. 3v: F.» t u t v . 4 .r 4 V .33. .. . 2.": t a” :2 1 2.7. ; n 4.. . tr 1 Ell ‘4: . . .;u..:« I... :1... . , . . J 3. 1a? , . . .. ,. . . V n . y . y . . £43.... $143.3? . .r “3...”..9 . 7;... . w w.» ‘ I. .. . , , . , . $23.55.“ ‘ ‘ 3.5%.. A a an. a...” _....§w? Em? fiéflfig E Emfifié $5.444: II oh-KJ-U ? fllllllllllIfliiiilliiliifiifilfi'ilifliiliiNIH!lllll 31293 01812 7161 This is to certify that the dissertation entitled ##7##“ mcflmmflw W55. far/Mm; 02/ W ”9%,“? % (2,56; presented by has been accepted towards fulfillment of the requirements for %' 9 degree in flMCé/Wc‘g Date €//?/% MSU is an Affirmative Action/Equal Opportunity Institution 0- 12771 “fl A 'rfi .__v—~._'—~ W LIBRARY Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. I DATE DUE, DATE DUE DATE DUE If EQ gap ,3 299; PRINCIPAL-AGENCY IN AMERICAN COURTS: PERSPECTIVES ON THE HIERARCHY OF JUSTICE By Sara Catherine Benesh A DISSERTATION Submitted to Michigan State University in partial fiJlfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1999 ABSTRACT PRINCIPAL-AGENCY IN AMERICAN COURTS: PERSPECTIVES ON THE HIERARCHY OF JUSTICE By Sara Catherine Benesh For many years, the scholarship on judicial decision making was limited to one admittedly unique court: the United States Supreme Court. However, scholars have begun to appreciate the importance of the lower federal courts, as the Courts of Appeals has become, for all intents and purposes, the final arbiter of the law. The Supreme Court decides less than 100 cases each year, while the caseload of the Courts of Appeals is consistently expanding. Given this phenomenon, it becomes essential that we understand decision making in that lower court. We understand how the Supreme Court makes decisions - its justices vote their attitudes — but the decision making of the lower court judge is still somewhat elusive. In this dissertation, I construct an integrated model of Court of Appeals decision making which includes the ideological orientation of the justices as well as the relevant precedents of the Supreme Court. Such an inclusion is necessary as the Court of Appeals judge finds him or herself in a position much different from the position in which the Supreme Court justice finds him or herself: the lower court is inferior to the Supreme Court. That is, the lower court is subject to the jurisprudence of its boss, or principal. Understanding decision making in the lower court, then, necessitates an understanding of the relationship between it and the High Court. By modeling said relationship, we can begin to understand why most of the decisions made in the United States are made as they are. ”7‘1 CCL In this dissertation, I first construct a formal model of the relationship between the Supreme Court and the Court of Appeals, concluding that the relationship, while closely resembling principal agency, is appreciably different. Although it is not in its best interests to do so - at least where the effectuation of policy is concerned - the lower court does indeed comply with the Supreme Court. Because of the lack of supervision afforded by the Supreme Court, it becomes necessary to vary the principal-agent relationship. My evidence shows that the lower court follows Supreme Court prescription because of what the Supreme Court is rather than because of what it does. Taking this relationship as a given, I model Court of Appeals decision making in confession cases. I find that the lower court decides cases according to its own policy preferences, the ideological composition of the Court, and the precedential guidance the Court has offered. In short, the lower courtjudge is a faithful agent, and yet an ideological actor. Looking more closely at the treatment afforded Supreme Court precedent, I find that, while the lower court does sometimes distinguish unliked precedents, or perhaps decides without regard to the spirit of the Supreme Court ruling on the matter, the lower courts do indeed comply. The institutional makeup of the Court of Appeals mandates that it decide cases differently from the Supreme Court, which decides solely on the basis of its members’ policy preferences. The lower court, by contrast, must account for other factors and other actors. Here we find that b0th Supreme Court. ideology ; ad Supreme Couttjurisprudence matter to the lower court. Although many questions remain unanswered, this study makes a substantial contribution to understanding both decision making on the lower court, and the relationship between said lower court and the High Court. This dissertation is dedicated to the one person without whom it could never have been completed: Professor Harold J. Spaeth i\ sup COI Oil W2 In It ACKNOWLEDGMENTS There are many people to whom I owe a great debt of gratitude, for without their support and encouragement, this dissertation (and my life!) would have suffered considerably. First of all, I’d like to thank my parents, the two people I can always count on for encouragement, confidence, pride, and the truth. You two have been the best parents a child can hope for and thensome. Thank you, Mom, for all those stories when I was little, and for the day by day Speeches now that I’m grown! Thank you, Dad, for listening to all my boring stories about political science, for actually trying to be interested in the papers I’ve written, and for being so darn proud of me! I love you both more than words can say, but I’m fairly certain you already know that! Thanks to Ben and Aaron, my wonderful little brothers. Visits with you rejuvenated me and reminded me that no matter what I did, I would matter to some pretty important people. I am so proud of both of you, and love you very much. Thanks to my wonderfully supportive and helpful dissertation committee. To Reggie, for giving me a hard time and challenging me at every turn. To Melinda, for the provocative comments which helped to prepare me well for any criticism that might come my way. To Jim, for the excellent methods training and helpful guidance. And to Harold, for being the best advisor a person could ever hope for! I cannot say enough about the influence you’ve had over me and the integral part you will have played in any success I may have in the future. Finally, but of course not least, I’d like to thank my fabulous group of friends who have made this whole graduate school process bearable. To Gina, for understanding and cheering me on and being the best, most thoughtfiIl friend I could ever have hoped to meet frien weel goin' in pt your and yow emt Mic any the? meet. To Annie, for the laughter, the many cards, the life saving visits, and the undying friendship and love. To Jen, Jenni, and Colleen for the funny emails, the awesome long weekends, and the reality checks. To Wendy, for the kind of support only another person going through exactly the same things can provide. To Misa for being a good roommate in perhaps my most stressfiIl year, and for teaching me Japanese and letting me talk to your mother! To Dave for being one of the only reasons I stayed through my first year and one of my best memories about this place. To Malia for showing me the way and for your continued guidance and support. To Joy for making me laugh and not uncommonly embarrassing me; to Matt for your early friendship; and to all the others — Kimberly, Michelle, Mark, Kirk, Chris, Sara, Chris, Brandon, Bryan, Lu-Huei, Erik, Scott, and anyone I’m just merely not remembering right now — for making graduate school a little more bearable. And finally, to Don, for loving me. I have been blessed with many wonderful people in my life, and I thank God for their love and support. I am who and what I am today because of them. Thank you!!! vi Int Ch Chapter 1: Chapter 2: Chapter 3: Chapter 4: TABLE OF CONTENTS Introduction ......................................................................................... 1 The Hierarchy of Justice ......................................................... 4 1.1 Decision Making in the Supreme Court and the Court of Appeals. . . .4 1.2 “Principal-Agent” Theory of the Hierarchy of Justice ................... 6 1.3 Principal Agent Theory ...................................................... 7 1.3A Shirking ...................................................................... 8 1.3B Problems of Monitoring ................................................... 9 1.4 A New Model of Agency .................................................. 10 1.4A Who is the Principal ...................................................... 11 1.48 The Enhanced Problem of Supervision ................................. 12 1.4C Influence of Agents on Agents .......................................... 12 1.4D The Role of History ....................................................... 13 1.5 The American Judicial System as an Hierarchy of Justice ............ 16 1.6 Applying Principal—Agent to the Judiciary .............................. 18 1.7 Impact and Compliance Work ............................................. 21 1.8 The Necessity of an Integrated Model .................................... 25 A Formal Model of the Relationship Between the Court of Appeals and the Supreme Court .......................................................... 27 2.1 Formalization of Judicial Principal-Agency ............................. 28 2.2 Compliance by the Lower Court .......................................... 31 2.3 Supreme Court Supervision ................................................ 32 2.4 Alternative Strategies at Work? ............................................ 34 Confession Cases in American Courts — The Data .......................... 40 3.1 Cases Involving Confession: Their History and an Argument for Their Use .......................................................................... 40 3.2 The History of Confession .................................................. 41 3.3 Defining the Universe ....................................................... 50 3.4 Fact Pattern Analysis ........................................................ 51 3.5 The Facts ...................................................................... 59 Appendix 3.A Coding Rules .................................................... 71 Decision Making in the United States Court of Appeals .................... 76 4.1 The Supreme Court in Confession Cases ................................. 79 4.2 Decision Making on the Courts of Appeals .............................. 83 4.3 Coding of the Variables ..................................................... 85 4.4 Results ......................................................................... 87 4.5 The Pooled Model ............................................................ 93 4.6 Differences Among Circuits ................................................ 95 4.7 Discussion ..................................................................... 96 vii Ch ch, Chapter 5: Chapter 6: 4.8 Conclusions ................................................................. 98 Appendix 4.A Diagnostics ................................................... 106 When the Supreme Court Speaks - An Analysis of Lower Court Response Treatment of Supreme Court Decisions .................................... 110 5.1 Vertical Stare Decisis ................................................... 111 5.2 Compliance and Impact in Judicial Scholarship ..................... 113 5.3 Supreme Court Cases Under Analysis ................................. 115 5.3.1 Townsend v. Sain 372 US. 293 (1963) .............................. 116 5.3.2 Bruton v. United States 391 US. 123 (1968) ....................... 119 5.4 Identification of Courts of Appeals Cases ............................. 123 5.5 Courts of Appeals’ Compliance with Bruton and Townsend ....... 125 Bruton 5.5.1 The First Circuit ........................................................ 126 5.5.2 The Second Circuit ..................................................... 128 5.5.3 The Fourth Circuit ...................................................... 130 5.5.4 The Fifth Circuit ........................................................ 131 5.5.5 The Sixth Circuit ........................................................ 133 5.5.6 The Seventh Circuit ..................................................... 134 5.5.7 The Eighth Circuit ...................................................... 135 5.5.8 The Ninth Circuit ....................................................... 136 5.5.9 The Tenth Circuit ....................................................... 138 5.5.10 The Eleventh Circuit .................................................. 139 5.5.11 The DC. Circuit ....................................................... 140 Townsend 5.5.12 The First Circuit ....................................................... 141 5.5.13 The Second Circuit .................................................... 142 5.5.14 The Third Circuit ...................................................... 142 5.5.15 The Fourth Circuit ..................................................... 142 5.5.16 The Fifth Circuit ....................................................... 143 5.5.17 The Sixth Circuit ....................................................... 144 5.5.18 The Seventh Circuit ................................................... 144 5.5.19 The Eighth Circuit ..................................................... 145 5.5.20 The Ninth Circuit ...................................................... 14S 5.5.21 The Tenth Circuit ...................................................... 145 5.5.22 The Eleventh Circuit ................................................... 146 5.6 Conclusions ................................................................. 146 Conclusion: Faithful Agents, Ideological Actors ......................... 151 6.1 Why Comply? ............................................................... 153 6.2 Decision Making in the Court of Appeals ............................... 154 6.3 But, Is It Complete Compliance? .......................................... 155 6.4 Different Courts = Different Justices? .................................... 156 6.5 Unanswered Questions, and Directions for Future Research ......... 157 viii Appendix A: From Judge to Justice — Institutional Effects on Judicial Decision Making Processes .............................................................. 161 Al The Judge Turned Justice ................................................ 163 A2 Exploring Decision Making and Opinion Writing from Judge to Justice ................................................................... 164 A3 Judge/Justice Burger ...................................................... 166 A4 Judge/Justice Blackmun .................................................. 170 A5 Judge/Justice Marshall .................................................... 172 A6 Conclusions and Speculations ............................................ 173 Table of Cases .................................................................................... 177 References ......................................................................................... 182 ix Ta! Ta! Tal Ta! Ta! Tat Tat Tat Tat Tat Tat Tab Tab LIST OF TABLES Table 3.1: West’s Key System — Criminal Law - Evidence ........................... 65 Table 3.2: Variables in the Involuntary Confession Cases ............................. 68 Table 3.3: Facts Used in Benesh Model .................................................. 69 Table 4.1: Decision Making on the Courts of Appeals ................................. 100 Table 4.1A: Marginal Effects (All Other Variables at Mean) ........................... 101 Table 4.2: Supreme Court and Courts of Appeals Decisions Pooled ................. 102 Table 4.2A: Marginal Effects (All Other Variables at Mean) ............................ 103 Table 4.3: Differences Among the Circuits .............................................. 104 Table 4.4: Frequencies of Circuit Exclusion of Confession ........................... 105 Table 5.1: Circuit Treatment of Bruton v. United States .............................. 149 Table 5.2: Circuit Treatment of Townsend v. Sain .................................... 150 Table A. 1: Judge to Justice Voting Behavior ............................................ 176 Table of Cases .................................................................................... 177 Fig Fig Fig Fig Figure 2.1: Figure 2.2: Figure 4.1: Figure 4.2: LIST OF FIGURES Utility Curves of the Courts of Appeals ................................. 33 Game Tree — Principal Agency in Action ............................... 37 Checking for Influential Cases ............................................ 108 Influence Analysis .......................................................... 109 xi INTRODUCTION In attempting to explain the ways in which judicial actors decide, the literature on judicial decision making is pervaded with adherents of either the legal model or the attitudinal model. Both tend to deny the operational relevance of the other, and a debate has ensued. However, this debate seems to center around the Supreme Court, and little work has been done to analyze decision-making tendencies of the Courts of Appeals to see if, in fact, the lower court judges decide cases in a way compatible with the legal model, the attitudinal model, or both. This dissertation aims to increase understanding of decision making in the Courts of Appeals and the relationship of that court to the Supreme Court, within a comparative and institutional framework. I will test both the attitudinal and legal models to determine which, or to what extent each, describes decision making in the Courts of Appeals. I assume that the lower court is constrained by its position within the judicial hierarchy, and, as a result, a revised, integrated model of decision making applies to these courts. I test a modified principal— agent theory of judicial hierarchy, therefore, which — I will Show — nicely captures the interdependence of the two institutions. Due to the differential institutional make up of the two courts, we expect that decisions and decision-making processes will necessarily differ between them. Here, I test the presumption that, while the Supreme Court decides basically in a manner consistent with its policy preferences, the Court of Appeals judge must take into account other factors and other actors the Supreme Court need not concern itself with. In order to judge the adequacy of a principal agent model of the judiciary, as well as gain some insight into the decision-making tendencies of the lower courts themselves, the project employs several different methodologies: (1) a formalization of the principal- agent relationship and calculation of expected utilities for certain behaviors; (2) probit models of decision making in the Courts of Appeals, including measures of the influence of Supreme Court precedent, Supreme Court ideological predisposition, and the policy preferences of the lower court judges, as well as the exploration of the possibility of differences among the circuits; (3) an analysis of lower court response to two non-salient Supreme Court decisions in the area of confession (Bruton v. United States and Townsend v. Sain) to determine whether the lower courts actually comply or employ any number of creative non-compliance techniques; and (4) an individual-level analysis of the behavior of the three Supreme Court justices who also served substantial time on the Courts of Appeals, examining the same decision makers in different institutional environments to see if changing the rules changes the behavior of the players. Using the forgoing methodologies and the theory of principals and agents, I will Show that decision making in the lower courts is mightily influenced by the Supreme Court, but that the interaction between the two actors is quite different from principal agency relationships that have been heretofore studied. The lower court heeds its principal not because of the direct supervisory function of the High Court, Since that direct an C0 cal supervision is the exception rather than the rule, but rather out of a combination of legal factors, including role perceptions, legal norms, respect for precedent, and lack of clarity on the part of the Supreme Court. In other words, because of the hierarchical set-up of the judiciary, the lower courts are constrained by High Court pronouncement, but that constraint really has very little to do with what the Supreme Court does but rather what the Supreme Court is. While notions of “role” and “respect” are not readily operationalized, by demonstrating widespread compliance with Supreme Court decisions and the major impact that institution has on the decision making of the lower courts, coupled with the nearly total lack of Supreme Court supervision, I believe the inference can be validly drawn, if not scientifically demonstrated. CC 50 t0 ti Ofa CHAPTER 1: THE HIERARCHY OF JUSTICE 1.1 DECISION MAKING IN THE SUPREME COURT AND THE COURTS OF APPEALS Two models have dominated recent discourse on Supreme Court decision making. These are the legal and the attitudinal models. The legal model of decision making, as described by Segal and Spaeth (1993), adheres to the belief that "...the decisions of the Court are based on the facts of the case in light Of the plain meaning Of statutes and the Constitution, the intent Of the framers, precedent, and a balancing Of societal interests" (32). The attitudinal model, on the other hand, claims that "...the decisions of the Court are based on the facts of the case in light Of the ideological attitudes and values Of the justices" (32). Because Of the distinguishing characteristics of the Supreme Court one may conclude that the Court of Appeals1 is different and therefore its judges may not rely solely on attitudes in their decision making. Segal and Spaeth address this possibility, and agree that a model Of lower court decision making may necessarily differ, because lower court judges: 1) do not have the broad control of their docket the Supreme Court enjoys; 2) may desire higher Ofl'rce, thus exposing themselves to political influences; and 3) are subject tO the superior Supreme Court, which has the authority to overturn their 1I use “Court Of Appeals” to refer tO the institution, “Courts Of Appeals” to refer to the collection of circuits, and Circuit Court tO refer to the individual circuit. It is duly noted that using “Court Of Appeals” is not the custom, but I believe it conveys a message Of a united institution, which is what I, in the end, argue the collection Of circuits actually is. 4 decisions.2 In addition, these courts have a much heavier workload, and they may be more constrained by precedent, given their position in the judicial hierarchy. We must, therefore construct another more integrated model; a goal to which this dissertation aspires. Why study the Court Of Appeals? After all, most Of the judicial literature to date explores any number Of minutiae pertaining to the processes and decision-making tendencies of the Supreme Court, thereby contributing to the notion of Supreme Court primacy. Perhaps one should continue that research agenda, finding another aspect Of Supreme Court decision making to more fully understand. While it is true that interesting questions remain at the High Court level, it is also the case that the lower federal courts (as well as the state courts, both appellate and trial) are increasingly becoming the focus of Study. Most decisions are in fact decided within these courts. The Supreme Court’s docket is Shrinking while the number of appeals to it grow each year (Carp and Stidham 1996). This means that the Courts Of Appeals (and State supreme courts) are almost always the last stop for justice; the Supreme Court formally decides less than 2% Of the cases appealed to it (Carp and Stidham 19962132). There is no question then, but that these courts are important institutions. It becomes imperative that we better understand how they make decisions. However, because of the institutional position Of these courts, we still need consider the influence of the Supreme Court on even those decisions in which it does not directly participate. The High Court is the supervisor Of the lower courts. The lower courts are not completely free agents. 