant is non-citizen -0.057 0.016 *** Estimated unauthorized immigrants 0.014 0.003 *** Unauthorized immigrants*non-citizen 0.002 0.004 Hispanic political representation Interaction Intercept 0.121 0.018 *** Defendant is non-citizen -0.080 0.020 *** Hispanic political representation -0.006 0.001 *** Hispanic political rep.*non-citizen -0.002 0.001 ? ? p = .1; * p = .05; ** p = .01; *** p = .001. 71 Figure 1: Sentence Length in Months for Incarcerated Offenders 0 .0 05 .0 1 .0 15 .0 2 .0 25 D en si ty 0 100 200 300 400 500 SLENGTH Figure 2: Natural Log of Sentence Length in Months for Incarcerated Offenders 0 .2 .4 .6 D en si ty 0 2 4 6 LNSLENGTH 72 Bibliography Albonetti, Celesta A. 1991. An integration of theories to explain judicial discretion. Social Problems 38(2): 247-266. Albonetti, Celesta A. 1997. Sentencing under the federal sentencing guidelines: Effects of defendant characteristics, guilty pleas, and departures on sentence outcomes for drug offenses, 1991-1992. Law & Society Review 31(4): 789-822. Allegro, Linda. 2010. Latino migrations to the U.S. heartland: ?Illegality,? state controls, and implications for transborder labor rights. Latin American Perspectives, Issue 170, 37(1): 172-184. Alonso Meneses, Guillermo. 2003. Human rights and undocumented migration along the Mexican-U.S. border. UCLA Law Review 51: 267-282. Arbona, Consuelo, Norma Olvera, Nestor Rodriguez, Jacqueline Hagan, Adriana Linares, and Margit Wiesner. 2010. Acculturative stress among documented and undocumented Latino immigrants in the United States. Hispanic Journal of Behavioral Sciences 32(3): 362-384. Armenta, Amada. 2012. From sheriff?s deputies to immigration officers: Screening immigrant status in a Tennessee jail. Law & Policy 34(2): 191-210. Ashcroft, John. 2003. Memorandum from the Attorney General to all United States Attorneys: Department principles for implementing an expedited disposition or ?Fast-Track? prosecution program in a district. Office of the Attorney General. September 22, 2003. Bhuyan, Rupaleem. 2008. The production of the ?battered immigrant? in public policy and domestic violence advocacy. Journal of Interpersonal Violence 23(2): 153-170. Bibler Coutin, Susan. 2011. The rights of noncitizens in the United States. The Annual Review of Law and Social Science 7: 289-308. Bloch, Alice and Liza Schuster. 2005. At the extremes of exclusion: Deportation, detention and dispersal. Ethnic and Racial Studies 28(3): 491-512. Blumer, Herbert. 1958. Race prejudice as a sense of group position. The Pacific Sociological Review 1(1): 3-7. Bosworth, Mary. 2012. Subjectivity and identity in detention: Punishment and society in a global age. Theoretical Criminology 16(2): 123-140. 73 Brabeck, Kalina, and Qingwen Xu. 2010. The impact of detention and deportation on Latino immigrant children and families: A quantitative exploration. Hispanic Journal of Behavioral Sciences 32(3): 341-361. Britt, Chester L. 2000. Social context and racial disparities in punishment decisions. Justice Quarterly 17(4): 707-732. Brody, Francesca. 2011. Extracting compassion from confusion: Sentencing noncitizens after United States v. Booker. Fordham Law Review 79: 101-144. Brotherton, David C., and Luis Barrios. 2009. Displacement and stigma: The social- psychological crisis of the deportee. Crime Media Culture 5(1): 29-55. Brotherton, David C., and Yolanda Martin. 2009. The war on drugs and the case of Dominican deportees. Journal of Crime & Justice 32(2): 21-48. Bucher, Jacob, Michelle Manasse, and Beth Tarasawa. 2010. Undocumented victims: An examination of crimes against undocumented male migrant workers. Southwest Journal of Criminal Justice 7(2): 159-179. Budeiri, Priscilla. 1981. Collateral consequences of guilty pleas in the federal criminal justice system. Harvard Civil Rights-Civil Liberties Law Review 16: 157- 203. Bureau of Justice Statistics. 