2Segal and Spaeth also mention the fact that electoral sanctions exist for some lower court judges which do not exist for Supreme Court justices, but this does not apply 5 ad rel WE Co inc dec not are den that judh eXpe this 1 enufl deva Weci ”Don foiiOV I0 the 1.2 A “PRINCIPAL-AGENT” THEORY OF THE HIERARCHY OF JUSTICE I propose a variation Of the variation, if you will, of principal-agent theory advanced by Brehm and Gates in their book on bureaucracy, and assert that the relationship between the Courts Of Appeals and the United States Supreme Court can be well understood via such a theoretical construct. The High Court, by law, oversees the Courts of Appeals, and those lower courts are bound by the law as prescribed by that higher authority. This relationship, in turn, affects the means by which lower courts make decisions. While the assertion that the courts exist within a principal-agent relationship is not completely new, very little work has attempted to explain and rigorously test for such a relationship. An article by Songer, Segal and Cameron (1994) treats the topic and I derive much from their seminal work. The treatment here will be much more extensive than any previously undertaken, however. Before delving into the theory of principals and agents, we should note that the judicial hierarchy does inherently differ from other hierarchies, and therefore, we Should expect the judicial brand of principal agency to reflect this. Because of the character Of this judicial hierarchy, some scholars are quite reluctant tO grant worth to projects that employ this theoretical construct. However, that the theory does not fit exactly does not devalue its application completely. In fact, models are by design abstractions of reality; we do not expect reality to be totally represented within the model. Instead, we focus upon certain crucial factors, and glean as much fi'om the model as possible. So, while the following sections describe principal agent theory qua principal agent theory, those to the Courts of Appeals. pr cor ( l 9 exp. cont subs: This l Choict Idaho 0rgani Optima COnstra pnnClpa "Omestn' sections which proceed them explain the adaptations necessary tO meaningfully apply such a theory to the relationship between the Supreme Court and the Court of Appeals and, by extension, to the decision-making processes Of the Court Of Appeals judge, as well as describe the contribution such a model makes to the literature on lower court jurisprudence and our understanding of said decision making. 1.3 PRINCIPAL AGENT THEORY An agency relationship, according tO Eggertsson, “. . .is established when a principal delegates some rights. . .tO an agent who is bound by a (formal or informal) contract to represent the principal’s interests in return for payment of some kind” (1990240). Miller and Moe define the principal-agent model as “. . .an analytical expression of the agency relationship, in which one party, the principal, enters into a contractual agreement with another, the agent, in the expectation that the agent will subsequently choose actions that produce outcomes desired by the principal” (19862175). This relationship exists, then, within an hierarchical structure, which itselfis a social choice aimed at correcting a market failure. Miller and Moe argue that any such relationship will result in a “Sen paradox” (180). Sen’s paradox asserts that any organization which is decentralized (i.e. hierarchical) is subject to violations Of Pareto optimality or transitivity, or, in attempts to maintain those attributes, must resort to constraining individual preferences.3 In a hierarchical structure, actors may be both a principal and an agent Simultaneously, creating an interesting dynamic. 3 The desirable traits for an organization in the social choice literature include: 1) nonrestriction of individual preferences; 2) unanimity, or Pareto Optimality; 3) 7 The problem with the loyalty expectation of the principal - that is, the expectation of the principal that his will be done by his agents - then, is that it runs counter to the firndamental economic notions Of self-interest (Songer, Segal, and Cameron 1994). Therefore, problems Of enforcement arise. The agent, like the principal, has interests and the desire to act in accordance with those interests, unless subjected to substantial controls or inducements (Shepsle and Bonchek 1996). The difficulty, then, is the creation Of enforcement mechanisms or incentive structures that will effectively constrain the actions Of the agent in order to sustain an efficient equilibrium. Because those mechanisms will never be perfect, the possibility of Shirking always exists. 1.3A SHIRKING Because Of information asymmetry between the two, with the agent generally assumed to have more information than its principal, opportunity for the agent tO “Shirk” Often occurs (Eggertsson 1990). Because Of the costs Of monitoring, the principal must make decisions as to which aspects it will supervise, and in what ways the outputs will be measured. A point is reached at which the principal will no longer gain utility from independence Of irrelevant alternatives; and 4) transitivity (Miller and Moe, 1986: 178). Because these are such basic requirements for an organization, it may be surprising that the authors find that only in dictatorship are all Of these conditions met. By mentioning the fact that hierarchies cannot achieve all four conditions simultaneously, I attempt to make the case that individual preferences will be restricted in order for the judiciary to function. 01 ha an les Otl (E3 coll im't VOh Pro p60 mac (Es. 58kg Chec Solv Unac Choi. Fron that ' monitoring the agent, as the costs of monitoring will outweigh the benefits of compliance or some semblance thereof." 1.3B PROBLEMS OF MONITORING The high costs of monitoring discussed above give rise to the possibility Of moral hazards and adverse selection. A moral hazard, according to the principal agent literature, arises when the principal measures compliance by some single proxy or indicator, thereby lessening effort in monitoring. This measurement method may allow for Shirking in all other areas save the one which the agent knows will be monitored by the principal (Eggertsson 1990). TO illustrate: adverse selection occurs in the hiring Of the agents. If an employer relies only on one easily identifiable criterion for hiring, say, for example, a college degree, She is missing many other possible correlates of productivity, such as involvement in extra-curriculars or dedication to a certain cause as evidenced by volunteering. This adverse selection procedure may exacerbate the principal-agent problem by Starting out with agents who may be more likely to Shirk than like-qualified people.s " The total cost Of agency to the principal, then, is “. . .the sum of the investments made in limiting Shirking plus the costs associated with remaining or residual Shirking” (Eggertson 1990:44). 5 Some ways to deal with this agency relationship include StevenS’ “screening and selection Of agents, contract design,” “monitoring and reporting,” and “institutional checks” (Stevens 1993 :284-287). Screening Of agents during the selection process can solve the adverse selection problem, and contract design can make certain behaviors unacceptable. Monitoring can take place via police patrol or fire alarm type oversight, the choice between which reflects the cost a principal will accept in monitoring its agents. From the names of these two types of principal oversight, one Should be able to ascertain that the police-patrol type is much more costly than the fire-alarm oversight type (Shepsle 9 7, “ 1.4 con the that burr \Nas Shh; 1.4 A NEW MODEL OF AGENCY Brehm and Gates, in their 1997 book on the bureaucracy, Offer a useful permutation Of the principal—agent model. They contend that prior theories of principal- agent relationships, or agency theories, are flawed in their assumption that agents will firndamentally Shirk Simply because they prefer Shirking tO working. They explicitly contend that agents will Shirk only when they do not agree with the policy preferences of the principal. While this hardly seems a bold assertion, the literature in this area, especially that on the bureaucracy, never explicitly makes this point, preferring instead to paint the bureaucracy negatively. They assume that bureaucrats would much rather sit around and waste time than to actually get things done. Brehm and Gates provide evidence that this is Simply not true. Their theory has Obvious implications to this study. Brehm and Gates focus on the problem Of adverse selection, for they posit that for some bureacrats, compliance is the norm. The problem, then, comes in finding enough Of these individuals. If that is accomplished, then the agency problem loses its bite. In describing those agents who do comply with their principal, Brehm and Gates assert that they do so because they share the preferences of that principal. Because previous work focuses on working and Shirking, they fail to account for the “. . .positive efforts of policy- oriented bureaucrats, social workers with guiding agendas, police Officers who honestly prefer to enforce the law, and professors who like tO teach” (Brehm and Gates 1997219). and Bonchek 1996). Stevens discusses the ability Of other institutions to play a role in the effectuation Of policy as a solution to this agency problem. 10 In 3“ (“v obj tov 0U! COP. pre 1.4. con th enac SUpe Burg mug! Final and ( In order to make their point, they first identify the possible policy output decisions available to the bureaucrat: working, Shirking, or sabotage. “Some bureaucrats devote extraordinary effort toward accomplishing policy (“work”), where others may expend as much effort deliberately undermining the policy Objectives of their superiors (“sabotage”). Other bureaucrats may be directing efforts toward non-policy goals (“Shirking”)” (Brehm and Gates 1997221). In all three policy outputs, the main concern is what the policy goals are, and whether bureacrats’ actions are consistent with the preferences Of the public, which should be approximated by the policy I I preferences Of their principal. 1.4A WHO IS THE PRINCIPAL Whether or not agents follow their principal, however, is not the only consideration Brehm and Gates consider. Another important question involves exactly who does the controlling. Bureaucrats, according to Brehm and Gates, are expected to enact the policy preferences Of four different entities. The first is, of course, their supervisors or managers. The second policy preference which matters is their own. Bureaucrats Obviously are influenced by their own policy preferences, and SO this influence must be included. Bureaucrats also respond to one another, which is the third entity. Finally, bureaucrats respond directly to the public, the customers they actually serve (21). Supervision, then, becomes very important in any discussion of agency. Brehm and Gates address the problems in the earlier literature as they attempt to add to the assumptions Of the principal-agent model the possibility that Shirking can take a number of 11 th (I! 1.4 CD) (‘2 or h info: and '. resor Very . agent PTOblt MC 1 bureau, p rOPOUI forms; that supervision is costly; and that agents’ preferences differ across issues, thereby making the propensity to comply differ across issues. “This model,” they say, “is an enhanced principal-agent game emphasizing the choice of actions available to agents as they produce output and the constraints limiting a supervisor’s authority over bureaucrats (the agents)” (1997:27). 1.4]! THE ENHANCED PROBLEM OF SUPERVISION Brehm and Gates spend some time on the supervisory problem, highlighting the fact that the enhanced principal-agent model supervisor is oftentimes unable tO control his or her agent because Of the many constraints S/he must face. There is the limited information constraint, the limited resources constraint, rules that constrain the supervisor, and the additional fact that the supervisor tOO has preferences and will allocate more resources to favored policy outputs. Therefore, both monitoring and sanctioning become very diflicult. However, in this enhanced model, the preferences of the principal and the agent are the single best predictor of agency compliance. That being the case, supervisory problems may not be such an ominous concern. 1.4C INFLUENCE OF AGENTS ON AGENTS Besides the supervisor, as mentioned earlier, other actors may constrain the behavior of the bureaucrat, or at least influence it. One set of such actors is other bureaucrats, borne out in the creation of standard Operating procedures. Brehm and Gates propound that, in order to better understand compliance, we must understand not only enforcement mechanisms and inducements, but also the history of the agency and the 12 do. how seen it see and it on the 1.4D] as a CC aClOr u Situatio Compile prObIEn PFIDCipa different may be C culture thereof, wherein . .workers learn fiom one another about appropriate behavior” (1997:48). They speak about more “peripher ” routes tO compliance. That is, instead of a focus on the supervisory role of the principal, they focus on other influences on the agent which may also, in fact, induce compliance. Under conditions of uncertainty it seems plausible to assert, as Brehm and Gates do, that people will look to Others who are more expert than they in order tO determine how they should act in a certain Situation or react to a given Stimulus. Applying that seemingly commonsensical notion, they extend the principal agent model, for the first time it seems, to include other influences on agency compliance. These agents have networks and interact regularly with one another. That interaction is bound to have some influence on their behavior, and SO must be included in our models Of principal-agency. 1.4D THE ROLE OF HISTORY In addition to their influence on each other, Brehm and Gates also identify history as a constraint on agent behavior. This peripheral route to compliance asserts that an actor will lOOk to his or her past behavior as a determinant of how to react to the current situation. Therefore, the first behavior of the individual and whether or not that behavior complied with the preferences and directives of the principal is most important. The problem becomes exacerbated then when changes in personnel occur; where a liberal principal with a given set of policy goals replaces a conservative principal with completely different sets of policy goals. In this Situation, an agent complying with the first principal may be out Of compliance with the second though performing identical tasks. 13 In sum, Brehm and Gates make the following assertions (pp. 71-74): 1. The more individual bureaucrats lOOk to fellow bureaucrats for information about appropriate behavior, the more likely it is that the bureaucrats will be in a state Of conformity. 2. The greater the sense of uncertainty a bureaucrat feels about appropriate behavior, the more likely that bureaucrat will look toward others, which in turn leads to greater conformity. 3. The greater the frequency of contact among subordinates, the greater the degree Of confomrity we would expect to see in their behavior. 4. Whether or not a bureaucracy, in general, tends tO comply with political supervisors depends primarily on the policy predispositions Of the bureaucrats. 5. Learning by imitation might lead to extremism, not to middle-Of-the-road behavior. . .Leaming by imitation can lead to extreme levels Of compliance as well as extreme levels of noncompliance. . .If subordinates adopt SOPS [standard Operating procedures] on the basis Of learning by imitation, their supervisors are but feeble influences on the ultimate compliance level Of their bureaucracies. This particular model of agency is particularly suited tO a study Of the courts. First of all, the concept that the agent will work, Shirk, or sabotage according to his or her policy preferences and the degree to which those preferences coincide with those Of the principal is essential in such a study. Since judicial decision making is generally agreed to be motivated by the policy preferences Of the judicial actors, it is reasonable that the lower court judges would indeed more likely follow their Supreme Court principal when their preferences coincide with the Court’s decisions. Secondly, the notion that there are peripheral influences on the decision over whether or not to follow the principal is very useful to understanding the behavior Of the lower courts. History, for example, in the form of adherence to precedent, or stare decisis, is at least given lip service by all lower court judges, SO it Should be included in a model which aims to detemrine the level of compliance by lower court judges to the policy prescriptions Of the Supreme Court. 14 Fun \yeH Chet the! fins (Sou appe perfi cour ifinal pnnc there Presc under devi c ETEau under uinl a Finally, the notion of inter-agent influence is one that we could apply to the judiciary as well. It seems plausible that different circuits Of the Courts of Appeals look to the other circuits for guidance in how to behave, possibly measuring by their actions how to react to the Supreme Court, and thereby whether or not they should heed their principal, although this notion has never been explicitly tested. In addition, the enhanced problem of supervision also nicely explains the Supreme Court’s position. The Court has resources limiting it to hear only a tiny fraction of all appeals coming to it. The High Court also has somewhat limited information on the performance of its subordinates, as it cannot plausibly know about every decision in every court. It knows only those appealed to it, and of those it can only choose to hear a few. Finally, the influence Of Supreme Court preferences comes to bear in its fitnctioning as principal. Surely the justices care about certain policy areas more than they do others and, therefore, tend to hear those cases, making certain that the agents heed their policy prescriptions. This leaves a great number of potentially important issue areas underaddressed by the Court. Its success as principal, then, depends quite heavily on devices other than direct supervision. The next section explores the fit Of principal agent theory to the judicial system in greater detail, highlighting instances where the theory proves to be quite helpfirl in understanding the relationship between the Supreme Court and the Courts of Appeals, as well as the influence that relationship has on decision making in the inferior court. 15 maj its 2 prin adrr half are : 0PP< desir on ti "in ii; 1.5 THE AMERICAN JUDICIAL SYSTEM AS AN HIERARCHY OF JUSTICE Studying the relationship between the Supreme Court and the Courts of Appeals as a hierarchical one seems both usefirl and plausible. However, as mentioned earlier, a major element in such a relationship is the ability of the principal to control the behavior of its agents by constraining them to make decisions in keeping with the interests of the principal. For the Supreme Court, this function, while essential to the efficient administration of uniform justice, is extremely difficult to perform. Since less than one- half of all Courts of Appeals cases are ever appealed to the Supreme Court and fewer still are actually heard, it Should be easy to see that the Courts Of Appeals judges have ample Opportunity to Shirk, or make decisions according to their own attitudes. However, the desire to avoid reversal is great, so there must be at least some level of constraint placed on the decision making of these lower courts to assure that they remain at least somewhat “in line” with the decision making of the higher court. In a Study designed to test such a principal-agent relationship between the Supreme Court and the Court of Appeals, Songer, Segal and Cameron (1994) use search and seizure cases, control for facts, and perform statistical tests to determine the extent to which lower court judges decide according to their attitudes, and the extent to which they regularly follow precedent set for them by their principal. They find that, while the Courts of Appeals are highly responsive6 to the changing search and seizure policies of the 6 Responsiveness Should be here distinguished from congruence, as responsiveness implies that as the Supreme Court modifies its doctrine, so the appeals court modifies its doctrine in the same direction. Congruence, on the other hand, implies that an appeals court and the Supreme Court, given the same facts, would decide the case in the same 16 Supn finer: Oftht fins; ofthe by?“ COUH “Ugh adnus efiecr Preset lndlca lUdges IeViett High ( Pllhlly ’5 more IOU” ll annude Supreme Court, they also find opportunities to Shirk in order to satisfy their own policy interests. They include variables measuring ideological changes on the Court, preferences of the lower court judge (utilizing an inferential measure with which there seemingly are a few problems, but none large enough to negate the importance of this variable), the facts of the cases themselves and their effect upon the decision reached. Their model performs very well, predicting almost 90% of all cases, reducing error by 22%. They find a high degree of congruence with Supreme Court doctrine in the lower court. This congruence is not perfect, but is much closer to being so at both extremes (high probability of exclusion of evidence obtained in the search v. high probability of admission Of evidence Obtained in the search). Secondly, they find that changing Supreme Court policy does have a substantial effect on the decisions of the lower court. Therefore, responsiveness is also seemingly present, even though lower court judges also follow their own policy preferences, thereby indicating an effect of Supreme Court policy and ideology, as well as the lower court judges’ own ideology. They also find evidence for their contention that those cases not reviewed by the Supreme Court are those most likely to have followed the ideology Of the High Court, thereby evidencing the influence Of Supreme Court monitoring. The authors pithily summarize: way. Congruence is time dependent and specific to a given decision, while responsiveness is more a measure of trends and the propensity of the lower court to follow the Supreme Court ideologically. The former leaves minimal room for the lower court to exercise attitudes, while the latter does afford some leeway. l7 appe becat pnnc dose fbrcc achfii chcui Supre theon hlPOt 1.6.x rdafic be dea 0theri tharh appr0p As in the metaphor of the dog on the leash, some appeals court panels led and some followed "the owner," but when the Supreme Court tugged on the leash, both liberal and conservative panels were responsive (688). While this study did in fact spark the interest in testing such a model of federal appeals court interaction, my Study purports to add much to their formulation. First, because the treatment here is much more inclusive, I am able to test for the existence Of a principal-agent relationship using many different methodologies, thereby coming much closer to proving that the two courts share this type of relationship. I do employ the tests for congruence and responsiveness conducted by Songer et al., adding to their formulation a different Slant on the relationship, tests for relationships and differences among the circuits, individual-level analysis of decision making, and additional qualitative evidence of Supreme Court constraints on lower court decision making. Hence, I am able to test the theory from several angles, thereby increasing confidence in any results favorable to the hypothesis of High Court influence on lower court decision making. 1.6 APPLYING PRINCIPAL-AGENT TO THE JUDICIARY While I have discussed principal-agent theory, and explained the ways in which this relationship constrains judicial decision making at the lower court levels, one last issue to be dealt with is the actual fit of the Court Of Appeals/Supreme Court relationship. In other words, is the relationship really one of a principal and its agents, or is it so different that it needs to be understood in some other way? In order to examine the question of appropriateness of the model to the judiciary, some important aspects of the model discussed both by Shepsle and Bonchek (1996) and by Miller and Moe (1987) will be 18 CXE THO rne diff han (:01 (for Ina} posi non: then ad\f does SEVEI asah knee; Ihe In examined, which demonstrate the unique relationship of the judicial hierarchy to this model. In discussing bureaucratic drift, Shepsle and Bonchek list some possible control mechanisms that may be employed to deal with non-compliance, which calls to mind differences in the relationship between the Supreme Court and the lower courts on the one hand, and the bureaucracy-legislature on the other. In attempts to control actions Of the Courts of Appeals, some mechanisms appear that are simply not available to the Supreme Court. First of all, the appointment process is mentioned as a means by which a principal may control its agent, for it can thereby appoint or confirm those with similar policy positions. This possibility is not available to the Supreme Court, as the President nominates (and the Senate confirms) both the Supreme Court justices themselves, and their lower court counterparts. Therefore, the Supreme Court does not deal with the adverse selection problem that Miller and Moe point out as problems Of monitoring, but does deal with the even larger problem of trying to control agents the Court was never given the Opportunity to hire (or fire).7 Secondly, Shepsle and Bonchek discuss the creation of procedural controls, and again the Supreme Court is at a disadvantage in attempts to constrain the behavior of its lower court agents. In this case, it is the legislature which makes procedural changes in 7 The moral hazard problem does continue to plague the Supreme Court, however, as its attempt to oversee the lower courts is limited to those cases which are appealed by the litigants. This leaves most jurisprudence outside the review of the Court. This will be discussed further. 19 “'3 and QVI I3or alar insfi nehl cong alOV junSd, Coukj as ‘Trtl the administration of the judiciary when it sees fit, not the Supreme Court. Even if it wanted to, the Court could not itself change the rules of the game.8 In those instances in which the Supreme Court does exercise its monitoring duties and reviews decisions of the lower court, one can again see that the Court is constrained by its very nature from the enforcement capability of some other institutions. Shepsle and Bonchek (1996) describe the different oversight Strategies of “police-patrol” and “fire- alarrn” surveillance, mentioned earlier (see footnote 4). The Supreme Court is institutionally constrained to perform only the “fire-alarm” type surveillance, as it has neither the time nor the ability to patrol the lower courts to make certain that decisions are congruent with the decisional mandates of the Supreme Court. Instead, it must wait until a lower court makes a decision that one of the litigants deems to be worth the cost of appealing to the Supreme Court, and which the Court moreover, deems important enough to review. On the other hand, given the large number of cases appealed to the Court, one could assume that police patrol surveillance occurs as the Court wades through petitions arguably representative of all disputes and hears and decides those of most significance.9 The Court restricts the individual preferences of the lower court judges by overturning their decisions and resolving conflicts among the different circuits, but 8 Here, though, one may find cases in which the Court extends or restricts jurisdiction. However, these extensions or changes are still subject to legislation that could reverse them, Since Congress determines all of the lower courts’ jurisdiction, as well as virtually all of the Supreme Court’s. 9 The author thanks Dave Rohde for pointing this out. 20 enft for sho the) COL (”let cour alwa 1.7] must Appc itself APpe decisi their r 1997; prObab COW-Ii! Cages d rec02m to coin} enforcement mechanisms are not systematic. This limitation provides more Opportunity for Shirking. Lower court judges no longer merely ask themselves whether or not they should decide cases a certain way, but rather whether they should decide them the way they want, even when that preference runs counter to the policy stance of the Supreme Court, and thereby risk possible reversal. More Often than not that probability of review (let alone reversal) is close to zero.lo On paper, it is easy to provide reasons for lower court avoidance of Supreme Court policy prescription. However, in reality, they almost always follow High Court pronouncements. 1.7 IMPACT AND COMPLIANCE WORK In order to apply to the judiciary this idea of a principal-agent relationship, we must first determine whether the Supreme Court exerts any influence on the Courts Of Appeals. As mentioned earlier, the lower court seemingly has little incentive to concern itself with possible disciplinary action by the Supreme Court, given the rarity of review. Still, the decision of the Court Of Appeals jurist is not an easy one. Court of Appeals judges are subject to influences other than their own policy preferences in their decision making. Several judicial scholars have attempted to explain those influences and their relative effect on decision making in the Courts of Appeals (Reddick 1997; Benesh 1997; Songer, Segal and Cameron 1994; Sheehan, Mishler and Songer 1992; Songer and ‘0 It would be interesting, however, tO determine in which types Of cases the probability of review and reversal is highest, and then determine whether or not the lower court judges perceive these differences. If so, and Shirking on those likely to be reviewed cases declined, support for the principal-agent theory could be shown, while still recognizing that the Court does not provide substantial inducements to get the lower court to comply with its preferences. 21 lian 197i Supt ivec pnnt 1972 inter tflace theo: (:OUI John more does (Brel Haire 1992; Davis and Songer 1989; Johnson and Canon 1984; Howard 1981; Goldman 1975; Richardson and Vines 1970; Wasby 1970; Goldman 1966). To what extent does Supreme Court policy control these judges? If this question can be adequately answered, we can begin to understand better the decision making of these judges; here, via use Of principal agency. No longer can it be said that the High Court’s decision is the final decision (Wasby 1972). Lower courts exert some semblance Of power over the Supreme Court in their interpretation and implementation of its decisions. However, mechanisms are arguably in place to undermine substantial deviations from the Court’s prescribed policy, at least theoretically. Largely, though, lower courts will interpret cases consistently with Supreme Court intent only when they agree with that policy (or are indifferent to it), according to Johnson and Canon.11 Therefore, the principal-agent model of Brehm and Gates may more aptly apply to the courtS’ hierarchical structure: the lower court will Shirk when it doesn’t like the policy, work when it does, and sabotage the policy when it really hates it (Brehm and Gates l997).'2 11Of course, this could mean a substantial broadening of the doctrine enunciated by the Court as well, but I would consider this as following the Court enthusiastically, contrary to Reid’s conclusions (1987). 1”This notion of hierarchical theory, as well as that espoused by Songer, Segal and Cameron (1994), or the organizational model advocated by Baum (1976), is not the same as the oft-criticized earlier hierarchy theory as mentioned, for example, in Richardson and Vines (1970). The earlier work was far more rigid in its prescription of the supremacy of the High Court than is the current model. 22 I3r lox (30 of: juri cul and thei disr fOr 0rd: that mytl aCCU COur to CO Probl themS the la Johnson and Canon offer reasons for compliance by the lower court, akin to what Brehm and Gates find for the bureaucracy. First, the possibility of reversal deters the lower courts from Shirking. While not many such reversals actually occur due to the Court’s shrinking docket,l3 Baum finds that stigma attaches to one’s reversal rate (Baum 1978; see also Baum 1976; Caminker 1994). Therefore, our concern about a seeming lack of monitoring by the principal may be overridden by the perspective the Court Of Appeals jurist takes regarding her job performance. In addition, judges are socialized into the legal culture, thereby having certain role perceptions (Tarr 1977), and take a pledge to defend and uphold the Constitution, which Should count for something (Johnson and Canon 1984). They also have an interest in maintaining the appearance of integrity in order that their decisions be accorded adequate respect, which would not occur if they blatantly disregarded Supreme Court policy. Or, judges could be acting solely out of self-interest, for in citing another’s reasoning one does not have to reason oneself (Caminker 1994)." However, a lower court needn’t completely thwart Supreme Court precedent in order to influence its impact. In fact, the lower courts have several Options available to ’3 Richardson and Vines (1970) offer a cute illustration: “The popular sentiment that one will appeal a case ‘all the way to the Supreme Court’ is somewhat of a popular myth to the extent that it symbolized the court’s appellate function. A more realistic and accurate statement would be the assertion that the case will be appealed ‘all the way tO the Court of Appeals’” (149). 1"Several Studies add communication or knowledge of decisions as a prerequisite to compliance (Johnson and Canon 1984; Tarr 1977). The Courts of Appeals have no problem in this regard. Either the judges keep up with the decisions of the High Court themselves or they are briefed about them by the lawyers who appear before them or by the law clerks with which they are provided (Tart 1977 ; Johnson and Canon 1984). 23 on “1' exi prec as g fact, aPPel APPe PTOXir 1990; still 10‘ are Tel; “eWsm compile High C0. them. They can interpret the decision narrowly, basically limiting it to its Specific facts (Johnson and Canon 1984). They may cite their own Opinions in lieu of citing the Offending precedent (Manwaring 1972). They may attempt to distinguish their case from the one for which Supreme Court prescription is available (Caminker 1994; Songer and Sheehan 1990; Baum 1978; Tarr 1977; Manwaring 1972). They may dispose of the case on procedural grounds (Johnson and Canon 1984); they may criticize the Supreme Court while following it (Tarr 1977); or they may simply ignore the ofl‘ending precedent’s existence. One might expect that trial courts will be more successful in ignoring or defying precedent than will appellate courts, as the trial court judge’s professional scrutiny is not as great as, for example, that of the Court of Appeals judge (Pacelle and Baum 1992). In fact, little evidence of outright defiance has ever been found in the intermediate federal appellate courts (Songer and Sheehan 1990). ” Many scholars posit that the Court of Appeals will be the most likely to comply with Supreme Court decisions, both due to its proximity to the Supreme Court, and its level of professionalism (Songer and Sheehan 1990; Johnson and Canon 1984; Gruhl 1980; Canon 1973; Murphy 1959). Of course, we still look to explain noncompliance, perhaps for the reason cited in Gruhl. “Just as there are relatively few stories about dogs biting people. . .but people who bite dogs are newsworthy, so it iS that noncompliance with court rulings gets more attention than compliance, even though the latter may be more frequent” (19802504). 1’It seems, though, that the DC. circuit has had its Share of skirmishes with the High Court (Belcaster 1992; Murphy 1959). 24 1.8 COT SOIl moc neet mak mod jurist com; integi Georg integr; the infl PerCEpr lO adhel Offshoor f("mil/all 1.8 THE NECESSITY OF AN INTEGRATED MODEL We are lefi then with the idea that first, attitudinal theory may travel to the Court Of Appeals, although the use of attitudes will be tempered by the peculiar institutional constraints faced by the Courts of Appeals. These constraints include a heavy, albeit somewhat difficult to understand, influence of the Supreme Court. TO understand decision making in the Court Of Appeals, then, we must integrate the legal and attitudinal models discussed earlier. Unlike the Supreme Court, more than just preferred policy needs to be considered. There is, therefore, no Simple and parsimonious model of decision making in these more complicated courts (at least not as Simple and parsimonious as the model that explains and predicts Supreme Court decision making). The Court of Appeals jurist takes several things into account, and to model her decision, we must look to a more complex formulation. Several scholars have quite convincingly argued for such an integrated model of judicial behavior (Songer and Haire 1992, Hall and Brace 1992, George and Epstein 1992). Therefore, I offer such. My variation of principal agency includes all the necessary components of an integrated model of decision making. First, it incorporates the role of law, allowing for the influence of both vertical and horizontal precedent (and the clarity thereof), role perceptions, and legal norms (which manifest themselves in a seemingly irrational tendency to adhere to Supreme Court precedent). While doing so, however, this principal agency offshoot does not forget the importance of attitudinal theory, incorporating into its formulation the interagreement between the two courts, and the preferences of the judges 25 on each. followed methodol on each. Therefore, this model offers a comprehensive representation of the process followed by lower court judges. The following chapters test this model using a variety of methodologies. 26 W: Court pret lower cour means dire desire to b assumprim f0111121! the: between t! power of \ Proceed in Shirking in exlittered u Pllncipa] a! game, COm them—y, COP, CHAPTER 2: A FORMAL MODEL OF THE RELATIONSHIP BETWEEN THE COURT OF APPEALS AND THE SUPREME COURT We are left, after Chapter 1, wondering why it is that lower courts follow Supreme Court precedent, and how we can model any influence the High Court might exert on lower court decision making, granting the possibility that that influence is not by any means direct. Formal theory Offers us an attractive way to consider any relationship we desire to better understand, and it forces the researcher to consider scrupulously the assumptions he or She is making in studying a given relationship. Therefore, I employ formal theory to examine this principal agency to see if, by forrnalizing the relationship between the Court of Appeals and the Supreme Court, we can glean some explanatory power of which we can make good use in the methodological chapters that follow. I proceed in this chapter to examine the expected utility for the Courts of Appeals of Shirking in preference to working (not following in preference to following), as well as the expected utility to the Supreme Court in supervising them, SO that I may begin to put these principal agency puzzle pieces together. In so doing, I formulate a judicial principal-agent game, complete with payoffs, to determine whether and when the lower court Should, in theory, comply with Supreme Court policy pronouncements, or whether, in fact, that it Shouldn’t. 27 2.1 F ORM Ca are most li utilitiesl7 f Court. Th that shoulc to sanctior modificatic Firs leisure is in goal (Cg, l the Operatit behafior as its “’Orkloa only Way in some Proce 2.1 FORMALIZATION OF JUDICIAL PRINCIPAL-AGENCY Calculating expected utilities Sheds some light on the choices that actors in a game are most likely to make. Here, I present inequalitiesl6 that must hold within expected utilities” for a given action to be taken by the lower court, and then by the Supreme Court. Through this exercise, we are able to exposit more clearly exactly those factors that should influence the decision by the lower court to comply, and by the Supreme Court to sanction or review. I use Brehm and Gates’ principal-agent game, with a couple of modifications necessitated by my subject. First, I do not posit a role for leisure Shirking by the lower court (although such leisure is included in the payoffs calculated later)”. While I think it plausible that such a goal (e. g., leisure) exists among lower court judges, as is the case for most human beings, the operationalization of this concept is firzzy. One could plausibly define leisure Shirking behavior as the court’s desire to rid itself of as many cases as possible, thereby decreasing its workload. However, the lower courts have an almost fitlly mandatory docket, SO the only way in which a lower court panel could “get out” of hearing a given case is to find some procedural grounds on which to dismiss it. However, I argue that finding ’6 The ordering of the actors’ preferences. ’7 “The sum across all possible outcomes of a decider’s utility for each possible outcome times the probability of that outcome’s occurring if a given action is chosen” (Morrow 19942350). 1’ Recall that “leisure shirking” refers to not working for the sole purpose of increased leisure; e. g., the preference of doing nothing to doing work in accord with the principal’s action. 28 proceduralg Supreme Cc ruherthanl uetoaccun innhmhthe already give meophuon. area. I declil rids itself of The court. dissei Precedent. j the lower cc this as Great COun prece Case from ii allhOUgh 811' 19 L( AC'C'Olding t MCCham’ .t l’lTIUally a. involve inert pOIIIOn 0f 11 procedural grounds for dismissal expends more effort than simply following whatever Supreme Court precedent is on point; hence the lower court would be making more work rather than finding a way to get a much needed break”. In fact, I would posit that if we are to accurately measure leisure Shirking, such Shirking would be found in the very cases in which the lower court follows Supreme Court precedent. After all, if the higher court already gives the answer, the lower court opinion author need not expend much energy in the opinion draft or legal reasoning. Because of these complications in the leisure Shirking area, I decline to include such a consideration, including those times where the lower court ridS itself of a case it doesn’t like on procedural grounds as a form of dissent Shirking.20 The second modification of the Brehm and Gates formulation is that, for the lower court, dissent Shirking and sabotage only differ in degree. Both avoid Supreme Court precedent. However, I define dissent Shirking as a ‘Sneakier’ sort of behavior, in which the lower court somehow creatively avoids the Supreme Court’s prescription. I refer to this as creative noncompliance. If the lower court wishes to avoid following Supreme Court precedent, it has several options Open to it. First, it may distinguish the facts of its case from the applicable precedent. This is perhaps the most popular form of defiance, although situations certainly exist where a lower court properly distinguishes its case from ’9 Lower courts are increasingly falling behind due to an overwhelming workload. According to the Director of the Administrative Office Of the U. S. Courts, Ralph Mecham, “The demands placed on United States judges today are staggering. Jurists at virtually every level of our federal system are facing a greater number of cases, which involve increasingly complex issues, explore novel areas of the law, and consume a larger portion Of their time” (Redmond 1996: 1). 29 the Supre language the Court types of c the lower altogether Court, dec Rith the S Supreme ( TEgardless Court‘s prl Court auth Control iha Let decision be diCA, the pr: SUpreme C‘ expended b‘ 20 - e dlfTermites b the Supreme Court precedent. Second, the lower court may declare the offending language in the Supreme Court case to be mere dicta, in which case it is not law. Third, the Court of Appeals may limit the applicability of the Supreme Court precedent to certain types of cases only, being sure to Show why its case is not one of those cases. And finally, the lower court can dispose of the case on procedural grounds, avoiding the merits altogether. All of these can be dissent Shirking as the lower court defies the Supreme Court, deciding the case as it sees fit. However, when the lower court really disagrees with the Supreme Court and wishes to tell it SO, the lower court may Openly defy the Supreme Court. This I label sabotage, as the lower court is making its own policy regardless of its principal’s earlier words on the subject. No longer is the Supreme Court’s prescription the law in that case. Rather, the lower court has thwarted Supreme Court authority and made law itself. This indeed sabotages Supreme Court efforts to control that particular area of law. Let the probability of the Court of Appeals (CA) following the High Court decision be fiCA (i denoting the individual circuit), the propensity for dissent Shirking be dim, the probability of sabotage be SiCA. and the influence of agent on agent be SOPicA. Supreme Court (SC) impact, then, will be a function of these probabilities and the effort expended by the Court for lower court supervision: xi = FICA(fiCA, diCA, SiCA, SOPICA, SSC, on, 5i) (1) 2° Recall that “dissent shirking” refers to avoidance of work due to ideological differences between the principal and the agent. 30 where X. is expended ir litigants bill of the deeis compliance miscommur 2.2 COMPI The OI‘ \\ hen tl outweigh t dissent Shi lorlUSI as 50me “3).. Th C0511}. f0r ' Fir where X; is a the implementation of a given Supreme Court decision, Ssc is the efi'ort expended in supervising decision making in the lower courts, on, is the probability of the litigants bringing an appeal, and 6; includes external factors involved in the implementation of the decision that cannot be controlled by the Supreme Court (e. g., trial court non- compliance, other governmental actors’ inaction in the implementation of decisions, miscommunication of the decision to governmental agencies, etc). 2.2 COMPLIANCE BY THE LOWER COURTS The Court of Appeals will follow the decision when 8X: / dfim > 0 implying, then, that (2a) 8X; / ad,“ 5 0 or (2b) 8X; / 65,01 5 O or both (2c) or, when the marginal utility of following the decision is greater than zero; e. g., benefits outweigh the costs in following the decision. Stated differently, the marginal utility Of dissent Shirking or of sabotage is less than or equal to zero; the lower court is better off (or just as well Off) following the Supreme Court prescription than trying to avoid it in some way. The Court of Appeals will sabotage when 6X; / as,“ > 0 implying that (3a) (3X; / afiCA < 0 (3b) or, when the marginal utility of following the decision is less than zero; e. g., it is less costly for the agent to work against the policy than it is for the agent to work for it. Finally, the Court of Appeals will dissent when 8X, / ad,“ > 0 and 8X, / arm < O (4) 31 that is. tl the decis I (recall X, RCA 15 III Court. I Where a, (probabilj TI policies 3 Utility for disliked p L3SUPR A: bm “l” at CCU” dECi COHSider tl that is, the agent gains more by Shirking (creatively avoiding) than he gains by following the decision. The utility firnction for the Court of Appeals will then be UCA = FIXi, (’PiCA)] (5) (recall Xi’s function in (1)) where X, is compliance with the Supreme Court decision, and Pm, is the punishment associated with noncompliance; namely, reversal by the Supreme Court. That punishment, then, is a function of the following: PICA z F(SSC, 0!, B, Y, 3) (6) where or, again, is the probability of appeal, B the probability of Supreme Court hearing (probability of granting certiorari), and y is the probability of High Court aflirmance. The lower courts are assumed to have negative utility for compliance with disliked policies and for noncompliance (sabotage or dissent) with liked policies, and positive utility for compliance with liked policies or noncompliance (sabotage or dissent) with disliked policies, taking the risk of punishment into account. 2.3 SUPREME COURT SUPERVISION As (6) Shows, the Supreme Court will not automatically supervise the lower court, but will attempt to maximize its utility, reviewing only the minimum number of lower court decisions in order to exact compliance from the lower courts. First though, we must consider the effect of SC supervision on lower court compliance. afiCA / 355C = 11(fiCA — {iCA) (7) 32 where {ICA l to be 1). an: L’sir worth of su which simp and further This amoU' under Stud Cllll'lllllSha a C0mPlianC. F where f'iCA is the propensity to comply that the SC would prefer (which may be assumed to be 1), and n is a constant describing how susceptible each CA is to SC supervision. Using this proposed effect of supervision, we can define the SC calculation of the worth of supervision as follows: fiCAISSC = fiCA(alone) — (faca— fiCA(alone))(l-e"‘ssc) (8) which simplifies tO fiCAlSSC = FICA - (f'iCA — fiCA(3]0fle)) Cmssc (9a) and firrther ECAISSC = 1 — (1 — ECA(alone)) e'“ssc (9b) This amounts to a utility curve for SC supervision which Shows that, whether the CA under study is highly likely to follow given supervision or not, the effects Of supervision diminish, assuming of course, the supervision leads to a higher likelihood Of lower court compliance. FIGURE 2.1: UTILITY CURVES OF THE COURTS OF APPEALS ———’____————_—— Workdes'redbySupCt Responsive Recalcilrant Work by CotA wtou Supervision ———_————_———_— 33 0f coursi decision decision 2.4 ALT! mmmen all cases here in tl which ha socializar inequalitj Pronounr iIlczorrecr set of he vertical ; Obsen-at 355111116 1 (Figure 1 the Com numbers Of course, SC supervision may depend also upon the value it places on the particular decision currently up for compliance, and as the game progresses, implementation of the decision will be less desirable due to the discount factor. 