1996. Noncitizens in the Federal Criminal Justice System, 1984-94. http://www.prisonpolicy.org/scans/bjs/nifcjs.pdf. Bureau of Justice Statistics. 2008. Confined non-U.S. citizens in state or federal prison: Table 20: Reported number of non-U.S. citizens held in state or federal prisons, by gender, region, and jurisdiction, June 30, 2007-2008. http://bjs.ojp.usdoj.gov/index.cfm?ty=tp&tid=145. Bureau of Justice Statistics. 2011. Prisoners in 2010. http://www.bjs.gov/content/pub/pdf/p10.pdf. Caravelis, Cyndy, Ted Chiricos, and William Bales. 2011. Static and dynamic indicators of minority threat in sentencing outcomes: A multi-level analysis. Journal of Quantitative Criminology 27 (4): 405-425. Cervantes, Joseph M., Olga L. Mej?a, and Amalia Guerrero Mena. 2010. Serial migration and the assessment of extreme and unusual psychological hardship with undocumented Latino/a families. Hispanic Journal of Behavioral Sciences 32(2): 275-291. Chin, Gabriel. 2011. Illegal entry as crime, deportation as punishment: Immigration status and the criminal process. UCLA Law Review 58: 1417-1459. 74 Chin, Gabriel J. and Margaret Love. 2010. Status as punishment: A critical guide to Padilla v. Kentucky. Criminal Justice 25(3): 21-31. Cho, Abe. 2010. Lowering sentences for illegal immigrants? Why judges should have discretion to vary from the guidelines based on fast-track sentencing disparities. Columbia Journal of Law and Social Problems 43(4): 447-490. Coleman, Mathew. 2012. The ?local? migration state: The site-specific devolution of immigration enforcement in the U.S. South. Law & Policy 34(2): 159-190. Cruz, Evelyn H. 2010. Competent voices: Noncitizen defendants and the right to know the immigration consequences of plea agreements. Harvard Latino Law Review 13: 47-66. Dadhania, Pooja R. 2011. The categorical approach for crimes involving moral turpitude after Silva-Trevino. Columbia Law Review 111: 313-355. Das, Alina. 2008. Immigrants and problem-solving courts. Criminal Justice Review 33: 308-328. Das, Alina. 2011. The immigration penalties of criminal convictions: Resurrecting categorical analysis in immigration law. New York University Law Review 86(6): 1669-1760. Davenport, Maria-Teresa. 2006. Deportation and driving: Felony DUI and reckless driving as crimes of violence following Leocal v. Ashcroft. The Journal of Criminal Law and Criminology 96(3): 849-875. Demleitner, Nora V., and Jon M. Sands. 2002. Non-citizen offenders and immigration crimes: New challenges in the federal system. Federal Sentencing Reporter 14(5): 247-254. Demuth, Stephen. 2002. The effect of citizenship status on sentencing outcomes in drug cases. Federal Sentencing Reporter 14(5): 271-275. Demuth, Stephen, and Darrell Steffensmeier. 2004. Ethnicity effects on sentence outcomes in large urban courts: Comparisons among White, Black, and Hispanic defendants. Social Science Quarterly 85(4): 994-1011. Department of Homeland Security. 2012a. Estimates of the Unauthorized Immigrant Population Residing in the United States: January 2011. http://www.dhs.gov/xlibrary/assets/statistics/publications/ois_ill_pe_2011.pdf. Department of Homeland Security 2012b. Estimates of the Legal Permanent Resident Population in 2011. http://www.dhs.gov/xlibrary/assets/statistics/publications/ois_lpr_pe_2011.pdf. 75 Department of Homeland Security 2012c. Yearbook of Immigration Statistics 2011.http://www.dhs.gov/sites/default/files/publications/immigrationstatistics/yea rbook/2011/ois_yb_2011.pdf. Department of Justice. 2002. 2000 Statistical Yearbook of the Immigration and Naturalization Service. http://www.dhs.gov/xlibrary/assets/statistics/yearbook/2000/Yearbook2000.pdf. Department of Justice. 