2.4 ALTERNATIVE STRATEGIES AT WORK? However, especially in the courts, we may find that supervision really plays a minute role. The Supreme Court tends not to hear cases. In fact, it hears only 0.05% of all cases appealed to it each year. Therefore, there must be another factor not included here in the utility firnctions Of the lower courts that can explain the broad compliance which has been found to exist. One such factor could be a legal role perception or socialization that would take the form Of adherence to precedent and the perception Of an inequality between the CA and the SC that should be resolved in favor Of following SC pronouncements. This would mean that the preceeding formulation Of the model is incorrect; that instead Of making utility maximizing choices, the CA instead relies upon a set Of heuristics, as Brehm and Gates call them, in their decision making. These would be vertical and horizontal stare decisis (consistency), role perceptions (legal culture), and the Observation Of the behavior Of other circuits (social proof). Where courts are concerned, I assume that the last two feed into a general “role” category. The game tree that follows (Figure 2.2) includes payoffs derived from such considerations, with the assumption that the Court Of Appeals wants to avoid the Supreme Court decision. I assign cardinal numbers to the payoffs and solve for the “tipping point;” that is, the point at which the Court of Appeals will be indifferent between following a disliked precedent and not 34 followir for the ( time spe Not Foll Not Rey Follow, Review preferen. with sup Power > worst (1" ofa‘ B a SUpreme deeision, F1 are deterr equilibriu, likelihood \. ‘ l and [hen Ii associated upenisio UPICme C supen'j510{ following it. Considerations included in the derivation Of these rank-order payoffs include, for the Court of Appeals, policy preferences, adherence to precedent, conception of role, time Spent in leisure, and the negative utility Of being supervised. (Follow = More Leisure; Not Follow > Follow; Not Follow, No Appeal > Follow, NO Appeal; Not Follow, Appeal, Not Review > Follow, Appeal, Not Review; Not Follow, Appeal, Review, Affirm > Follow, Appeal, Review, Affirm; Follow, Appeal, Review, Reverse > Not Follow, Appeal, Review, Reverse) .21 For the Supreme Court, the payoff calculation includes policy preferences, power considerations, and leisure time, taking into account costs associated with supervision (Supervision = -Leisure; Powermcm > Powerflnm; Policy > Leisure; Power > Leisure; Powerm,cw > Powerm. mi”). The outcomes are ranked from best (8) to worst (1). As I Show, this decision depends on the lower courts’ beliefs about the values Of or, B, and y; i.e., the lower court’s beliefs that the parties to the suit will appeal, that the Supreme Court will take the case, and that the Supreme Court will affirm its renegade decision. From the game tree, it should become immediately evident that the equilibria here are determined by lower court evaluation Of the various unknown values; that is, the equilibrium depends upon lower court expectations of the likelihood of appeal, the likelihood Of Supreme Court review, and the likelihood Of reversal Should the case be 2' Payoffs were determined by adding and subtracting the relevant considerations, and then ranking the outcomes. For example, the payoffs for the Court Of Appeals associated with Not follow, NO appeal = Policy — Leisure — Stare Decisis — Role + Supervision (no negative costs due to supervision) = 8 (best outcome). The payoff for the Supreme Court for the same outcome is —Policy — Power + Supervision (no cost Of supervision) = 1 (worst outcome). 35 appeale those 6‘ enforce: followir the low follows: appealed and the Supreme Court see fit to hear it. The Supreme Court can manipulate those expectations by its behavior in order to exact lower court compliance, should enforcement become necessary. More formally, we can set the expected utility Of following equal to the expected utility of not following in order tO find the point at which the lower court will switch from following to the more deviant not following behavior, as follows: EGOLLOWING) = olib/(2) + 043(1 - 10(6) + “(1 - 9X3) + (1 - 00(4) =-4aBy +3aB-a+4 (10) E(NOT FOLLOWING) = aB(1-y)(1) + aBy(5) + (1-a)(8) + a(1- [3)(7) =4aBy-a-6al}+8 (11) 36 “1dr dc are aPpc High CC FIGURE 2.2: GAME TREE - PRINCIPAL AGENCY IN ACTION OwRI‘OF Since these values are not easily interpretable, we might plug in values for a, B, and y derived from the judicial politics literature. We know that at most 20% of all cases are appealed to the Supreme Court (Carp and Stidham 1996). We can, therefore, assume that a=0.20. We also know that only one half Of one percent Of all cases appealed to the High Court are accepted for review there (Humphries, Sarver, and Songer 1998). 37 Therefo the case y, there we Obta Again it pronout explain ‘ norm of includin lends th influenc. Perceprj COmpllal PrefEren therefOn Plausible \ for the 0* fOiiQWr~ I lower CO! Ilppmg p‘ QUlte a dl Therefore, we set B=0.05. Finally, we know that the High Court is more apt to reverse the cases it accepts for review. In fact, it affirms only about one third of the cases it hears. y, therefore, is set to 0.33 (Epstein et a1 1997). Plugging these values into (10) and (11), we Obtain the following expected values: E(Following) = 3.8432 E(Not Following) = 7.7532 Again we find that these lower courts just Should not be following Supreme Court pronouncements.22 However, reality tells a different story and we therefore need tO explain the difference between rational expectations and the legal reality. Possibly, the norm Of vertical stare decisis overshadows the influence of many Other considerations, including, perhaps, policy preferences. Or maybe the perception of role Shared by judges tends them toward deference. The role perceptions I have included in the payoffs may not influence the judges to the same extent as, say, policy preferences; that is, perhaps role perceptions should be weighted more heavily. Somehow, the Supreme Court induces compliance, perhaps Simply via valid authority. Or, on the other hand, perhaps policy preferences simply coincide between the Court Of Appeals and the Supreme Court, and therefore elusive behavior is not necessary. Any Of these factors seem tO provide a plausible explanation for the incongruence between the theoretical expectation and reality. 22 One could also substitute only two Of the three unknowns finding tipping points for the others in order to fiIrther understand lower court considerations in the decision to follow. In SO doing, however, I find that the tendency to not follow is SO strong that the lower court will never switch from non-compliance tO compliance (the value for the tipping point in all cases is negative). It seems that the Supreme Court, therefore, is in quite a dilemma. 38 A comp! conform: behayior producin exists be h indeed it compliar Appeals treatmen Court po begin the foci Seek Why. If that SUpr the princi agency C( A completely different, yet plausible, solution to the puzzle is that studies finding conformity by lower courts overlook what I label ‘creative non-compliance.’ Such behavior Ofientimes looks like compliance and may consistently be underestimated, producing the appearance Of deferential behavior. If this be the situation, little difference exists between our theoretical expectation and reality. In the remainder of the dissertation, I 1) test the theory that the High Court does indeed influence decision making in the Courts of Appeals, and 2) attempt to explain that compliance. The following chapters, then, test for Supreme Court influence on Court Of Appeals decision making in the confession cases, Similarity among the circuits in their treatment of Supreme Court precedent, lower court treatment Of two non-salient Supreme Court policy pronouncements, and differences in decision making between people who begin their careers on the Court Of Appeals and end it on the Supreme Court. All of these foci seek tO determine if and when the lower court defers to its principal, and, in the end, Why. If we confirm that the lower courts are indeed faithful principals, while recognizing that Supreme Court review is highly improbable, we necessarily establish a new variant Of the principal-agent relationship: one in which the mere presence of the principal induces agency compliance. 39 Court or focus. I utility, pr developn validity a fact patte 3.] CASE THEI l l SPEaking, I0 unden; Se analysis I dEClSion in ChOOSir, SUpreme ( making th In addm'On CHAPTER 3: CONFESSION CASES IN AMERICAN COURTS In order to test the notion of judicial hierarchy and the influence Of the Supreme Court on decision making in the Court Of Appeals, we first need an issue area on which to focus. I have chosen confession cases and in this chapter I present the arguments for their utility; provide the history of their treatment in the Supreme Court; and discuss the development Of the law Of confession. Next, I discuss case selection, and the handling of validity and reliability. Finally, I present the method by which these cases are analyzed — fact pattern analysis — and delineate the facts to be coded here. 3.1 CASES INVOLVING CONFESSION: THEIR HISTORY AND AN ARGUMENT FOR THEIR USE I choose the confession cases for several reasons. First Of all, pragmatically speaking, there are enough cases in both the Supreme Court and in the Courts Of Appeals to undertake a sound and rigorous analysis. Second, because these cases deal with controversial civil rights, they fit the type Of analysis I employ. In this area Of the law, differences in fact contribute to the Court’s decision. The Supreme Court itself has indicated as much (see Haynes v. Washington”). In choosing this area Of law, we are able to control for case differences between the Supreme Court and the Courts of Appeals and then directly compare their decision making, thereby controlling for the fact that we are comparing two distinct sets of cases. In addition, choosing a single area Of the law in which differences in fact beget differences 40 in outcor preceder add or St E interactic method c method, I doctrine r Escobedr F i Area Fre resPonds “963: 13‘ Obtain cor 3‘2 THE I the Suprer balanCe so delineaI es the 1.1mm SurTouhdin in outcome, at least theoretically, we are also able to model lower court attention to precedent, as a given Supreme Court precedent in this area of law can generally be seen to add or subtract a given factor from the voluntariness consideration. Because theirs is the first attempt at modeling Supreme Court — Court Of Appeals interaction as a principal agency relationship, using Songer, Segal and Cameron’s basic method Of comparing decisions will be a usefirl starting point. In order tO use their method, however, the issue area must include a Significant change in Supreme Court doctrine over time (19942677). Again, confession fits well, for, I will Show, Miranda and Escobedo, among Others, clearly changed the Court’s doctrine. Finally, other scholars have recognized the utility of fact pattern analysis in this area. Fred Kort, for example, identifies a long list of considerations tO which a court responds when deciding whether a confession was so involuntary as to warrant reversal (1963: 137). These include defendant characteristics as well as the tactics employed to Obtain confessions, including physical and psychological coercion. 3.2 THE HISTORY OF CONFESSION Otis Stephens has authored a scholarly treatment Of involuntary confessions and the Supreme Court, in which he examines how the Supreme Court has attempted to balance social order and individual rights in this Specific area (1973). In his study, he delineates the history Of the amendments related to confessions, the "fair trial" doctrine, the McNabb-Mallory rule, and finally, the impact of Miranda and the controversy surrounding that decision. 23373 US. 503 (1963) 41 through incriminz Amendn has read confessit English 1 primarily ObjeCIior convictir The protection against police coercion Of confession is afforded tO U. S. citizens through a combination Of the Fifth Amendment immunity against compulsory self- incrimination, the Sixth Amendment right tO the assistance Of counsel, and the Fourteenth Amendment, which made the aforementioned applicable to the states. The Supreme Court has read these amendments to justify its activism in the area Of police interrogations and confessions. The earliest basis for Supreme Court policy making in confession cases is the English common-law rule excluding involuntary confessions, which was designed primarily to guard against the introduction of unreliable evidence. It was not aimed at objectionable interrogation practices, but at the protection Of the accused fi'om an unjust conviction (Stephens 1973). Initially, the Supreme Court reviewed only those convictions based on coerced confessions, directly condemning the police inflicted “third degree.”24 This protection did not reach very far, for the Court also recognized the “evidentiary rule” which stressed the reliability Of the confession. With this rule, “third degree” questioning may not affect admissibility if the accused did in fact commit the crime, as evidenced by other facts presented at trial. In the 19205 the Court began to focus on a policy that was much more concerned with the fairness Of the trial than with whether or not the accused actually committed the 24"Third degree" may be defined as: "the employment Of methods which inflict suffering, physical or mental, upon a person in order to Obtain information about a crime" (Stephens, 1973: 10). 42 crime. 1 admissit Obtained the dete: doctrine activity freedom prOIeCIll l aforeme POIICE Q the requ PTOCeSSQ atmOSph COUnSe! determir \ 2; crime. In other words, the concern began to become due process, excluding fiom admissibility confessions that may have even been true, but nonetheless were improperly Obtained. This has been dubbed the "fair trial" doctrine, which replaced trustworthiness in the determination Of admissibility. Justice Louis D. Brandeis quite nicely articulates this doctrine in a dissent: At the foundation of our civil liberty lies the principal which denies to government Officials an exceptional position before the law, and which subjects them to the same rules Of conduct that are commands to the citizen...ReSpect for law will not be advanced by resort, in its enforcement, to means which Shock the common man's [sic] sense Of decency and fair play.25 The so-called "judicial revolution" Of 1937 increased the pace Of Supreme Court activity in the field of confessions, as well as in other areas of criminal procedure, and in freedom Of expression and race relations, reflecting its growing concern with the protection Of individual rights (Stephens 1973). During this judicial revolution the Court struggled tO attempt to define the aforementioned “fair trial” doctrine. Defining the scope Of constitutional limitations on police conduct is difficult, as the Constitution clearly expresses no one standard. Instead, the requirements Of a fair trial relied for constitutional legitimacy on the concept Of due process. The Court looked at such factors as public interest or excitement, the atmosphere surrounding the trial, characteristics Of the defendant, and effectiveness of counsel (Stephens 1973), thereby considering both legal and extralegal factors in the determination Of a fair proceeding. 2’Bureau v. McDowell, 156 US. 465, 477 (1921) 43 easy one precedin confessir unanimo need not doctrine. fundame resulting doortoi Slight, accused, 0f the dc Violence balanClng CNOD’D’U‘P“ The first major case in which the Court invoked a fair trial standard was a very easy one out Of Mississippi.26 The Court found that the extreme physical coercion preceding the confessions of three Black men rendered their "free and voluntary" confessions inadmissible and their convictions invalid. Justice Hughes wrote for a unanimous Court, indicating that the ruling rested on the requirements of due process, and need not rely on the protection against self-incrimination. Applying the "fair trial" doctrine, Hughes argued that because the wrong done to the defendants was SO firndamental, it "made the whole proceeding a mere pretense Of a trial" rendering the resulting conviction and sentencing completely void (at 286). This decision Opened the door tO review of police practices and the protection Of prisoners, even if that Opening was slight. In the late 193 Os, the Court began tO extend the protections accorded to the accused, determining first whether or not coercion occurred and later extending the scope Of the definition of that coercion. The Court no longer relied only on reports of physical violence, but exonerated the accused on the basis Of psychological coercion as well, balancing the amount Of pressure exerted by police against the capacity of the suspect to handle that influence. Justice Black eloquently sums this up in Chambers v. Florida: Under our constitutional system, courts stand against any winds that blow as havens Of refuge for those who might Otherwise suffer because they are helpless and weak, outnumbered, or because they are non-conforming victims Of prejudice and public excitement. Due process of law, preserved for all by our Constitution, commands that no such practice as that disclosed by this record shall send any accused to his [sic] death. NO higher duty, no more solemn responsibility rests upon this Court, than that Of translating into living law and maintaining the 26Brown v. Mississippi, 297 US. 278 (1936) 44 Ir focusing sun'eillar requiremu McNabb- accused I obtained McNabb; 52 di In Mal/or reqUiremi 30'. 31 it did "0t ‘ constitutional Shield deliberately planned and inscribed for the benefit Of every human being subject tO our Constitution -- of whatever race, creed, or persuasion.27 In 1943 the Court introduced a separate standard Of confession for federal cases by focusing directly on police procedures that greatly widened the scope Of judicial surveillance over the interrogation process. The Court also broadened the due process requirements of state confession admissibility well beyond the "fair trial" doctrine.28 The McNabb-Mallory rule, named for the two cases which established it,29 provided that an accused person be promptly arraigned before a judicial Officer, with any testimony Obtained beforehand excluded from trial. Justice Frankfurter summarized the rational in McNabb: The history of liberty has largely been the history Of Observance Of procedural safeguards. And the effective administration Of criminal justice hardly requires disregard Of fair procedures imposed by law.’0 In Mallory the Court sets forth explicitly the meaning Of the "prompt arraignment" requirement, attempting a definition Of "unnecessary delay."31 The Court did not outlaw 27309 US. at 241 (1940) 28Justice Black even advocated a standard of "inherent coerciveness" (in Ashcroft v. Tennessee, 322 US. 143 (1944)) which would treat any interrogation as suspect because of the intrinsic element of police control. Because Of its radical nature, this standard did not receive the Court’s blessing for quite some time (Stephens 1973). 29McNabb v. US. 318 US. 332 (1943) and Mallory v. US. 354 US. 449 (1957) 30318 US. at 347 31While Mallory did outlaw delay in arraignment solely for interrogation purposes, it did not indicate how much leeway the police would have when the arrest tOOk place at a time when no magistrate was available for arraignment (Stephens 1973). 45 police 1 detentil right to resistar police, confess (see no joined I consen restricri Warren ll'lle’ldL “COCFCII William c0“fESsi police questioning, but it did insist on at least a semblance Of judicial surveillance of detention. Appearance before a magistrate ensured that suspects were informed Of their right to remain Silent and tO the assistance Of counsel. These decisions elicited great resistance from law enforcement and those within the government sympathetic to the police, as the federal Standard in particular had, they argued, no real Constitutional basis.32 During the 19405, the Court had difficulty endorsing one Single position on confessions, oscillating between Black's earlier-mentioned test Of "inherent coerciveness" (see note 28) and the traditional "fair trial" doctrine. When Justices Clark and Minton joined the Court in 1949, a new majority put an end tO judicial activism in the area. These conservative justices created a majority who began to ease Fourteenth Amendment restrictions (Stephens 1973). During this time our analysis begins (1946 to be Specific). This conservative phase was short lived, however. Following the onset Of the Warren Court in 1953, an expansion of constitutionally guaranteed protections of individual liberties developed. The Court began to consider the “totality Of circumstances,” deciding the admissibility Of confessions on a case by case basis with “coercive” broadly defined. Attempting to further protect against unjust accusation and conviction, Justice William 0. Douglas began to urge linkage of the right tO counsel with the ban on coerced confessions. He argued in Crooker v California33 : 32Stephens (1973 285-89) gives a good account Of the Opposition the Court encountered. ”357 US. 433 (1958) 46 Douglas "trustwc stand. fc inquisitc Amendn produce llapp V unreasor Obtained Court. II that WET: that an“ The mischief and abuse of the third degree will continue as long as an accused can be denied the right to counsel at this the most critical period Of his ordeal. For what takes place in the secret confines of the police Station may be more critical than what takes place at the trial (at 444). Douglas’ View prevailed in the early 19605 with the formal repudiation Of the "trustworthiness" principal. Convictions based on involuntary confessions could not stand, for, according to Justice Frankfurter: '...ours is an accusatorial and not an inquisitorial system..."34 In 1962 due process was extended to include the Fifth Amendment's protection against self-incrimination.” The incorporation of additional provisions of the Bill of Rights to the states produced a revolution in criminal procedure (Stephens 1973). In 1961 the Court decided Mapp v Ohio, and in 1964, Escobedo v. Illinois. 3 6 The former protected against unreasonable searches and seizures, while the latter made inadmissible any confession Obtained if an accused was denied counsel. Miranda v. Arizona37 dispelled any uncertainty resulting from Escobedo. The Court, making state and federal standards identical, spelled out the necessary warnings that were to precede an arrest, including 1) the right to remain silent; 2) an explanation that anything said can and will be used against the accused; 3) the right to consult with an attorney who may be present during interrogation; 4) the explanation that, if the accused 3"Rogers v. Richmond, 365 US. 534, at 540 (1961) 35Gallegos v. Colorado, 370 US. 49 (1962) 3“’367 US. 643 and 378 US. 478, respectively 37384 US. 436 (1966) 47 cannot at wishes to request c interroga goyemm "knowing questioni Notwiths suspects example, Court in inadmiss Provided H8 argu‘ the testir (JJ cannot afford an attorney, one will be appointed free Of charge; and 5) if the suspect wishes to waive his or her right, S/he may nonetheless assert the right to remain silent or request counsel at any time during the course of the proceeding, at which time interrogation must cease. The Court further stated that “...a heavy burden rests on the government to demonstrate” that, if a defendant does waive his or her rights, S/he did so "knowingly and intelligently.”8 The Court concluded that the atmosphere and act Of questioning is inherently coercive, agreeing with Justice Black's earlier Opinion. Notwithstanding Miranda the Warren Court occasionally waffled in its support of suspects’ rights, even employing the abandoned “voluntariness test.”39 With the arrival of Chief Justice Burger, standards were further relaxed. For example, while Warren Opposed any use of a coerced confession, Burger, writing for the Court in Harris v. New York, ’0 stated that “It does not follow from Miranda that evidence inadmissible against an accused in the prosecution's case in chief is barred for all purposes, provided Of course that the trustworthiness Of the evidence satisfies legal standards.”41 He argues that an involuntary confession may constitutionally be used in court to impeach the testimony of the defendant Should s/he choose to take the stand. 38Ibid at 475. ”Boulden v. Holman 394 US. 478 (1969) 4"401 US. 222 “Id. at 476 48 asthec: npeof: some of "volume required includin, questior counseh comprig. judkfialr OpUUOn- ‘l‘ilranda 48 49(1962 In addition to the Stephens book, other studies Of involuntary confession, as well as the case law itself, lend legitimacy tO Kort's assertion that such cases were ripe for the type Of analysis he proposed. Israel and LaF ave (1971), as well as Ringel (1966), identify some Of the most important developments in the case law with respect tO the "voluntariness test." Haynes v. Washington, ’2 for example, established that voluntariness required the examination of the "totality of circumstances" surrounding a confession, including such circumstances as 1) physical abuse;"3 2) threatsf" 3) extensive questioningf’5 4) incommunicado detention;46 5) denial of the right to consult with counsel;"7 and 6) characteristics and Status of the suspect”. These and other facts comprise the fact patterns I will analyze. This area Of the law Should prove instructive in enhancing our understanding Of judicial decision making. The general public deems it important, as countless public Opinion polls Show exasperation when a reputedly guilty person goes free simply because “373 US. 503 (1963) 43Lee v. Mississippi, 332 US. 742 (1948) “Payne v. Arkansas, 356 US. 560 (1958) 4’ Turner v. Pennsylvania, 338 US. 62 (1949) 46Davis v. North Carolina, 384 US. 737 (1966) "Fay v. Nola, 372 US. 391 (1963), Escobedo v. Illinois, 378 US. 478 (1964), Miranda v. Arizona, 384 US. 436 (1966) “Culombe v. Connecticut, 367 US. 568 (1961); Gallegos v. Colorado, 370 US. 49 (1962); Spano v. New York, 360 US. 315 (1959) 49 “the co writing lead the confess 3.3 DE defined involvir moreoy would t Spaeth‘ criticisn West’s , belWEQE 5. “the constable blundered.”"9 Finally, scholars deem it important, evident by the wealth of writing about the confession cases, especially Miranda. 5° The character of the facts that lead the Court to affirm or reverse Of a conviction on the basis Of an improperly Obtained confession provide a basis by which to test the larger theoretical questions presented here. 3.3 DEFINING THE UNIVERSE I use a “legalistic” approach to identify the universe of cases for this project. This defined universe was Obtained by using West’s Key Number system. Many cases involving a confession turn on considerations independent Of the confession. The Court, moreover, uses widely varying language. It was therefore apparent that a LEXIS search would not be feasible. I also considered utilizing Shepard ’5 Citations in conjunction with Spaeth’s confession cases, but was not happy with that prospect either, due in part to criticism Of Shepard ’s as a selection tool (Songer 1988)5 1. Therefore, I settled upon West’s Decennial Digests, taking every case under keys 516 through 538 decided between 1946 and 198152 in both the Supreme Court and the Court Of Appeals. These ”Then judge Cardoza in People v. Defore, 150 NE. 585 at 587 (1927). 50See Stephens (1973:165-200) for references to a number Of such studies. 51I do employ Shepard ’s in a subsequent chapter, however, and deem that there, Songer’s criticism is lacking in force. He does have a point, however, where the selection Of cases for this type Of analysis is concerned. If I were to rely on the Court Of Appeals to cite Supreme Court precedent, I would miss instances wherein they merely cite none or where they cite their own cases as precedent, thereby potentially missing important realizations Of Court Of Appeals non-compliance, although some studies have found this to be the exception rather than the rule (Benesh and Reddick 1998). 52The analysis begins with the Vinson Court because that is the earliest Court for which we have substantial data collection. The analysis ends with Stewart’s resignation in 50 keys are subhead West Pl. in the le construc cases an inadmiss 3.4 FAC for many and Cam beget dif very sma in any dis Which to Ofa faCL] mOdel of decisions ADPEals, releoam a 1981. there. The keys are under the larger subject of “Criminal Law,” under the heading “Evidence” and the subheading “Confessions.” The title of each of these keys is listed in Table 3.1. Because West Publishers has a very strong reputation, evidenced by the wide use Of their products in the legal community, I view their listing tO be exhaustive and validly and reliably constructed. These at least were a defensible universe of cases. I assume any omitted cases are randomly distributed and are not systematically related to either admissible or inadmissible confessions, having no systematic fact pattern. 3.4 FACT PATTERN ANALYSIS Fact-pattem analysis has been utilized in different forms and in different issue areas for many years. Only recently has this method been theoretically infused (Songer, Segal and Cameron 1994). The method itself relies on the assumption that differences of fact beget different outcomes. It has been noted over time (although not empirically) that a very small difference in the facts of a case may substantially impact a decision. Therefore, in any discussion of comparing decisions in one court with those in another a means by which tO empirically control for differences in fact must be proffered. I advocate the use of a fact-pattern analysis. It is important to note that I am not advancing a fact-based model Of either courts’ decision making. Rather, I am controlling for differences in decisions and attempting to test Of the influence Of the Supreme Court on the Court Of Appeals. In other words, do the facts that the Supreme Court has identified as being relevant considerations in the determination Of voluntariness in turn matter to the decision 1981. The law of confession was largely settled by that time. Hence, I end my analysis there. 5 1 of the ( that the control decisior benefit! apparen as his qt lacked r quantita t0...deci. dEfInitiO to identi these ele reill’ctlrlin Kon's pa Slandard: factual is OVer 25 f mild] to I of the Courts of Appeals? I review early attempts at fact—pattern modeling, then, not so that these early studies might be replicated, but rather so that the methodology I use to control for differences among cases be seen as validly and intelligently employed. One of the first scholars to attempt a quantitative fact-pattern analysis of judicial decision making was Fred Kort (1957, 195 8, and 1963). Kort recognized the inherent benefit of studying appellate court cases by analyzing the different patterns of fact apparently guiding the judges’ decisions. His early work, however, was greatly criticized, as his quantitative methods were ad hoc at best (Fisher 1958). While his work perhaps lacked methodological rigor, it was theoretically usefiJl, as it outlined a method to quantitatively predict court decisions. Kort first demonstrated what he called “...an attempt to apply quantitative methods t0...decisions by the Supreme Court of the United States” (195721). He provided an early definition of the fact-pattern, explaining that "...it is possible to take some decided cases, to identify factual elements that influenced the decisions, to derive numerical values for these elements by using a formula, and then to predict correctly the decisions of the remaining cases in the area specified" (1957: 1). This was admittedly quite far-sighted on Kort's part, although the method he subsequently delineates is primitive by today’s standards. Kort used the Supreme Court’s right to counsel cases, and identified a list of factual issues to be considered. He only used fourteen cases, however, and with his use of over 25 facts, he violated many important rules of statistical inference. Even so, he added much to the then current understanding of judicial decision making, concluding that this 52 area. in whit possibility of too much fro 0958)) and i perfectly. Kort pattern analy again used th well, in an et’ seems better ICgression, h only may pr be “possible law stated i- may be able the Conn Fin; Boolean a]! case fOr fac Confession~ area, in which decisions were considered inconsistent, exhibited some consistency and the possibility of prediction when subjected to quantitative analysis. He is careful not to infer too much from his results, however, (which he clarifies in his response to Fisher's criticism (1958)) and admits that his methods are crude, and that no method predicts decisions perfectly. Kort continued to attempt to find the best possible method to conduct a fact- pattem analysis of Supreme Court decisions in subsequent work. In his 1960 article, he again used the right to counsel cases, this time adding involuntary confession cases as well, in an effort to understand the fair trial rule. He employed factor analysis here, which seems better suited than his earlier method. He subjected his factors to multiple regression, but offered no results in this particular paper. Kort notes, however, that not only may prediction result from this quantitative fact pattern analysis, but that it may also be “possible to determine conformance and non-conformance by lower courts to rules of law stated in flexible form by higher courts” (1960: 14). He fiarther states that this analysis may be able to find the underlying rationale for certain rules and doctrines proscribed by the Court. Finally, Kort turned, in later work, to methods of simultaneous equations and Boolean algebra. Although neither is very useful in my analysis, he did make an excellent case for fact-pattern analysis of confession cases (1963). He analyzes involuntary confession, right to counsel, and workman's compensation cases. In his Table l, which is reproduced here as Table 3.2, he lists the facts which seem important to the decision of 53 the involl model (5: A methodo (MLE), ‘ this type 5 deterrnin understa seizure c of fact-p judicial c are preSL HSHHnet dlfielenc (1) (1984389 the Seam; requrem, the involuntary confession cases. I use his list of characteristics, much modified, in my model (see Table 3.3). Although the early history of fact-pattern analysis debated appropriate methodology, that debate is today largely settled, as maximum likelihood estimation (MLE), variations of which include logit and probit models, are eminently appropriate for this type of analysis (Segal 1984). Scholars have applied fact-pattern analysis (using MLE) to various issue areas to determine whether or not such analysis leads to accurate prediction and better understanding of Supreme Court decisions. Segal (1984, 1986) found that, in search and seizure cases, the fact pattern of the case did affect the decision reached. The usefulness of fact-pattern analysis, he reasoned, lies in the explanation of the variance found in judicial decision making. Since the roles and attitudes of a given judge or a given court are presumably basically stable, as he and Spaeth document (1993), it is reasonable to assume that differences in decisions among cases decided by a given court result from differences in case facts (Segal 1984). Segal then delineates his model, including relevant facts for search and seizure cases.” He finds that his model does a good job of explaining the Court’s decisions (19841896), producing a 48% reduction in error. 5 3 The facts he deems relevant to decision making include prior justification of the search, the nature of the search, and the exceptions to the probable cause and warrant requirements. Within each of these areas, he identifies specific relevant facts. 54 Ir considers uses the l lanUlDES by the SL model 31: addition ehherthe Which is TEgIessic area of c Court m. Couns a mOdels f case was formUlati SlaIiStic \ fit. In order to set up his multivariate analysis, Segal identifies the factual considerations he deems important to the determination of a search as unreasonable. He uses the facts given by the lower courts for the determination of probable cause and the lawfulness of an arrest, and when the lower court facts are conflicting with those set forth by the Supreme Court, he uses the lower court facts in attempts to remain consistent. His model also takes into account changes in personnel on the Court, which is an important addition from earlier fact-pattern analyses. All of the variables in his fact-pattern analysis are dichotomous, thereby measuring either the presence or absence of certain facts, excepting the change in Court measure, which is polychotomous. He uses probit to estimate the coefficients, since least squares regression is inappropriate when dealing with a dichotomous dependent variable. Segal’s model does a good job of explaining the decisions in this reputedly messy area of constitutional law. The -2LLR is significant and the estimated R2 is quite large.” Court membership change was also a significant factor, which makes sense, as some Courts are a priori more likely to uphold a search and seizure than are others. Earlier models failed to include this variable. Segal also found that the United States as party to a case was much more successful than any other litigant. These searches, Segal notes, do not appear to be any less severe than those conducted by state officials. ”One should take care in interpreting this statistic, however, as there is no one formulation of the pseudo-R2 that is universally accepted. I report it merely because it is a statistic with ready meaning. It is used here as a quick approximation of the goodness of fit. 55 Seg naystheyi ideological explainabli opponuni successfiil decision r. decisions. strongly < l‘ jusucest Variation challeng but desi baged \ identifi model Stabxe Segal (1986) used essentially the same data to analyze individual justices and the ways they use facts to make decisions. He focuses on those justices in the middle of the ideological continuum, inasmuch as the decisions of those on either extreme are explainable merely through ideological variables. Those in the middle, however, offer an opportunity to examine the role that facts play in the decision, and the number of facts successfully considered by a single justice, using a bounded rationality model of human decision making. Since justices can only consider a limited number of facts in their decisions, this theory argues, the presence of or absence of facts in each justice's subset strongly determines whether or not the justice upholds the search. This analysis begets significant results as well, as he finds that, for all four middle justices under study (White, Stevens, Stewart and Powell), a significant proportion of the variation was explained through reliance upon facts. He also finds evidence that challenges the belief that justices become more conservative with age. Hagle continues this line of inquiry with a fact-pattern analysis of the Court's obscenity cases (1991). In his analysis, he follows the same general approach as Segal, but designedly does so in a much more general fashion. Hagle attempts to explain a quite difficult and vague area of case law using broad- based variables since there is a lack of constitutional language on which to rely in the identification of relevant facts in the determination of obscenity. He uses a cybernetic model of decision making (1991:1041) in which the Court simplifies the case, follows stable decision rules, focuses on the attributes of the situation broadly defined, and has a 56 hnuted by Segé results. ever so ontheu decisio facbpa decisio howeve decisio model . the “'3 Like 3‘ area of some n de‘ermi mugt DE Therefc Efi‘eqs i limited choice set. While this cybernetic model seems plausible, and was, in effect, utilized by Segal in his analyses, Hagle’s gross use of facts hinders the generalizability of his results.’5 Nevertheless, his variable of lower court decision employed is intriguing since it ever so lightly touches on the relationship between the two courts and the influence of one on the other, which is of interest here. Hagle finds that his model explains the vast majority of the Court's obscenity decisions (1991:1050), thereby arguably demonstrating that the "theoretical framework [of fact-pattern analysis] supports the use of specific or broad-based variables" to explain decisions. Hagle's analysis does tend to stray a bit from the focus of fact-pattern analysis, however, which delineates specific aspects of the dispute which may or may not affect the decision reached by each justice. Joseph Ignagni's study of the free exercise cases also purports to use a fact-pattern model (1990). He uses what he calls a "fact-attitudinal" model to examine changes from the Warren to the Rehnquist Courts, concluding that the decisions are indeed explainable. Like Segal's choice of the search and seizure cases, Ignagni takes a seemingly disordered area of constitutional jurisprudence and attempts to make it more explainable. He extends Segal's view of judicial decisions as boundedly rational and identifies some relevant facts for examination. Like Hagle, he chooses very broad facts to determine decisions, but does a fairly good job of operationalizing them and setting up his 55His variables include direction of lower court decision, whether the audience must pay for access to the materials, and whether the case originates in a Southern state. Therefore, his is not really a test of fact pattern differences per se, but more a test of the effects of differing circumstances. 57 model. the dec another analyze Court, questic determ influeni is athec Specific two cor PFOject model c as the a com O M u their ad] Within 6 model. Although advantaged by a skewed model category, he correctly predicts 82% of the decisions using probit, the method legitimized by Segal efforts. Thereby, we have another victory for the fact-pattemer, and another area in which decisions have been analyzed. Utilizing the confession cases of both the Court of Appeals and the Supreme Court, this project employs a sort of fact-pattern analysis to answer the theoretical questions of interest here. I am cognizant of the theoretical problems in utilizing facts to detemtine case outcomes. Some argue that what a fact-based model really measures is the influence of Supreme Court precedent. Others say that the inclusion of controls for facts is atheoretical. However, I argue that, because I am using the fact-pattern methodology specifically as a measure of Supreme Court precedent, thus rendering the decisions of the two courts comparable, any criticism of the fact-pattern method applies little if all to my project. This is a different type of fact-pattern model, and a different way in which to model court decisions, and I think its use is readily defensible. Given the relatively long and distinguished history of fact-pattern analysis, as well as the apparent room for and need for firrther research on the influence of the Supreme Court on Courts of Appeals decision making, I have conducted a fact pattern analysis of both U. S. Supreme Court and U. S. Court of Appeals cases involving confessions and their admissibility. The methods to be employed vary by chapter and will be explained within each. The following section identifies the facts that were coded and the questions of their validity and reliability. 58 3.4 THE FACTS I carefully determined the facts to be coded by consulting many sources on the law of confession. There are many considerations on which the determination of the voluntariness of a given confession may turn. To settle upon the set of facts presented in Table 3.3, I consulted Kort’s work, other sources on confessions and the admissibility thereof (e. g., Inbau and Reid 1962; Schafer 1968; Rutledge 1994), and the Supreme Court decisions themselves. In reading a sample of the cases contained in my analysis, I determined that certain facts seemingly contributed to the Court’s decision in a particular case. Those facts, which are listed in Table 3.3, are described here in more detail, with the specific coding scheme employed shown in Appendix 3.A. Of course, not all of the facts end up being statistically significant; those that are not are excluded from the final models.’6 First of all, we expect actual coercion to be an important fact in the determination of voluntariness. Where coercion is present, manifested either physically or 56I code these facts directly from the opinion of the given court under study, accepting the decision of the lower court wherever the subjective facts were in dispute. “That is, we coded case facts from the perspective of their status prior to the decision of the appeals court” (Songer, Segal and Cameron 19942681, fit 7). The Supreme Court regularly iterates that it . .accept[s] [the lower court’s] judgement insofar as facts upon which conclusion must be reached are in dispute” (Gallegos v. Nebraska, 342 US. 55, at 61), as does the Courts of Appeals: “This court thus accepts and adopts the state trial court’s findings of facts. . . [We] should not disturb the state court’s findings absent convincing testimony that the factual determination was erroneous” 413 F.2d. 459, at 461). It is only their appraisal thereof which may differ. As an additional check on the possibility that the appeals court is ignoring facts to obtain their preferred outcome, I code for factual disagreement among the majority and the court’s dissenter(s). Such disagreement may signal a court majority that is playing loose with the facts. 59 psychologically, the court will more likely overturn the confession.’ 7 Forms of coercion other than those explicitly physical or psychological exist; those are also coded. Deprivation of basic needs includes lack of food or sleep for the defendant. Again, we would expect their presence to increase the likelihood of excluding the confession. Length and place of detention are also coded as coercive forces, since a lengthy detention, or any detention in squalid conditions, also adds to the pressure brought to bear on a defendant. The longer the detention especially in conjunction with a failure to bring the accused before a magistrate, (which will be discussed shortly) the more likely is the confession’s exclusion. Similarly, the conditions of that detention may also contribute to the likelihood of exclusion. If the defendant were held incommunicado; that is, without the assistance of family, fiiends, or an attorney, we would also expect the atmosphere to be additionally coercive, and therefore, enhancing exclusion. Finally, the presence of an attorney has been determined to mitigate coercion. Therefore, if an attorney represents an accused, we expect the court to be more likely to uphold the confession. If an accused requests an attorney and that attorney is denied, however, we would expect the court to be much less likely to uphold the confession, since an important right was not only ignored, but also actively thwarted. In addition to these coercive forces, I have identified several characteristics of the ”Here, I use the word “overturn” loosely. What we are actually measuring again is whether the court accepts the confession as is. The Court may well remand the case to the lower court for further proceedings without overturning it explicitly. In such cases, the outcome is coded as overturning, which we will use here to mean “unacceptable in its current state.” 60 accused which may lead that individual to be more or less affected by coercive methods. These include mental status, intelligence, race, experience, youth, and some other miscellaneous characteristics.58 We would expect a defendant with a mental illness or defect to be more vulnerable to coercion than another like situated defendant. In addition, we would expect a defendant of subnormal intelligence to be more easily coerced than an intelligent defendant. We may expect that a defendant of a minority race to be more susceptible to subtle coercive tactics than a defendant of a majority race since the interrogator(s) may be more intimidating. We would similarly expect a young defendant to be more vulnerable. On the contrary, legally experienced defendants should be less susceptible to police coercion. I have also identified some procedural defects that may taint the voluntariness of a confession. These include things that the police are supposed to do or have power over, that assure the fair treatment of an accused; e.g., the reading of the Miranda rights (or informing suspects of constitutional rights before Miranda); the length and place of interrogation; the use of police relay tactics in questioning; and the bringing of the defendant before a magistrate without unnecessary delay. Most of these are self- explanatory. A confession obtained via a long interrogation in a dingy room without windows by a large number of officers serially and in tandem is more likely to be 58I have included an “other” category because some cases involve unique defendants and the unique characteristics of those defendants do indeed affect the determination of confession’s admissibility. Examples include a mother of several small children, a drug addict, a defendant with high blood pressure, a defendant who was ill, and a defendant who was very slight of build. Many of these defendant characteristics, I posit, will affect the level of coercion likely to have been present during the interrogation. 