2008. Office of the Deputy Attorney General. Memorandum of Reauthorization of Early Disposition Programs. February 1, 2008. http://www.fd.org/docs/select-topics---sentencing/u-s-department-of-justice-2-1- 08-memo-regarding-reauthorization-of-early-disposition-programs-%28fast- track%29.pdf?sfvrsn=4. Diaz, Jesse. 2011. Immigration policy, criminalization and the growth of the immigration industrial complex: Restriction, expulsion, and eradication of the undocumented in the U.S. Western Criminology Review 12(2): 35-54. DiDenti Christiansen, Elise. 2010. Adolescent Cape Verdean girls? experiences of violence, incarceration, and deportation: Developing resources through participatory community-based groups. International Journal of Intercultural Relations 34(2): 127-140. Dixon, Jo. 1995. The organizational context of criminal sentencing. American Journal of Sociology 100(5): 1157-1198. Dow, Mark. 2007. Designed to punish: Immigrant detention and deportation. Social Research 74(2): 533-546. Dubber, Markus D. 2010. Citizenship and penal law. New Criminal Law Review 13(2): 190-215. Eagly, Ingrid. 2010. Prosecuting immigration. Northwestern University Law Review 104(4): 1281-1360. Engen, Rodney L., and Randy R. Gainey. 2000. Modeling the effects of legally relevant and extralegal factors under sentencing guidelines: The rules have changed. Criminology 38(4): 1207-1229. Ewald, Alec C. 2011. Deportation, effective counsel, and collateral sanctions: Padilla v. Kentucky (2010). Justice System Journal 32(2): 235-242. Ewing, Walter A. 2010. The many facets of effective immigration reform. Society 47: 110-117. Farrell, Amy, and Geoff Ward. 2011. Examining district variation in sentencing in the post-Booker period. Federal Sentencing Reporter 23(5): 318-325. 76 Fassin, Didier. 2011. Policing borders, producing boundaries. The governmentality of immigration in dark times. Annual Review of Anthropology. 40: 213-226. Flynn, Don. 2005. New borders, new management: The dilemmas of modern immigration policies. Ethhnic and Racial Studies 28(3): 463-490. Federal Bureau of Prisons. 2009. The state of the Bureau 2009. http://www.bop.gov/news/PDFs/sob09.pdf. Federal Bureau of Prisons. 2012. Quick facts about the Bureau of Prisons. http://www.bop.gov/news/quick.jsp. Fekete, Liz, and Frances Webber. 2010. Foreign nationals, enemy penology and the criminal justice system. Race & Class 51(4): 1-25. Franko Aas, Katja. 2011. Global criminology. In The SAGE Handbook of Criminological Theory, eds. Eugene McLaughlin and Tim Newburn. Los Angeles, CA: Sage. Frey, Barbara A., and X. Kevin Zhao. 2011. The criminalization of immigration and the international norm of non-discrimination: Deportation and detention in U.S. immigration law. Law & Inequality 77: 279-312. Fussell, Elizabeth. 2011. The deportation threat dynamic and victimization of Latino migrants: Wage theft and robbery. The Sociological Quarterly 52(4): 593-615. Gibney, Matthew J. 2008. Asylum and the expansion of deportation in the United Kingdom. Government and Opposition 43(2): 146-167. G?mez-Jara D?ez, Carlos. 2008. Enemy combatants versus enemy criminal law: An introduction to the European debate regarding enemy criminal law and its relevance to the anglo-american discussion on the legal status of unlawful enemy combatants. New Criminal Law Review 11(4): 529-562. Gonzales, Roberto G., and Leo R. Chavez. 2012. ?Awakening to a nightmare?: Abjectivity and illegality in the lives of undocumented 1.5-generation Latino immigrants in the United States. Current Anthropology 53(3): 255-281. Gorman, Thomas E. 2009. A history of fast-track sentencing. Federal Sentencing Reporter 21(5): 311-317. Guerra Thompson, Sandra. 2008. Immigration law and long-term residents: A missing chapter in American criminal law. Ohio State Journal of Criminal Law 5: 645-662. 77 Hagan, Jacqueline, and Scott Phillips. 