61 overturned than is one in which the defendant is questioned over a very short time in a light and airy room by one investigator. Similarly, a defendant who is given his or her rights by a magistrate and formally charged with the crime he or she is accused of committing is more likely to avoid coercion than is a defendant who is badgered into confessing before seeing a judge. I also code for the voluntary waiver of Miranda rights, as such a waiver would obviously negate the absence of the Miranda warnings. In addition to these facts, I have also coded mitigating circumstances, courtroom procedural fairness, habeas corpus petitions, the determination of harmless error, prior coerced confession, and fiuits of illegality. The mitigating circumstance variable could be somewhat gross, but I have defined it carefully, so the problem has arguably been abated. I coded this variable as being present when factors serving to mitigate the involuntariness of the confession were established in the record. These include such considerations as the differences between state law and federal requirements, defendants who have studied criminal law, and those questioned while in custody on another unrelated charge. This variable, when present, may help explain why a seemingly involuntary confession is upheld. Courtroom procedural fairness refers to the determination of voluntariness by the court. Early confession law submitted to the jury the question of the voluntariness of the confession. However, it was later established that the judge must make a separate determination of voluntariness outside the presence of the jury in order to avoid prejudicing the jury (Jackson v. Denna, 378 US. 368 (1964)). The fi'uits of illegality variable is present when the court has either determined that 62 the arrest and/or detention of the defendant was illegal or evidence was obtained during an illegal search, and the confession is a fruit of one of these illegal actions. Ifan earlier confession was deemed to have been obtained illegally, I code for the presence of prior coerced confession, expecting that such a subsequent confession will be deemed involuntary as a fi'uit of the first. Harmless error is coded as present only if the judge (or panel) decides that an infraction in criminal procedure did not lead directly to the conviction and so was excusable, or harmless. This factor should also serve to mitigate a finding of involuntariness. Finally, I include a variable for habeas corpus as a control to see if there is a difference in the way in which habeas petitions are handled (versus a regular appeal). Each of these factors plausibly contributes to the likelihood fimction for the decision of the court, either negatively or positively, depending on the expected relationship. (See the appendix for specific coding decisions.) For each fact to be coded, I follow a trichotomous coding scheme, using the facts as stated by the appellate court that made the decision. I have coded each particular fact 3, ‘6 as “mentioned as present, mentioned as not present,” and “not mentioned.” This addition to the Songer et al model should provide additional germane information; specifically, that it will permit analysis of factors present in the given case as well as those which are mentioned by the court as not being present in the case at hand. Each should affect the likelihood that the confession will be overturned. One would think the former to be a bit more important, but it makes perfect sense that mention of the absence of facts is also important. I have split the information as coded into two dummy variables. One 63 dummy measures whether or not the fact was mentioned as present, and the other will measure whether or not that same fact was mentioned as NOT present. The model also contains a variable controlling for the ideology of the court under study (measured as the percentage of liberal judges on the panel), as well as a control for the lower court disposition, since it has been repeatedly demonstrated that the Supreme Court tends to take cases in order to reverse them (e. g., Perry 1991). Reliability tests were conducted, and the results were quite impressive. Of the 5% sample coded by the additional coder, 97.7% of all cells were identical to my coding. In 84.3% of the non-zero cells we agreed (with most disagreements a matter of simple misunderstanding of the coding rules) and in 99.3% of the cells containing a zero there was agreement. I think this provides good evidence of the accuracy of my coding. As far as validity is concerned, the careful construction of the coding scheme along with the extensive use of legal scholars’ identification of relevant case facts in confession cases (e.g., Criminal Justice Institute 1976; Grano 1996; Inbau and Reid 1962; Israel and LaFave 1971; Ringel 1966; Rutledge 1994; Schafer 1968; VanMeter 1973; Wigrnore 1942) bodes well that I am neither missing any substantively important facts, nor am I including any without relevance to the decision made in a confession case. In short, I think my data were validly and reliably coded and will be, in the chapters that follow, appropriately analyzed. 64 TABLE 3.1 WEST’S KEY SYSTEM - CRIMINAL LAW - EVIDENCE” K. CONFESSIONS. Key 516. Nature and sufficiency as admissions of guilt. 5 17. Admissibility in general. (1). In general. (2). In particular criminal prosecutions. (3). Necessity of laying predicate for admission. (4). After proof aliunde of facts confessed. (5). Necessity and admissibility of entire statement. (6). Admissibility as affected by person to whom confession is made. 5171‘50 Voluntary character of confession. (1). In general. (2). Necessity for showing voluntary character. (3). Exclusion if involuntary. 517.2. Absence or denial of counsel; inadequate representation. (1). In general. (2). Failure to request counsel; waiver. (3). Advising of right to counsel. 517.3 Proof of corpus delicti; corroboration in general. (1). In general. (2). Necessity of proof. (3). Requisites of proof. (4). Sufficiency in particular prosecutions. 5 18. Caution. (1). Necessity in general. (2). Necessity where accused is in custody. (3). Sufficiency. (4). Effect. 5 l9. Voluntary character in general. (1). What confessions are voluntary. (2). Effect of prior involuntary confessions. (3). Confessions while in custody in general. (4). Confessions made to persons in authority. (5). Judicial confessions. ”Source: West’s American Digest System. Various Dates. St. Paul: West Publishing Company. 60517.1 through 517.3 appear only in and after the Ninth Decennial, which covers dates from 1981 - 1986. 65 (6). Confessions while in custody for other offense. (7). Confessions while tied or in stocks. (8). Confessions while in custody illegally or under invalid process. (9). Questioning and soliciting in general. 520.Promises or other inducements. (1). In general. (2). Sufficiency of promise or inducement in general. (3). Confessions subsequent to inducements not acted upon when made. (4). Promise of pardon. (5). Promise not to prosecute. (6). Collateral inducements. (7). Subsequent to confession. (8). By detectives, private prosecutors, and individuals. 521.Judicial compulsion. 522. Threats and fear. (1). In general. (2). Fear of mob violence. (3). Corporal punishment inflicted on accused. (4). Exhibiting weapons to accused. 523.Deception or promises of secrecy. 524. Mental incapacity. 525. ------ In general. 526. ------ Intoxication. 527. Children. 528. Codefendants and accomplices. 529. Confessions under oath. 530. Written confessions. 531.Preliminary evidence as to voluntary character. (1). Presumptions and burden of proof. (2). Admissibility of evidence. (3). Weight and sufficiency of evidence. (4). Right of accused to introduce evidence. 532.Determination of question of admissibility. (1). Time of determining admissibility. (2). Discretion of court. 533.Corroboration. 534. ------ In general. (1). In general. (2). Corroboration of confession by inculpatory facts. 535. ------ Corpus delicti. (1). Necessity of proof of. (2). Sufficiency of proof of. 66 536.Use in different proceedings. 53 7.Use of evidence obtained by means of confession. 538.Effect. (l). Conclusiveness on accused. (2). Conclusiveness on jury. (3). Weight and effect. 67 TABLE 3.2 VARIABLES IN THE INVOLUNTARY CONFESSION CASES‘Sl CIRCUMSTANCES OF PRESSURE: 1. Physical violence overtly designed to elicit a confession. 2. Physical violence not overtly designed to elicit a confession. 3. Nonviolent physical pressure, e. g., deprivation of food, sleep, or clothes, or long exposure to bright lights, etc., or any combination of these devices of pressure. 4. "Relay tactics" in questioning. 5. Protracted interrogation. 6. Verbal or mental coercion other than protracted interrogation, e. g., threat of physical violence, threat of arrest of relatives, statement by law-enforcing officers to the defendant that there is a threat of mob violence even though none actually might exist, false statement that accomplice has confessed, promise of leniency, or persuasion by a psychiatrist, or any combination of these devices of pressure. 7. Removal of the defendant to isolated places at night, for questioning. 8. Preceding admittedly coerced confession. 9. No other confession which is admittedly voluntary and indicative of the voluntary nature of the confession in question. 10. No formal presentation of charges. 11. Delay in the formal presentation of charges. 12. Detention incommunicado. 13. An actually existing threat of mob violence, which the defendant notes without being made aware of it by law-enforcing ofiicers. 14. No advice of the right to remain silent. 15. No advice of the right to counsel. 16. Request of consultation with counsel denied. 17. No consultation with counsel prior to the confession. PERSONAL CIRCUMSTANCES INDICATIVE OF THE INABILITY TO RESIST PRESSURE: 18. Negro status. 19. Youth and concomitant immaturity. 20. Illiteracy, subnorrnal education, ignorance of English, mental incapacity, or limited contact with the prevailing culture pattern. (Multiple values if more than one manifestation.) 21. No previous experience in criminal proceedings, or unfamiliarity with criminal procedure in spite of such previous experience. POSSIBLE MODIFYING CIRCUMSTANCEZ 22. No evidence which, apart from the challenged confession, would be sufficient for a conviction. 61This list merely reproduces Kort’s Table 1 (19632137). I have added to and subtracted from it, as Table 3.3 shows. 68 TABLE 3 .3 FACTS USED m BENESH MODEL COERCIVE FORCES 1. Psychological Coercion: Any threat or promise made, trickery, or deceit 2. Physical Coercion: Any form of physical torture or beating, or the use of truth serums Deprived of basic needs, including food or sleep Length of Detention Place of Detention Had no Attorney Request or Actual lawyer Denied . Held Incommunicado POLICE PROCEDURE 9. Miranda: Before 1966 decision, this variable is coded as pertaining to “constitutional rights” 10. Right to remain silent 11. Right to have an attorney 12. Length of Interrogation 13. Place of Interrogation l4. Magistrate: Whether defendant brought before a judicial officer without unnecessary delay. 15. Police Relays 16. Voluntary Waiver: Whether defendant knowingly and intelligently waived his constitutional rights DEFENDANT CHARACTERISTICS 17. Mental Status of Defendant 18. Intelligence of Defendant 19. Race of Defendant 20. Experience with legal system of Defendant 21. Youth of Defendant 22. Other characteristics impacting the admission of confession, such as drug addiction, mother, high blood pressure, illness, etc. MITIGATING CIRCUMSTANCES 23. Mitigating Circumstances: Factors serving to mitigate the involuntariness of confession, such as state v. federal issues, defendant studied criminal law, defendant being held for another crime during interrogation, etc. COURTROOM PROCEDURE 24. Procedural Fairness in determination of voluntariness msgwew 69 OTHER 25. Habeas Corpus 26. Fruit of Illegality: Confession fifth of either illegally obtained confession, illegal search and seizure, or illegal arrest U.S. detention 27. Harmless Error 70 APPENDIX 3 .A: CODING RULES DECISION 0N WHETHER To CODE: Before delving into the coding scheme, do take note that not all cases in this defined universe were coded. IF THE CASE DOES NOT INVOLVE THE VOLUNTARINESS OF A CONFESSION, omit it. This need not be the sole issue, but it absolutely must be AT issue. If a confession is merely discussed but the appellant is not questioning its voluntary nature, I do not code the case. Common reasons not to code include corpus delecti questions, procedural issues (habeas corpus), admission of codefendant’s confession against another, voluntariness of guilty pleas (but do code if they hinge on a coerced confession). Also, if a case doesn’t have an adequate discussion of the facts at issue, do not code it. Make a quick note as to the reason for eliminating the case. If two codefendants are tried together, each alleging violations of their rights in the admission of a confession, code as two cases, noting the caseid for each. If the court quickly discards one of the claims, then just code it as one case. If a defendant makes several confessions, code only the one at issue. (Do note that later ones my be the fruits of earlier ones.) Opinion Author = place a checkmark by the name of the judge writing the opinion. If the opinion is per curiam, note PC. in the margin of the case. vote = vote ON THE ADMISSIBILITY OF THE CONFESSION, as a whole number (e. g., 30 is a three to zero vote) Note: This is not always the same as the vote in the case. judgel, judge2, judge3 = last names of judges. Add (dist) if a district court judge, add (DC) if a DC judge is sitting by designation on the 2'“1 circuit for example, to denote a visiting judge. Add (sr.) if the justice is a senior judge sitting by designation. For en banc decisions, just cram all the names in, keeping the majority and minority together for easier coding. Note en banc in the margin, and I’d suggest reading these first to make sure they involve confession, b/c it is a pain to list all the judges. jvotel, jvote2, jvote3 = enter 1 for overturn confession, 0 for uphold. Overturn is used loosely, so if there is any indication that something is wrong or the case should be reheard or whatever, enter 1. 0 means the confession is acceptable as is and is without error. confess = whether confession was upheld or not. Again, 1 for overturn, O for admit with rule as above. 71 partyl,2,3...etc = party of judge measured as party of appointing president. l=Democrat, O=Republican. For Supreme Court, use Segal/Cover score. enhanc = Circuit heard the case en banc. 1=en banc, O=not. percuriam = whether per curiam decision or not. lacs, O=no. denymag = whether accused was effectively denied the opportunity to see a magistrate (e.g., there were no mitigating circumstances). l=was denied, O=was not denied. disposit = disposition of lower court decision: A = affinned, R=remanded, Rev=reversed, =modified, V=vacated, plus any combination thereof. psych = whether psychological coercion is present. This includes threats, promises, and any other form of inducement which is not physical and is discussed as such by the court. 2=there was psychological coercion, IHhere was no psych coercion, O=not mentioned at all. phys = whether physical coercion is present. This is beatings and the like referred to by the Court as physically obtaining a confession. 2=was physical coercion, l=was no physical coercion, 0=not mentioned. len_det = whether length of the detention was mentioned as contributing to the involuntariness of the confession. 2=held for an unreasonable length of time, l=was not held long, 0=not mentioned. pl_deten = whether the place of detention was such that it provided a coercive atmosphere 2=held in a bad place, 1=facilities were fine, 0=not mentioned. nomagis = whether defendant was brought before a committing magistrate before s/he confessed. 2=was NOT brought before the magistrate, l=was brought before magistrate, 0=not mentioned. Note: When the court decides that the defendant did confess before being brought before a judicial officer, but that the delay in such bringing was not unnecessary, add a mitigating factor “not unnecessary” but still mark nomagis as 2. nomirand = for cases arising after Miranda and subject to it, code whether appropriate warnings were given. 2=wamings NOT given, 1=appropriate warnings given, O=not mentioned or not applicable. nosilent = before or after Miranda, whether suspect was informed of his right to remain silent. 2=was NOT warned of this right, l=was warned, O=not mentioned. 72 noatty = before or after Miranda, whether suspect was informed of his right to have an attorney. 2=was NOT warned of this right, l=was warned, O=not mentioned. Note: Only after Miranda do suspects have to be told that they can have an atty during interrogation and that if they cannot afford an atty one will be appointed for them. If this extra warning is at issue, but the police did tell the accused he had the right to an atty, mark noatty 1 and nomirand 2. codef' = whether the accused codefendant’s confession was admitted at the trial. 2=codefendant’s confession admitted, l=codefendant’s confession not admitted, O=not mentioned. mental = whether the accused has a mental deficiency (retardation, brain damage, insanity). 2=has a mental problem, 1=no mental deficiency, =not mentioned. intell = whether the accused is intelligent. 2=is intelligent, l=not intelligent (moron status, etc), 0=not mentioned. race = whether it is mentioned that the accused belongs to ANY minority race. 2=minority, l=not minority, O=not mentioned. Note: Generally, only the presence of this matters, so choices are really 2 or 0 here. hasexper = whether accused has ANY experience with the law. 2=has experience, 1=does not have experience, O=not mentioned. youth = whether accused is young or not (and the court mentions that he’s young — generally 18 or younger). 2=is young, 1=is not a youth, O=not mentioned. other = any other characteristic that the accused possesses which hinges on the voluntariness of his confession (drug addiction, dnrnk during questioning, sick, tired, etc.). 2=is another characteristic (write it in), 1=is not other characteristic (write it in -- e. g., the accused was not drunk), O=not mentioned. mitigate = whether any mitigating circumstance existed so that the accused confession could be called voluntary. For example, the delay in bringing before a magistrate was not unnecessary; or the accused was in custody for another crime; or this was a state not a federal arrest (50 McNabb doesn’t apply); or the questioner wasn’t a police officer. 2=mitigating circumstance exists (write it in), 1=mitigating did not exist (write it in — not very common at all), O=not mentioned. waiver = accused made a knowing and intelligent waiver of his rights (atty and silence). 2=made waiver, 1=did not waive (or did not do so intelligently), 0=not mentioned. 73 procedur = whether the procedure for determining voluntariness by the trial court was adequate; that is, was the accused afforded a separate hearing outside the presence of the jury and then was the confession admitted to it (as per Jackson v. Denno). 2=procedure was FAULTY, l=procedure was fine, O=not mentioned. volun = whether or not the accused volunteered the information taken before being asked any questions. 2=volunteered info, 1=did not volunteer info, O=not mentioned. unrep = whether or not the accused had an attorney, but code only if the court mentions this explicitly. 2=had NO attorney, 1=had an attorney with him, O=not mentioned. len_int = whether the confession was tainted due to the length of the interrogation. 2=long interrogation, 1=interrogation of reasonable length, O=not mentioned. relays = whether the police interrogated the suspect using relays of officers (the court will call it thus). 2=relays were employed, 1=no relays, O=not mentioned. incommun = whether the accused was held incommunicado, that is, without access to friends, family, attorney, etc. In the company of police officers only. 2=held incommunicado, 1=not held incommunicado (was allowed to call mother, see sister, etc), O=not mentioned. habeas = whether this appeal is on habeas corpus. 2=habeas, O=not. harmless = whether the error involved was deemed by the court to have been harmless. 2=harmless, 1=not harmless, O=not mentioned. factdis = whether there was a factual dispute among the judges on the panel. These can arise in concurrences or dissents. 2=factual disagreement, 0=none. pl_int = whether the place of interrogation contributed to the involuntariness of the confession. 2=bad place where the suspect was interrogated, 1=place was fine, =not mentioned. depbasic = whether the accused was deprived of basic needs during questioning, like sleep, food, clothes, etc. 2=deprived of basic human needs, l=not deprived, O=not mentioned. priorcc = whether there was a prior coerced confession which may have tainted the one under attack. 2=there was a prior coerced confession, l=there was no prior confession, O=not mentioned. 74 fruit = whether the confession was the fruit of some illegality, like illegal arrest, illegal search and seizure, illegal detention, etc. 2=was the fruit, l=was not the fruit, O=not mentioned. req_atty = whether the accused requested and was refused an attorney. 2=did request/was refused, 1=did not request, or requested and was not refused, O=not mentioned. 75 CHAPTER 4: DECISION MAKING IN THE UNITED STATES COURTS OF APPEALS In this chapter, I use the decisions of the Supreme Court to predict decisions in the Courts of Appeals. Rather than estimating a model in the Supreme Court and then applying it to Courts of Appeals (as per Songer, Segal, and Cameron 1994),‘52 I identify the relevant precedential Supreme Court cases and include measures for their applicability, interacted with indicator for the presence of the fact that emanates therefrom. This is an innovative use of fact pattern analysis (FPA) as applied to this hierarchical relationship. I view FPA as measuring the influence of precedent, with each fact a mere proxy for the holding of the precedent in question. Congruently with this perception, I estimate an integrated model which “best” explains the decision to allow or disallow a given confession in the Courts of Appeals. This model includes Supreme Court precedent (the facts), Courts of Appeals ideology, and the ideology of the Supreme Court."3 62 The reason for this difference is a difference in the cases under analysis. The confession cases are not as easily modeled as other issues areas in the Supreme Court via a fact pattern analysis, as it seems that the High Court takes cases in which new facts or new combinations of facts occur so that they may expose the relationship of said facts to the admission of confession. Songer et al. take advantage of an already fact-pattemed area of law and use the model that predicts Supreme Court cases in the lower court. There is no such model for confession cases and I argue that there can be none. These are ideologically decided cases in the High Court, and when the Supreme Court decides to hear a given case, it is with an eye to add to the factors to be considered in the determination of voluntariness. ‘3 One could argue that this integrated model should also include some external influences on the Courts of Appeals. However, because said external influence has not been adequately studied for these lower courts and because measures of public opinion or 76 I contend that the Courts of Appeals, because their members are influenced by their particular attitudes, and because the Supreme Court does not effectively monitor their decision making, will NOT decide its cases only on the basis of Supreme Court policy pronouncements. The earlier decided Supreme Court cases will indeed exude some influence on the decision making of the lower court, but, in addition, because the potential for Supreme Court supervision always exists, the lower court will also look at the ideological composition of the High Court and consider the majority’s ideological bent in reaching its decision, possibly in attempts to predict what the Supreme Court would do should it choose to hear an appeal of the case now before the Courts of Appeals. In addition, I test a model in which the Courts of Appeals decisions are pooled with the Supreme Court decisions (in their fact patterns) incorporating a dummy variable taking a value of one for the Courts of Appeals and zero for the Supreme Court to determine whether there is a significant difference between the two courts in their decision making. If they are not different, we may be assured that, no matter where a confession case arrives for judgment, be it in the highest court in the land, or in the Courts of Appeals, it will be decided in the same manner, and the same conclusion will be reached. However, if that dummy is significant, that still does not necessarily preclude this interpretation. Rather, it allows for two potential conclusions: (1) that the lower court does in fact decide cases differently in this area than does the Supreme Court; or (2) that “mood” are elusive at best, I have decided not to include those potential factors here. Their applicability to decision making in the lower courts, however, will be explored in fiIture research. 77 the lower court hears different types of cases, driving it to either affirm or reverse more often than it would were it hearing the exact same cases as the Supreme Court is hearing. There is preliminary evidence of a difference among the cases. Simple frequencies run on the individual facts do indicate that the Supreme Court takes cases with quantitatively more facts present; that is, the cases heard by the Supreme Court include a whole host of factual considerations. High percentages of Supreme Court cases include psychological coercion, for example (3 2%). However, in the Courts of Appeals, it appears that there are far fewer meritorious cases, as the majority of entries for the cases in those courts contain a zero entry. To illustrate, only 5% of the cases there contain psychological coercion. Now, because of the infrequency of the facts overall, this still could be significantly related to the decision on the confession. However, this indicates that the lower court should in fact be less likely to overturn a confession. The cases it is receiving are more fiivolous.64 Of course, this finding of differences still does not preclude a finding of attention to Supreme Court precedent, and whether the Courts of Appeals heeds the Supreme Court is the most important question here under study. Finally, I explore the possibility that the circuits differ one from another via the inclusion of a dummy variable for each circuit, in attempts to determine whether justice in this area of the law is determined, at least partially, by which circuit heard the case, or alternatively, whether all the circuits are deciding cases compatibly, thereby contributing 6" It seems upon this quick frequency check, that the Supreme Court’s docket in this area could be quite fruitfiilly studied via fact pattern analysis, while its decision making cannot. 78 to “equal justice.” If a difference among circuits is discovered, that difference will need to be explained either with reference to the cases decided by the unique circuit(s), by reference to its differential ideological makeup, or via an explanation of the unique decision making process employed by that specific circuit. 4.1 THE SUPREME COURT IN CONFESSION CASES A major consideration in this chapter are the facts that the Courts of Appeals should arguably be attentive to in deciding confession cases. In order to identify said facts it is necessary to carefully assess the Supreme Court’s jurisprudence in this area of the law. The lower courts, after all, can only apply a given fact to their analysis once that fact is identified as “mattering” to the determination of voluntariness.65 As was discussed in Chapter 3 and in the previous section, the Supreme Court has decided confession cases in an incremental fashion. In other words, it made no sweeping conclusions about what types of, and under what circumstances, confessions should be allowed. Rather, it took cases in which a seeming injustice occurred and noted that certain facts in that lower court decision contributed to its determination. Because no constitutional language explicitly addresses coerced confessions, the Court was forced to deal with each circumstance as it arose.“5 First, the Court focused on the “third degree” tactics employed by police 6’ Of course, it is possible that the lower courts anticipate decisions by the Supreme Court, or perhaps, in deciding a case without a Supreme Court policy prescription, they consider the presence or absence of a given fact to be important. 66For this reason, it becomes somewhat difficult to analyze Supreme Court decision making in this area of the law via fact pattern analysis. In search and seizure, this is made easier by the relatively specific constitutional protection afforded. In confession, it seems that the Court chooses cases on the basis of a fact or combination of facts that it 79 departments, finding that any confession obtained via physical coercion could not stand (Brown v. Mississippi 297 US. 278 (193 6)). However, in that same case, the Court extended coercion to include threats and other psychological measures. The Court also discussed the need for judicial advice before an accused is interrogated (McNabb v. US. 318 US. 332 (1943)), and, using the Fourth Amendment (Silverthorne Lumber Co. v. US. 251 US. 385 (1920)), recognized the inherent danger in using an arguably voluntary confession when said confession was preceeded by a decidedly involuntary one. The Court also recognized that the amount of coercion in the case must be relative to the suspect’s ability to withstand the pressure (Chambers v. Florida 309 US. 227 (1940)). In Chambers the Court opened the door to consider any range of characteristics of the accused which might mitigate or aggravate the finding of voluntariness. All of these facts were established as material by the Court prior to the start of my analysis (1946) and so are expected to exude influence on the decision making of the Courts of Appeals throughout the period under study. After the beginning of my analysis, the Court decided other precedent setting cases. These include Wong Sun v. US. (371 US. 471 (1963)), which excluded confessions obtained as a result of a prior illegality (a prior invalid search or an invalid arrest, etc); Townsend v. Sain and Jackson v. Denna (3 72 US. 293 (1963) and 378 US. 368 (1964) respectively), which established standards for the determination of has not yet ruled upon. Therefore, these idiosyncrasies disallow the construction of a statistical model that fits the Supreme Court. 80 voluntariness; Lynumn v. Illinois (3 72 US. 528 (1963)), which rendered inapplicable harmless error analysis to confessions; Escobedo v. Illinois (378 US. 478 (1964)), which held that once an accused requests counsel all interrogation must cease; Miranda v. Arizona (3 84 US. 436 (1966)), which delineated the warnings that must be given upon arrest in order that a confession be deemed voluntary; Bruton v. US. (391 US. 123 (1968)), which precluded use of a codefendant’s confession against another codefendant; and Milton v. Wainwright (407 US. 371 (1972)), which did use harmless error analysis to uphold a conviction based upon a confession. The Court obviously decided other cases during this time as well, but most involved certain combinations of fact which lead either to upholding or reversing the use of a confession, and therefore did not present the lower court with any new precedent, as such, to apply. In other words, the Supreme Court, once it announced its determination of the facts to be considered in the voluntariness calculation, then took cases that had certain combinations of fact, perhaps to confirm that those factors rendered a confession involuntary (in most cases). It may well be, however, that the reaffirrnation of certain facts by the Supreme Court will lead the Courts of Appeals to more fully comply. Such facts would include the Miranda warnings, the prohibition of physical or psychological coercion, and some defendant characteristics. We should note that the facts which are highly significant should be those that have been variously espoused as important by the High Court. In the earlier mentioned frequencies, we can glean some expectations: psychological coercion should weigh heavily due to the large number of cases the 81 Supreme Court heard wherein that factor was relevant; intelligence of the suspect should matter, as 25% of the cases the Court heard involved that defendant characteristic; race should matter, as 25% of the cases heard by the Supreme Court had a defendant of a minority race; lack of representation should matter, as 23% of the cases heard by the High Court involved an unrepresented defendant; length of interrogation was at issue in 32% of the Supreme Court’s cases, and therefore should matter more to the lower court; relays were present in 23% of the cases; incommunicado detention was present in 35% of them; 23% of the cases included a denial of the presentation before a magistrate; finally, the lack of Miranda warnings and the denial of a request for attorney were also often present facts (30% and 23% respectively). It seems reasonable that the Courts of Appeals may pay more heed to those factors repeatedly treated by the Supreme Court, provided the High Court is consistent in its treatment. Each of the aforementioned circumstances, then, are included in a fact-based (i.e., precedent-based) model of Courts of Appeals decision making. For precedents decided before 1946, the facts themselves are included, as they should have been germane to the Courts of Appeals for the entire period under study. For those decided after that date, however, a dummy variable is created for the applicability of the precedent which encompasses the period from the year in which the case was decided to the present67 and is interacted with the relevant fact. These are then entered into the equation. It is 67 Because I was concerned that using the year of decision in the Supreme Court would prove to be too gross, wherein the trials from which the appellant is appealing to the Courts of Appeals may have been heard before the relevant Supreme Court precedent, 82 assumed that the given facts will not matter to the Courts of Appeals before the Supreme Court decided the applicable case, but that assumption will be tested as well. 4.2 DECISION MAKING ON THE COURTS OF APPEALS The following model (later referred to as equation 4-1) was tested on the Courts of Appeals data: Y, = B, psych - 62 napsych + 03 denymag + ()4 mental + 65 other - 66 mitigate +67 nowaive - 63 volun + 69 priorcc + 010 lawerca + 611 ideal + 012 highct - 613 fair + 6,4 appesc + 615 appwang + 015 apptawn + 617 appbrut + 613 appmilt + 019 appmir - 620 nadep - 621 habeas + 622 enbanc + p. [4-1] where psych is the presence of psychological coercion; napsych is the absence of psychological coercion; denymag is the delay in bringing an accused before a magistrate without the mitigating circumstance that said delay was not unnecessary; mental is the presence of a mental problem or defect; other is the presence of some other defendant characteristic which increased coercive pressure; mitigate is the presence of some fact mitigating against a finding of involuntariness; nowaive is the absence of a waiver; volun is the presence of volunteered information; priorcc is the presence of a prior coerced confession; Iowerca is the disposition of the lower court regarding the confession (1=exclude, O=admit); ideal is ideology measured as the percent liberal on each panel (using the party of the appointing president as a proxy for ideology); higher is the average Segal/Cover score of the Supreme Court for each year; fair is the presence of a fair I also ran the analysis with this dummy lagged at one year. This did not change the results appreciably and so I employ the current measure. 83 procedure in determining whether a given confession is voluntary; appesc is the Escabeda measure, measuring the applicability of the precedent and the presence of the fact of a denial of a defendant’s request to see an attorney; appwong is the Wang Sun measure, again measuring the applicability of the precedent and the presence of the fact that the confession was the fruit of an illegality; apptawn is the Townsend measure, taking a value of one for years after the decision and controlling for the presence of an unfair procedure; appbrut is the Brutan measure, indicating the admission of a codefendant’s confession, controlling for the year in which Brutan was decided; appmilt does the same for the Milton decision, the precedential value there being the applicability of harmless error analysis to a confession; appmir is the measure for Miranda, indicating a failure to advise a suspect of his or her rights after the decision in Miranda was handed down; nadep is the absence of a deprivation of basic needs; habeas is a dummy indicating a habeas corpus petition; and enbanc is a dummy indicating that the circuit heard the case en banc. The expected direction of each is indicated by its contribution to or detraction from the likelihood of a confession’s exclusion. I assume that cases heard en banc are more likely to be reversed (extending Supreme Court treatment logic — that they hear them to reverse them) and that confessions appealed on habeas are less likely to be excluded because these are more inclined to be frivolous. All others should be self-evident from the discussion in the previous chapter. 84 4.3 CODING OF THE VARIABLES Although I have previously discussed the coding of many of the variables (i.e., those based on facts in the opinion) there are a few that warrant fiirther consideration. These include ideal and highct. This section will defend the measurement of these two very important variables. Ideology has been assumed to be an important consideration in judicial decision making for some time, and with the advent of the attitudinal model, most scholars assume that attitudes explain voting on the Supreme Court, and probably contribute to decision making in the lower courts as well (Segal and Spaeth 1993). However, the measurement of attitudes is something not so widely agreed upon. In the Supreme Court, because we know the justices well, we can use scaling techniques and past voting to predict firture decision making. However, we are just beginning to study the Courts of Appeals, so there is much more uncertainty. In addition, the large number of judges precludes each] examination and scaling of each individually. I assume that once we “get to know” those judges better, we will begin to be able to predict their decision making, at least to the extent that ideology is found to matter. However, until then, I think a different measure is needed. Therefore, as my measure of ideology, I use the party of the appointing president, assuming democratic appointees are liberal and republican appointees more conservative. I then create an ideology score for each panel, which is the percentage of liberal justices on the panel. I expect ideology to be positive, as liberal panels will more likely exclude a confession. 85 There are other measures of ideology one could employ, however, and these should at least be discussed. First, I could use the ADA scores of the appointing president in combination with the ADA scores of the senators in the state from which the judge is appointed as Giles, Hettinger, and Peppers (1998) do. However, as they themselves admit, party does just as well in civil liberties cases. Because I am dealing with that sort of issue area (confession), using party is seemingly justified. I also considered the PAJID measure created by Brace, Langer and Hall (1998) for use in the states, but I think it neither possible nor efficient to measure these federal judges in this manner. True, I could use the ideology of the state from which these judges are being appointed, but that begs the question of whether it is the senator or the president which has the most influence over the appointment. If it is the senator, surely we could use the PAJID measure. If it is the president, however, and I contend that it is, I would be hesitant to use a public mood sort of formulation to measure ideology for the appeals court judges. It seems more defensible to rely upon the ideology of the appointing president. In a meta-analysis of 140 books, articles, dissertations, and conference papers to examine whether there is a link between political-party affiliation and judicial ideology, Pinello (1998) finds that indeed there is, and therefore, any more complicated measures lack efficiency. Therefore, I employ the party of the judge (here, measured as the party of the appointing president) as my ideology measure, as above. To measure High Court ideology, I use the Segal/Cover scores in a yearly average. 86 Segal and Cover, using content analytic techniques, “derive independent and reliable measures of the values of all Supreme Court justices from Earl Warren to Anthony Kennedy” (1989:557). Segal, Epstein, Cameron and Spaeth (1995) update, backdate, and extend the Segal and Cover analysis to include the two Bush appointees, and the seven Roosevelt and four Truman nominees. We therefore have a score for every justice on the Vinson, Warren and Burger Courts, which are under study in this analysis. By adding the scores of the nine sitting justices in a given term and dividing by nine, we can obtain a yearly score. This yearly score is my measure of the influence of Supreme Court ideology. It is admittedly gross, but is superior to the circular measure which uses previous votes to predict current and fiIture votes. This variable should also be positive, as the more liberal the High Court, the more likely the Courts of Appeals should be to overturn a confession, fearing that if it decides otherwise it will be overturned upon review. These measures in mind then, we turn to the results. 4.4 RESULTS The results of this estimation are contained in Table 4.1“. Do note that I’ve ’8 I conducted several diagnostics tests on this model to determine its robustness. The results of several of those tests can be found in the appendix to this chapter. First, I tested for multicollinearity examining the correlation matrix (which can be found in the appendix). Second, I tested for omitted variable bias using the RESET test, finding that including the dependent variable on the right hand side did nothing to the estimates, and its coefficient was insignificant. Third, I tested for heteroskedasticity among circuits by running a heteroskedastic probit. I concluded therefrom that heteroskedasticity was not a problem, as the term for variance was insignificant. Finally, I tested for influential cases. I explain this test fiilly in the appendix to this chapter. Suffice it to say here that, while there were a few influential cases, their inclusion does not significantly alter our results. We can have much confidence in these data, then, and in the results obtained via this estimation. 87 estimated this model using robust standard errors. The reason for this is that any multiple regression technique assumes the independence of observations, and usually this is not a problem. Here, however, since I have multiple observations fi'om the same circuit and even the same panel, any contention that they are strictly independent is dubious at best. The estimation of robust standard errors corrects for this and produces more plausible results. Looking then at the model, we are immediately impressed with the fit. The pseudo R2 is an impressive 0.79, although we should take care not to garner too much satisfaction from this fact, given the problems with R2 measures in general and the pseudo- R2 specifically.69 In addition to this evidence of goodness of fit, we also find that the model predicts 95% of all cases correctly, which translates into a 78% reduction in error over the null model. Not all of the facts listed as plausible influences on the determination of voluntariness are included, however. I conducted an extensive set of likelihood ratio tests to assure that the exclusion of those variables not providing a statistically significant impact did not harm the results. In other words, exclusion of variables was based on their overall theoretical contribution to the model rather than their mere statistical insignificance (Granato 1991). This final model, more parsimonious than one including all facts listed in 69 There is some debate over the usefulness of the R2 statistic, or coefficient of determination, if you will. For example, Achen (1982) argues that, since R2 is dependent upon the variation in the independent variables and that variance is clearly a function of the sample, it makes little sense to attribute to its value a measure of the strength of the relationship between the independent and dependent variables. King concurs (1986). More problems occur when using the pseudo-R2 as we are here, because that measure is calculated differently depending upon the statistical package being utilized. It is better to rely upon the percent correctly predicted, which will also be duly reported (Hagle and Mitchell 1992). 88 Table 3.2 (in the previous chapter), provided the best fit.70 We find much of interest in these results (contained in Table 4.1). First, all variables are in the expected direction, although a few are insignificantly so. Both the presence of psychological coercion (psych) and its absence (napsych) exude a statistically significant influence on the exclusion of a confession, with its presence adding to the probability of exclusion, and absence decreasing that contingency. Do note, however, that presence is far stronger than absence. This was a fact in many of the Supreme Court precedents, and so our “differential strength” hypothesis receives some validation (that the lower court would pay special heed to those variables frequently espoused by the High Court). In addition, the denial of a prompt hearing before a magistrate (denymag) significantly adds to the probability of exclusion (another frequent fact for the High Court), as does the absence of a knowing and intelligent waiver (nowaive). Volunteered information from the defendant (volun) detracts from confession exclusion, as does any factor mitigating police malfeasance (mitigate). The existence of a prior coerced confession (priorcc) adds to the probability that a later confession will be excluded, and 7° Do note that several variables, in either their presence or absence, predicted a given outcome perfectly. These include the following: physical coercion, length of detention, place of detention, relays, place of interrogation, and deprivation of basic needs. This results because in every case wherein the variable takes on a specific value, the dependent variable is the same. Thus, in every single case in which there is some physical coercion, the court excludes the confession. Such a variable cannot be included in the model because it has no variance. But it is important to note the large impact it has on the court’s decision even though we cannot explicitly model it. Interestingly, two of the aforementioned perfect predictors are those facts for which the Supreme Court heard a substantial number of cases (race and police relays). This provides some additional evidence of the differential strength hypothesis. 89 the presence of fair procedure (fair) in the determination of voluntariness detracts from it. A couple of characteristics of the accused come to bear here as well; thus, a mental defect (mental) adds to the probability of exclusion, as do other circumstances (other), broadly defined. Habeas corpus petitions (habeas) do not seem to be treated any differently from other cases, although we did expect them to be more often affirrned, and the absence of a deprivation of basic needs (nadep) is also insignificant. The Courts of Appeals are far more likely to rule as the lower court did (lowerca), evidenced by the significant and positive coefficient on that variable, and they do pay some heed to the ideological composition of the Supreme Court (highct), as that variable is positive with a significance level of 0.08.71 Finally, en banc decisions (enbanc) do tend to exclude confessions more often than single panel decisions do. It is interesting to note the differential success of the more recently established precedents. Obviously the Courts of Appeals pay heed to a number of old precedents, evidenced both by those facts authoritatively established prior to the beginning of this analysis that are significant in the model and those other facts that, because of perfect prediction, could not be included (bearing on such precedents as Chambers v. Florida Id. (race), Lisenba v. California 314 US. 219 (1941) (relays)). However, the Courts of Appeals do not seem to heed all of the High Court’s pronouncements. We see here that Wang Sun, T awnsend, Milton and Miranda are all significantly related to the lower court’s 7' Granted, this is beyond conventional significance criteria. However, as I deem the cutoff of 0.05 to be unduly arbitrary, I report it anyway and the reader may contend with it as s/he will. 90 determination of voluntariness, while Escobeda and Brutan are not. Neither are those Supreme Court precedents which dealt with deprivation of basic needs, intelligence of defendant, experience with the legal system (of defendant), the making of a knowing and intelligent waiver of rights, lack of representation, length of interrogation, and incommunicado detention. For many of the variables, however, it is a mere lack of frequency that explains their insignificance. Exploration into why this is so for those wherein there are enough instances of fact presence will flow fiom the analysis in Chapter 5, wherein I will specifically test Courts of Appeals compliance with specific precedents, attempting to explain differential compliance as a firnction of characteristics of the Supreme Court cases themselves and ideological interagreement between the Supreme Court and the deciding panel.72 Finally, ideology is significant, as expected, and decidedly so. Even though I measure this variable somewhat grossly, as discussed earlier, it is significant at the 0.001 level. And its impact on the probability of exclusion is far from negligible as well. In Table 4.1A we see that its marginal effect is 0.079. This means that for each tiny increase in the percentage of liberals on the panel, the probability of excluding a confession rises by 8%, ceteris paribus. So, liberal panels are substantially more likely to exclude a 72 I also ran the facts established in the various precedents without reference to the year in which they were decided in order to determine whether it was merely the fact that was signficantly related to the treatment of the confession or rather the Supreme Court pronouncement on the matter mattered. I find that, indeed, the fact is relevant regardless of the time span. However, upon further investigation, I realized that most of my observations are post-1960 which is around the time most of the precedent-setting cases I discuss were decided (only 78 of the 563, or 14%, were decided before 1960). Therefore, 91 confession than are conservative ones. It is, of course, more usefiil to note the change in probability of exclusion given a 33% increase in liberals, since most of our panels are three-member. A change from a one—liberal panel (ideology=0.33) to a two-liberal panel (ideology=0.67) increases the likelihood of exclusion, when all other variables are held at their means, fi'om 6% to 15%. When there are no liberals on the panel, only 2% of confessions are excluded. When liberals fill the panel, the likelihood of exclusion is 31%. Surely this is a large influence. This demonstrates that while the Courts of Appeals’ judges do pay heed to the precedents of the Supreme Court, at least in their realization as case facts, they remain consistent with their own ideology, notwithstanding the indirect way in which I measure it. In addition to mentioning their significance or insignificance, we may also discuss the differential influence each exude on the exclusion of a confession as we did for ideology. Table 4.1A includes the coefficients as marginal effects, giving the change in the probability that Y=1 for a tiny change in the independent variable. For the dichotomous variables, that change is from 0 to 1, while for the continuous variables (ideal and highct) the probability is computed for an infinitesimally small change in X. Via this table, we see that Wang Sun, or that the confession is the fruit of an illegality, exudes the greatest impact on the decision to exclude the confession, with its presence increasing the probability that the panel will rule against the admission by 83%, ceterisparibus. Psychological coercion is also highly significant, its presence increasing the likelihood of we can confidently attribute the significance of these facts to the Supreme Court’s decision. 