2008. Border blunders: The unanticipated human and economic costs of the U.S. approach to immigration control, 1986- 2007. Criminology & Public Policy 7(1): 83-94. Hagan, Jacqueline Maria, Nestor Rodriguez, and Briann Castro. 2011. Social effects of mass deportations by the United States government, 2000-10. Ethnic and Racial Studies 34(8): 1374-1391. Hagan, John, and Alberto Palloni. 1999. Sociological criminology and the mythology of Hispanic immigration and crime. Social Problems 46(4): 617-632. Hagan, John, Ron Levi, and Ronit Dinovitzer. 2008. The symbolic violence of the crime-immigration nexus: Migrant mythologies in the Americas. Criminology & Public Policy 7(1): 95-112. Hartley, Richard D., and Luisa F. Armendariz. 2011. Border justice? Sentencing federal narcotics offenders in Southwest border districts: A focus on citizenship status. Journal of Contemporary Criminal Justice 27(1): 43-62. Hartley, Richard D., and Rob Tillyer. 2012. Defending the homeland: judicial sentencing practices for federal immigration offenses. Justice Quarterly 29(1): 76-104. Helms, Ronald, and David Jacobs. 2002. The political context of sentencing: An analysis of community and individual determinants. Social Forces 81(2): 577- 604. Iles, Gale. 2009. The effects of race/ethnicity and national origin on length of sentence in the United States Virgin Islands. The Annals of the American Academy of Political and Social Science 623: 64-76. Jacobs, David, and Jason T. Carmichael. 2002. The political sociology of the death penalty: A pooled time-series analysis. American Sociological Review 67(1): 109- 131. Jacobs, David, Jason T. Carmichael, and Stephanie L. Kent. 2005. Vigilantism, current racial threat, and death sentences. American Sociological Review 70(4): 656-677. Jacobs, David, and Richard Kleban. 2003. Political institutions, minorities, and punishment: A pooled cross-national analysis of imprisonment rates. Social Forces 82 (2): 725-755. Johnson, Brian D. 2005. Contextual disparities in guidelines departures: Courtroom social contexts, guidelines compliance, and extralegal disparities in criminal sentencing. Criminology 43(3): 761-796. 78 Johnson, Brian D. 2006. The multilevel context of criminal sentencing: integrating judge and county level influences. Criminology 44(2): 259-298. Johnson, Brian D., 2010. Multilevel analysis in the study of crime and justice, in Piquero, Alex R., and David Weisburd, ed., Handbook of Quantitative Criminology. 1st Ed., XIII: 615-648. Johnson, Brian D., and Sara Betsinger. 2009. Punishing the ?model minority?: Asian- American criminal sentencing outcomes in federal district courts. Criminology 47(4): 1045-1090. Johnson, Brian D., Eric A. Stewart, Justin Pickett, and Marc Gertz. 2011. Ethnic threat and social control: Examining public support for judicial use of ethnicity in punishment. Criminology 49(2): 401-441. Johnson, Brian D., Jeffery T. Ulmer, and John Kramer. 2008. The social context of guidelines circumvention: The case of federal district courts. Criminology 46(3): 737-783. Johnson, Kevin R., and Bernard Trujillo. 2007. Immigrant reform, national security after September 11, and the future of North American integration. Minnesota Law Review 91: 1369-1406. Kanstroom, Daniel. 2000. Deportation, social control, and punishment: Some thoughts about why hard laws make bad bases. Harvard Law Review 113: 1890- 1935. Kanstroom, Daniel. 2011. The right to deportation counsel in Padilla v. Kentucky: The challenging construction of the fifth-and-a-half amendment. UCLA Law Review 58: 1461-1514. Kautt, Paula, M. 2002. Location, location, location: interdistrict and intercircuit variation in sentencing outcomes for federal drug-trafficking offenses. Justice Quarterly 19(4): 633-671. King, Ryan D., Kecia R. Johnson, and Kelly McGeever. 2010. Demography of the legal profession and racial disparities in sentencing. Law & Society Review 44(1): 1-31. Kramer, John H., and Jeffery T. Ulmer. 1996. Sentencing disparity and departures from guidelines. Justice Quarterly 13(1): 81-106. Landale, Nancy S., Kevin J. A. Thomas, Jennifer Van Hook. 2011. The living arrangements of children of immigrants. The Future of Children 21(1): 43-70. 79 Lawston, Joddie Michelle, and Martha Escobar. 2009. Policing, detention, deportation, and resistance: Situating immigrant justice and carcerality in the 21st century. Social Justice 36(2): 1-6. Legomsky, Stephen H. 2006. Deportation and the war on independence. Cornell Law Review 91: 369-410. Light, Michael. 2010. Has racial/ethnic disparity in punishment decisions changed over time: An analysis of U.S. Federal Courts 1992-2004 (Master?s thesis, The Pennsylvania State University). Logue, Melissa A. 2009. ?The price of being Mexican?: Sentencing disparities between noncitizen Mexican and non-Mexican Latinos in the federal courts. Hispanic Journal of Behavioral Sciences 31(4): 423-445. Lopez, Mark Hugo, and Susan Minushkin. 2008. National Survey of Latinos Report. Pew Hispanic Center. Martinez, Jenny S. 2008. Process and substance in the ?War on Terror?. Columbia Law Review 108(5): 1013-1092. McDermid, Lea. 2001. Deportation is different: Noncitizens and ineffective assistance of counsel. California Law Review 89: 741-778. McLaughlin, Eugene. 2011. Critical criminology. In The SAGE Handbook of Criminological Theory, eds. Eugene McLaughlin and Tim Newburn. Los Angeles, CA: Sage. Mendelson, Margot K. 2010. Constructing America: Mythmaking in U.S. immigration courts. The Yale Law Journal 119: 1012-1058. Mitchell, Ojmarrh, and Doris L. MacKenzie. 2004. The relationship between race, ethnicity, and sentencing outcomes: A meta-analysis of sentencing research. Research Summary. Final Report. National Institute of Justice. Mitchell, Ojmarrh. 2005. A meta-analysis of race and sentencing research: Explaining the inconsistencies. Journal of Quantitative Criminology, 21(4): 439-466. Montague, M. Gavan. 2001. Should aliens be indefinitely detained under 8 U.S.C. ? 1231? Suspect doctrines and legal fictions come under renewed scrutiny. Fordham Law Review 69: 1439-1486. Moore, Derrick. 2008. ?Crimes involving moral turpitude?: Why the void-for- vagueness argument is still available and meritorious. Cornell International Law Journal 41: 813-843. 80 Morawetz, Nancy. 2000. Understanding the impact of the 1996 deportation laws and the limited scope of proposed reforms. Harvard Law Review 113: 1936-1962. Morrison, Trevor. 1997. Removed from the Constitution? Deportable aliens? access to Habeas Corpus under the new immigration legislation. Columbia Journal of Transnational Law 35: 697-722. Murphy, Kara B. 2012. Representing noncitizens in criminal proceedings: Resolving unanswered questions in Padilla v. Kentucky. The Journal of Criminal Law & Criminology. 101(4): 1371-1401. Mustard, David B. 2001. Racial, ethnic, and gender disparities in sentencing: evidence from U.S. Federal Courts. Journal of Law and Economics 44: 285-314. Myers, Martha A., and Susette M. Talarico. 1987. The Social Contexts of Criminal Sentencing. Research in Criminology. New York: Springer-Verlag. National Conference of State Legislatures. 2009. Latino Legislators. http://www.ncsl.org/legislatures-elections/legisdata/latino-legislators- overview.aspx. Ogden, David W. 2009. Memorandum from the Deputy Attorney General to all United States Attorneys: Authorization for Certain Early Disposition Programs. Office of the Deputy Attorney General. May 29, 2009. Ortiz Maddali, Anita. 2011. Padilla v. Kentucky: A new chapter in Supreme Court jurisprudence on whether deportation constitutes punishment for lawful permanent residents. American University Law Review 61: 1-58. Parrado, Emilio A. 2012. Immigration enforcement policies, the economic recession, and the size of local Mexican immigrant populations. The Annals of the American Academy of Political and Social Science 641: 16-37. Pauw, Robert. 2000. A new look at deportation as punishment: Why at least some of the Constitution?s criminal procedure protections must apply. Administrative Law Review 52: 305-345. Pew Hispanic Center. 2010. Demographic profile of Hispanics, 2010. http://www.pewhispanic.org/states/. Pew Hispanic Center. 2008. Statistical Portrait of Hispanics in the United States, 2008. Table 13. Hispanic Population by State: 2008. Pew Hispanic Center. http://www.pewhispanic.org/files/2010/01/2008-Hispanic-Profile-Final.pdf. Pew Hispanic Center. 2012. Unauthorized Immigrants: 11.1 Million in 2011. http://www.pewhispanic.org/2012/12/06/unauthorized-immigrants-11-1-million- in-2011/ 81 Phillips, Scott, Jacqueline Maria Hagan, and Nestor Rodriguez. 2006. Brutal borders? Examining the treatment of deportees during arrest and detention. Social Forces 85(1): 93-109. Pinard, Michael. 2006. An integrated perspective on the collateral consequences of criminal convictions and reentry issues faced by formerly incarcerated individuals. Boston University Law Review 86: 623-690. Provine, Doris Marie. 2011. Race and inequality in the war on drugs. Annual Review of Law and Social Science 7: 41-60. Quinn, Genevieve. 2011. Broken borders, broker laws: Aligning crime and punishment under section 2L1.1(b)(7) of the U.S. sentencing guidelines. Fordham Law Review 80: 923-958. Quinney, Richard. 1975. Conflict theory of crime. In Criminology. Boston: Little, Brown: 37-41. Reyes, Maritza. 2012. Constitutionalizing immigration law: The vital role of judicial discretion in the removal of lawful permanent residents. Temple Law Review 84: 637-700. Robinson, William I. 2006. ?Aqui estamos y no nos vamos!? Global capital and immigrant rights. Race & Class 48(2): 77-91. Sampson, Robert. 2008. Rethinking Crime and Immigration. Contexts 7(1): 28-33. Savelsberg, Joachim. J. 1992. Law that does not fit society: Sentencing guidelines as a neoclassical reaction to the dilemmas of substantivized law. American Journal of Sociology 97(5): 1346-1381. Schlesinger, Traci. 2005. Racial and ethnic disparity in pretrial criminal processing. Justice Quarterly 22(2): 170-192. Shermer, Lauren O?Neill, and Brian D. Johnson. 2010. Criminal prosecutions: Prosecutorial discretion and charge reductions in U.S. federal district courts. Justice Quarterly 27(3): 394-430. Skitka, Linda J., Christopher W. Bauman, Nicholas P. Aramovich, and G. Scott Morgan. 2006. Confrontational and preventative policy responses to terrorism: Anger wants a fight and fear wants ?them? to go away. Basic and Applied Social Psychology 28(4): 375-384. Sklansky, David Alan. 2012. Crime, immigration, and ad hoc instrumentalism. New Criminal Law Review 15(2): 157-223. 82 Spohn, Cassia. 2005. Sentencing decisions in three U.S. district courts: testing the assumption of uniformity in the federal sentencing process. Justice Research and Policy 7(2): 1-27. Steffensmeier, Darrell, and Stephen DeMuth. 2000. Ethnicity and sentencing outcomes in U.S. federal courts: Who is punished more harshly? American Sociological Review 65:705-729. Steffensmeier, Darrell, and Stephen Demuth. 2001. Ethnicity and judges? sentencing decisions: Hispanic-Black-White comparisons. Criminology 39(1): 145-178. Steffensmeier, Darrell, Jeffery Ulmer, and John Kramer. 1998. The Interaction of race, gender, and age in criminal sentencing: The punishment cost of being young, Black, and male. Criminology 36(4): 763-797. Stith, Kate and Jose Cabranes. 1998. Fear of Judging: Sentencing Guidelines in the Federal Courts. Chicago: University of Chicago Press. Stowell, Jacob I., Steven F. Messner, Kelly F. McGeever, and Lawrence E. Raffalovich. 2009. Immigration and the recent violent crime drop in the United States: A pooled, cross-sectional time-series analysis of metropolitan areas. Criminology 47(3): 889-928. Stumpf, Juliet P. 2011. Doing time: Crimmigration law and the perils of haste. UCLA Law Review 58: 1705-1748. Tambini, Damian. 2001. Post-national citizenship. Ethnic and Racial Studies 24(2): 195-217. Taylor, Margaret H., and Ronald F. Wright. 2002. The sentencing judge as immigration judge. Emory Law Journal 51: 1131-1174. The Sentencing Project. 2007. Uneven Justice: State Rates of Incarceration by Race and Ethnicity.http://www.sentencingproject.org/doc/publications/rd_stateratesofincbyr aceandethnicity.pdf. Tonry, Michael. 1996. Sentencing Matters. New York: Oxford University Press. Transactional Records Access Clearinghouse (TRAC) Immigration Project. 2008. U.S. Deportation Proceedings in Immigration Courts. http://trac.syr.edu/phptools/immigration/charges/deport_filing_charge.php. Ulmer, Jeffery T. 2005. The localized uses of federal sentencing guidelines in four U.S. district courts: Evidence of processual order. Symbolic Interaction 28(2): 255-279. 83 Ulmer, Jeffery. 1997. Social Worlds of Sentencing: Court Communities under Sentencing Guidelines. Albany: State University of New York Press. Ulmer, Jeffery T., James Eisenstein, and Brian D. Johnson. 2010. Trial penalties in federal sentencing: Extra-guidelines factors and districts variation. Justice Quarterly 27(4): 560-592. Ulmer, Jeffery, and Brian D. Johnson. 2004. Sentencing in context: A multilevel analysis. Criminology 41(1): 137-178. Ulmer, Jeffery T. 2012. Recent developments and new directions in sentencing research. Justice Quarterly 29(1): 1-40. USSC-United States Sentencing Commission. 2008. The Changing Face of Federal Criminal Sentencing.http://www.ussc.gov/Research/Research_Publications/2008/20090127 _Changing_Face_Fed_Sent.pdf. U.S. Census Bureau. 2007. State Tables. Annual Population Estimates 2000-2007. http://www.census.gov/popest/data/historical/2000s/vintage_2007/state.html. U.S. Census Bureau. 2001. The Hispanic population: 2000. Census 2000 Brief. http://www.census.gov/prod/2001pubs/c2kbr01-3.pdf. U.S. Census Bureau. 2011a. The Hispanic population: 2010. 2010 Census Briefs. http://www.census.gov/prod/cen2010/briefs/c2010br-04.pdf. U.S. Census Bureau. 2011b. Profile America Facts for Features. http://www.census.gov/newsroom/releases/archives/facts_for_features_special_editions/cb11f f18.html. Wang, Xia. 2012. Undocumented immigrants as perceived criminal threat: A test of the minority threat perspective. Criminology 50(3): 743-776. Warren, Patricia, Ted Chiricos, and William Bales. 2012. The imprisonment penalty for young Black and Hispanic males: A crime-specific analysis. Journal of Research in Crime and Delinquency 49(1): 56-80. Welch, Jennifer. 2004. Defending against deportation: Equipping public defenders to represent noncitizens effectively. California Law Review 92: 541-584. Wishnie, Michael J. 2011. Proportionality: The struggle for balance in U.S. immigration policy. University of Pittsburgh Law Review 72: 431-466. Wolfe, Scott E., David C. Pyrooz, and Cassia C. Spohn. 2011. Unraveling the effect of offender citizenship status on federal sentencing outcomes. Social Science Research 40: 349-362. 84 Wu, Jawjeong, and Miriam A. DeLone. 2012. Revisiting the normal crime and liberation hypotheses: Citizenship status and unwarranted disparity. Criminal Justice Review 37(2): 214-238. Wu, Jawjeong, and Cassia Spohn. 2010. Interdistrict disparity in sentencing in three U.S. district courts. Crime & Delinquency 56(2): 290-322. Zedner, Lucia. 2010. Security, the state, and the citizen: The changing architecture of crime control. New Criminal Law Review 13(2): 379-403.