92 exclusion by 75%.73 Those exuding the least amount of impact are habeas and nodep, which were not significant, and Milton, which was significant but only decreases the likelihood of exclusion by 2%. The variables we expected to be disproportionately influential, at least according to the sheer number of Supreme Court treatments thereof, are indeed highly influential (psychological coercion, denial of magistrate, and lack of Miranda warnings). This makes good sense.74 4.5 THE POOLED MODEL Although we are unable to model the decision of the Supreme Court to either allow or disallow a confession via a fact pattern model, we can pool the data from the two Courts together in order to determine whether a difference exists between the two exists.” Testing the same model utilized above on that pooled data (excepting highct and enbanc which are lower court-specific), and including a dummy variable, court, which takes a 7’ Do note though, that these are not necessarily seen as being stronger than the ideology or high court variables, as those are measured continuously. 7’ I did attempt to also run the model on only the nonunanimous cases, thinking perhaps that in these cases, the lower court judges’ attitudes would more clearly show themselves. However, only 68 cases in the Courts of Appeals were decided nonunanimously and because of the large number of facts and the aforementioned infrequency of their presence in the lower court cases, the model was impossible to estimate. Ideology was significant by itself in those cases, however, which provides some evidence of the difference between cases decided by a unanimous vote and those that are not. However, the lack of numbers and of frequency of fact presence precludes any systematic interpretation. 7’ I was concerned that the two samples had different variances and therefore estimation would be invalid. For that reason, I ran a heteroskedastic probit on these data as well, finding again that the variance term was insignificant. We can therefore be confident in the results reported here. 93 value of one for Courts of Appeals cases, we obtain a quick test of whether there is a difference between the two courts. If the dummy variable is Significant, then the two groups are different. If it is insignificant, they are not. We may, after estimation of this model, then compare the predicted probability of overturning a confession for each court. As the results in Table 4.2 show, the coefficient on court is indeed significant and in the expected direction. The lower court (court=l for the Courts of Appeals) is significantly less likely to overturn a given confession. This bears out in bivariate analyses as well since we can see there that the lower court admits 77% of all confessions at issue, while the Supreme Court only admits 32% (See Table 4.4). This clearly Shows that there is a difference in decision making between the two federal appellate courts. In terms of marginal effects, we find that the court variable indicates a decrease in the likelihood of overturning a confession, all else equal, by 17%. (Table 4.2A gives this result, along with all other variables’ marginal effects.) There are several possible reasons for this phenomenon. If, for example, the cases systematically differ, or if at least the lower court’s use of certain variables is different, a variable denoting the court would be expected to be significant. It becomes difficult, then, to understand whether the phenomenon under study is one of disobedience of the lower court to its principal, or merely one of differences between cases.76 However, this matters 76 While this problem of the incomparability of cases would be troublesome were my main goal here to compare the decision making of the two courts, it does not much hinder the results I seek to obtain. While it is an interesting problem — how to compare the two sets of cases — my main goal is to explain decision making in the lower court, which I contend the first estimation in fact does. However, in running a combined model, we obtain the interesting result that there is in fact a difference between the two datasets. 94 not to our analysis because, no matter why the lower court decides differently, we still seek to model its decision making, and that we have done. 4.6 DIFFERENCES AMONG CIRCUITS We are left in this chapter with one final question: is the Courts of Appeals an institution or is it a set of lower courts? That is, can we talk about “the Court of Appeals” or must we understand this federal appellate court to be twelve circuit “Courts of Appeals?” In order to determine to what extent the Courts of Appeals decide cases in accord with one another, I conducted a couple of simple analyses. First, I ran the model [4-1] which fits the Courts of Appeals generally, with a dummy variable for each specific circuit individually. That is, I ran one model with a dummy for the DC circuit, one with a dummy for the First circuit, and so on. In so doing, only the DC circuit differed significantly, evidenced by the significance of the coefficient on that dummy."7 To explore this difference, I ran another model in which all of the dummy variables for circuit were included, excepting the DC circuit which was used as the baseline (meaning that each dummy can be interpreted as first circuit or DC circuit, second circuit, or DC circuit, etc.). As can be seen from Table 4.3, most of the circuits significantly deviate from the DC That is, in the lower court, the majority of confessions are allowed while on the Supreme Court, the majority are excluded. This does indicate either a difference in cases or a difference in decision making. If the perceived difference in decision making is due solely to the difference in the cases, which is driven by the differential agenda-setting processes of the two courts, then another very interesting research question arises. That clearly indicates that institutional constraints matter. This notice of a difference motivates the study here undertaken. 77 Results of these estimations available upon request. 95 circuit. Even those that do not share the same relationship with the significantly divergent ones: the DC circuit is always more likely to exclude a confession. This is interesting, as we find that one circuit court is seemingly behaving more like the Supreme Court than are the others, although its frequency of overturning a confession (it does so 45% of the time) still pales in comparison with the High Court’s. Table 4.4 displays the circuit courts’ frequency of overturning confessions. Note the high degree of similarity, except for the DC circuit and for those circuits in which there are too few observations in the dataset to really draw any substantive conclusions. 4.7 DISCUSSION We must think carefiilly what the results delineated above mean for the decision- making process of the Courts of Appeals, and about the relationship between the Courts of Appeals and the Supreme Court. Is it one of agency, or are the two so separate they should be seen as distinct entities? I think the former is true, and I think the results here lead to that conclusion, although not directly. First of all, we do find evidence of a difference in decision making. A model pooling the Supreme Court and Courts of Appeals data demonstrates that the two courts are different. However, this finding alone does not prove that the decision-making processes differ per se. It could very well be that, as alluded to earlier, the cases are just different. The aforementioned set of frequencies provides some evidence of this. Perhaps, because of the mandatory jurisdiction of the Courts of Appeals, it merely receives more fiivolous appeals and therefore has little occasion to reverse a confession. Perhaps the 96 Courts of Appeals are just less likely to overturn a confession, as shown above and in the tables, even when we include measures for other variables. Perhaps the sheer number of cases they hear precludes careful examination of the facts. Perhaps also, as Justice Jackson posits in Stein v. New York (346 US. 156 (1953), citing a lower court opinion) lower court judges realize their role in the judicial hierarchy and instead of trying to avoid being overruled by the Supreme Court, which is a primary motivation assumed by most scholars, they rely instead on the Supreme Court to correct any errors they might make. Or maybe this area of criminal law is just difficult to model. In considering the many iterations of the final model reported here, there were variables that predicted the outcome (either success or failure) perfectly, as mentioned earlier. This is problematic, since we obviously do not have enough variation in those independent variables. On the other hand, much may be learned from the identification of these variables, for difl’erences in strength among the examined facts may exist. Physical coercion, for example, may be so abhorrent that its presence nullifies the effect of any other variable. No matter what the reason for the differences between the two sets of data (which are the two sets of court decisions in their fact patterns) we do find substantial evidence that the lower court considers the Supreme Court in making its decision in this area of law; and they do so across the board. No circuit ignores Supreme Court prescription. While the DC. Circuit is somewhat of an anomaly, still it heeds Supreme Court precedent and considers Supreme Court ideology. This finding, our most important, indicates that these two courts do exist in some sort of hierarchical relationship. 97 4.8 CONCLUSIONS The Courts of Appeals differ fi'om the Supreme Court in some very important institutional ways, which leads us to believe real differences in decision making exist. Indeed, we find some preliminary evidence for such a conclusion here. The Courts of Appeals are influenced by attitudes, just as scholars have found the Supreme Court to be influenced by its ideology. However, in some way, those attitudes combined with the facts of the case lead the lower court more often to conclude that the confession is admissible. The Supreme Court, in this area of case law, is playing its part as an error correction body quite well, lending some credibility to its supervisory role. The high Court is much more likely to overturn a confession than are the Courts of Appeals. This begs the question, however, of whether the Court is only taking such cases as it wishes to overturn or such cases as have unconstitutionally obtained confessions. Similarly, we must wonder if the Courts of Appeals, with their mandatory docket, are receiving a different type of case. However, most important is not that they differ, but how they differ. Here we have shown that they differ because of the institutional constraints brought to bear on the lower court jurist in the form of Supreme Court precedent and authority. We find that the various circuits of the Courts of Appeals do not much differ fi'om one another, although the DC. circuit is clearly an anomaly. This may be due to the different types of cases which that circuit hears, or its jurisdiction over the District of Columbia. Probably it differs because it needs not deal with state cases which are more 98 likely to contain fiivolous appeals. Either way, we can be comforted that generally, it matters not which circuit hears a case. In addition, facts relevant to Supreme Court pronouncements on the matter do influence lower court decision making, and the Court of Appeals judges do consider the ideology of the High Court in making their decisions. The lower court is more constrained than the Supreme Court. These differential constraints cause it to decide cases differently. They must take the Supreme Court into consideration when they make decisions. They must pay greater heed to precedent. They are not completely free agents. 99 TABLE 4.1 DECISION MAKING ON THE COURTS OF APPEALS VARIABLE COEFFICIENT ROBUST STD ERROR SIG LEVEL psych 2.923904 .4997947 0.000 nopsych -.9731026 .3581683 0.007 denymag 2.144201 .3926829 0.000 mental 1.741769 .43 7997 0.000 other 1.135905 .3588758 0.002 mitigate -1 .2353 84 .3655038 0.001 nowaive 2.120143 .609631 1 0.001 volun -2.73646 .49781 0.000 priorcc 2.441595 .67 70546 0.000 lowerco 1 .494748 .2708209 0.000 ideol 1.630921 .5024961 0.001 fair -1.224694 .3603 95 0.001 appesc .5205679 .498455 0.296 appwong 3.168536 .6461919 0.000 apptown 2.535135 .452375 0.000 appbrut .7458668 .5574829 0.181 appmilt -1.46812 .6168742 0.017 appmir 1.970867 .443 5693 0.000 habeas -.0586441 .2508939 0.815 nodep -1.184581 .7597508 0.119 enbanc .8908099 .4573 254 0.051 highct .98493 48 .5542747 0.076 _cons -3. 139999 .4989624 0.000 Percent Correctly Predicted: 95% Number of obs: 563 Percent Correct for Null: 77% Wald chi2(22): 177.75 Reduction in Error: 78% Prob > chiz: 0.0000 Log likelihood: -63 . 166168 Pseudo R2: 0.7907 100 TABLE 4.1A MARGINAL EFFECTS (ALL OTHER VARIABLES AT MEAN) Variable dF/dx Psychological Coercion .7511785 No Psychological Coercion -.0419803 Denial of Magistrate .4196516 Mental Defect .3339979 Other Defendant Characteristics . 1399804 Mitigating Circumstances -.0465 528 No Knowing Waiver .4862786 Volunteered Information -.0422555 Prior Coerced Confession .6090878 Lower Court Disposition .1774203 Ideology .079462 Fair Procedures Used -.0441628 Escobeda .0420856 Wong Sun .83292 Townsend .6160358 Burton .07 50242 Milton -.0206929 Miranda .4123 56 Habeas -.0027923 No Deprivation of Basic Needs -.0253467 En Banc Decision .0990219 High Court Ideology .0479881 101 TABLE 4.2 SUPREME COURT AND COURTS OF APPEALS DECISIONS POOLED VARIABLE COEFFICIENT ROBUST STD ERROR SIG LEVEL psych 1.607802 .4280457 0.000 nopsych -l . 156885 .2878442 0.000 denymag 1.86434 .2955987 0.000 mental 1.83 8404 .3700746 0.000 other 1.106901 .318719 0.001 mitigate -1.703535 .3964364 0.000 nowaive 1.968865 .7442161 0.008 volun -2.203687 .4388097 0.000 priorcc 1.921309 .58263 98 0.001 lowerco 1.414212 .2530235 0.000 ideol 1.3073 12 .4249967 0.002 fair -1. 12644 .2771847 0.000 appesc 1.025743 .4358408 0.019 appwong 2.749837 .4698836 0.000 apptown 1.936058 .3944829 0.000 appbrut -1.24959 1.25336 0.319 appmilt -1.761566 .5560682 0.002 appmir 1.22768 .4759132 0.010 habeas -.0684943 .218147 0.754 nodep -.7352896 .4951178 0.138 court -.9280094 .3911176 0.018 _cons - l . 188449 .4089607 0.004 Percent Correctly Predicted: 95% Number of obs: 620 Percent Correct for Null: 73% LR chi2(21): 554.58 Reduction in Error: 81% Prob > chi2: 0.0000 Log likelihood: -83.932129 Pseudo R2: 0.7676 102 TABLE 4.2A MARGINAL EFFECTS (ALL OTHER VARIABLES AT MEAN) VARIABLE dF/dx psych .4058122 nopsych -. 1045178 denymag .4660471 mental .5049735 other .2262666 mitigate -. 1323535 nowaive .5660888 volun -.0896428 priorcc .5472989 lowerco .2682105 ideol .1403444 fair -.0896428 appesc .2174876 appwong .807298 apptown .5361543 appbrut -.0521772 appmilt -.0543665 appmir .2759948 habeas -.0071985 nodep -.0495004 court -. 17436 103 TABLE 4.3 DIFFERENCES AMONG THE CIRCUITS VARIABLE COEFFICIENT ROBUST STD ERROR SIG LEVEL psych 3.259789 .561908 0.000 nopsych -1.228264 .3894348 0.002 denymag 2.130494 .3 82751 0.000 mental 1.70539 .4745122 0.000 other 1.652251 .3798769 0.000 mitigate -1.429927 .395331 0.000 nowaive 1.724074 .6595408 0.009 volun -3.086458 .8153862 0.000 priorcc 2.629904 .8497964 0.002 lowerco 1.680783 .3443931 0.000 ideal 1.504338 .5861006 0.010 fair -l.148076 .3816071 0.003 appesc .747758 .6120952 0.222 appwong 6.3271 13 .8863268 0.000 apptown 2.649952 .4963306 0.000 appbrut .9460987 .5590065 0.091 appmilt -2.468272 .6451 163 0.000 appmir 2.485563 .5123406 0.000 habeas . 1409642 .3138113 0.653 nodep -1.690286 .9161547 0.065 enbanc .691302 .5973 828 0.247 highct 1.301933 .7657358 0.089 one -.70955 .8582489 0.408 two -2. 183419 .6825248 0.001 three -2.062205 .7607726 0.007 four -1.276153 .668308 0.056 five -1.870591 .4934971 0.000 six -1.511172 .581607 0.009 seven -1.274669 .7643482 0.095 eight -1.501988 .6393989 0.019 nine -1 .37042 .5520437 0.013 ten -5.470482 .83 50338 0.000 eleven -2.0994 19 .6869217 0.002 _cons -1.975653 .6894297 0.004 Percent Correctly Predicted: 96% Number of obs: 563 Percent Correct for Null: 77% Wald chi2(22): 145.83 Reduction in Error: 83% Prob > ch12: 0.0000 Log likelihood: -50.811381 Pseudo R2: 0.8316 104 TABLE 4.4 FREQUENCIES OF CIRCUIT EXCLUSION OF CONFESSION CIRCUIT NUMBER OF DECISIONS CONFESSION EXCLUDED PERCENT First 10 4 40% Second 50 12 24% Third 24 5 21% Fourth 37 13 35% Fifth 73 12 16% Sixth 30 5 17% Seventh 30 6 20% Eighth 52 10 19% Ninth 80 15 19% Tenth 48 2 4% Eleventh 60 1 3 22% DC. 69 31 45% ALL CIRCUITS 563 128 23% SUP CT 57 39 68% 105 THE CORRELATION MATRIX: confess psych --------- + confess I 1.0000 psych I 0.3476 1 0000 nopsych I -0.2988 —0 1787 denymag l 0.4031 0.2213 mental I 0.2453 0 0996 other I 0.2725 0.1051 mitigate | -0.3372 -0 1122 nowaive | 0.2571 -0 0405 volun I -0.1868 -0 0860 priorcc I 0.1415 0 0252 lowerco I 0.6547 0 3001 ideol I 0.1604 -0 0023 highct I 0.0994 0 0294 fair I 0.3796 -0 0639 appesc I 0.1228 0 0627 appwong I 0.2600 0 0965 apptown I ~0.1749 0 0131 appbrut I 0.0046 -0 0222 appmilt I 0.0205 -0 0192 appmir I 0.1937 0 0332 nointell I 0.1035 0 1807 noexper I 0.1696 -0 0273 habeas I 0.2122 0 2080 enbanc l 0.1929 0 0211 l nowaive volun + nowaive I 1.0000 volun l -0 0654 l 0000 priorcc I -0 0296 —0 0198 lowerco I 0 1301 -0 1519 ideol I 0 0231 0 0016 highct I 0 0181 0 0785 fair I -0 0486 -0 0485 appesc I 0 0950 -0 0441 appwong I 0 0577 -0 0600 apptown I -0 0384 —0 0351 appbrut l —0 0169 —0 0359 appmilt | -0 0146 -0 0310 appmir I 0.1887 0 0590 nointell I 0.0039 -0 0212 noexper I 0.0950 -0 0441 habeas I 0.0282 -0 1283 enbanc I 0.0485 0 0172 APPENDIX 4.A: DIAGNOSTICS .0000 .0425 .0170 .0880 .0625 .0199 .1517 .0318 .0162 .0140 .0187 .0079 .0199 .0220 .0529 106 III 000 l OOOOOOOOOOOOOOO .0000 .1425 .1312 .2647 .0281 .0405 .0178 .3254 .0919 .0654 .0281 .0709 .1110 .0655 .0374 .1323 .0373 .1308 .0709 .0782 .0919 lowerco -——————-— _--——-—-—-—————-—-—--——_-—————-————————-————-———-—-———-——-—————- CDCDCDCDCDOCDOOOOCDCDF-J .0000 .0620 .0471 .2119 .1709 .1570 .0835 .0626 .0452 .1958 .0841 .1238 .2047 .1301 mental .0000 .0606 .0319 .0003 .0729 .0464 .0093 .0245 .0172 .0023 .0073 .0120 .0601 other mitigate .0000 .0557 .0416 .0808 .0570 .0121 .1525 .0826 .0355 .0327 .0612 .0797 .0000 .0439 .0446 .1592 .0267 .0849 .0300 .0767 .0439 .0653 .0563 appesc appwong apptown appbrut appmilt appmir nointell | + appesc | 1 0000 appwong I -0 0190 1.0000 apptown I -0 0223 -0.0248 1.0000 appbrut I -0.0114 -0.0155 0.0004 1.0000 appmilt | -0 0099 0.1635 -0.0479 -0.0080 1.0000 appmir I 0 1402 0.0861 -0.0276 -0.0238 -0.0206 1.0000 nointell I 0 0439 0.0123 -0.0351 -0.0267 0.0849 0.0468 1.0000 noexper I -0.0140 -0.0190 -0.0680 -0.0114 -0.0099 0.0555 0.2737 habeas l 0.1337 -0.0131 -0.0290 0.1436 -0.0559 0.0479 0.1425 enbanc I -0.0208 0.0577 -0.0071 -0.0169 -0.0146 —0.0433 0.0039 I noexper habeas enbanc _________ +_________.__.__---____.__.__-_- noexper I 1.0000 habeas I 0.1337 1.0000 enbanc I 0.0950 -0.0010 1.0000 Because no correlation exceeds 1070, we may safely assume that we have no multicollinearity. INFLUENTIAL CASES Consider the attached graphs. From Graph 4.1 we find that several fact patterns do seem to deviate significantly from the rest of the cases. In Graph 4.2 we obtain the amount of influence attributable to each of the outlying cases. As one can see, there are a few cases exerting a disproportionate amount of influence. I identified those cases and deleted them in order to determine whether their inclusion was adversely affecting our results. In so doing, I do indeed obtain “better” results. However, because it is atheoretical to do so, I do not delete them, but merely comment that there are a few cases that are outliers and so the results reported in this chapter would be stronger were it not for those influential cases. 107 H-L dX"2 GRAPH 4.1 CHECKING FOR INFLUENTIAL CASES 60 '1 I ‘ t) I i 40 j J 20 O f) \ O i mintt‘fll DC) 23 i Oi‘ 37'1“: h I "IUOLJCDDKDOM T r+v*7- _ , , . .5 _ . , 7 .. - .7777 7 .. 4—- ~_ ‘—“T_" 0 .5 1 Pr(confess) _ T j T) 108 1 5.2319 l \:«. H-L d0 -1 4.4807 g . 1 .7e-O7 GRAPH 2 INFLUENCE ANALYSIS if \y / 90. Li) a“ 3 Pr(confess) 109 CHAPTER 5: WHEN THE SUPREME COURT SPEAKS - AN ANALYSIS OF LOWER COURT TREATMENT OF SUPREME COURT DECISIONS This chapter offers yet another way to analyze the impact of the Supreme Court on decision making in the Courts of Appeals. Here, I trace lower court treatment of two of the Supreme Court’s decisions on confession, Townsend v. Sain (372 US. 293 ( 1963)) and Brutan v. United States (391 US. 123 (1968)), neither of which is a salient case. They differ from one another, however, and thereby allow for a comparative analysis, in that precedent established in T awnsend was followed to a significant degree by the Courts of Appeals (as seen in the last chapter), while Brutan was not. By taking a closer look at the actual treatment of these precedents by the Courts of Appeals we may be able to uncover what is different or similar between these cases and suggest why, in the previous analysis, the Court’s decision in Brutan did not exert a significant influence on Courts of Appeals decision making. In this chapter I “Shepardize” both Townsend and Brutan, reading each confession case that treats these two Supreme Court cases to determine whether the lower court is indeed paying heed to the Supreme Court prescription, or whether, through distinguishing the Supreme Court precedent from the case being decided, the Courts of Appeals avoid application of a distasteful precedent. While this analysis is not quantitative in nature, I do think it provides a deeper insight than can be gained in a quantitative fashion. In fact, qualitative research can often provide reasons for phenomena we have difficulty understanding via a multivariate analysis (King, Keohane and Verba 1994). 110 5.1 VERTICAL STARE DECISIS It is understandable that judges themselves disagree over the deference owed higher court rulings. One judge, openly hostile to the suggestion that he follow whatever the Supreme Court decided, said: ...it cannot rationally be said that the state courts of all the states must necessarily follow every twist and turn of federal decisional law as it has developed in the many cases setting forth divergent views (Manwaring l972:8). Other judges join in the spirit of this judge’s open defiance, making such statements as “[w]e follow them when we can’t get around them” (Baum 19782212); “[w]e cannot assume that the Supreme Court intended to deprive the highest court of an independent sovereign state of one of its traditional powers... the right to exercise a sound judicial discretion” (Murphy 1959:1021); or “[t]his Court’s independent review of the relevant historical documents. . .convinces it that the United States Supreme Court has erred in its reading of history” (Caminker 1994:819). More common, however, are statements from jurists sharing the spirit of this Wisconsin judge: “We are bound by the results and interpretations. . .in these high court decisions. Ours is not to reason why; ours but to review and apply” (State v. Nusbaum, 198 N.W.2d 650 at 653 (1972)). And that of a Florida justice who said “[w]e deem it to be our inescapable duty to abide by this decision of the United States Supreme Court [Brawnn] interpreting the federal constitution” (Murphy 1959:1020). Learned Hand offered a similar sentiment: “I have always felt that it was the duty of an inferior court to suppress its own Opinions and. . .try to prophesy what the appellate courts would do” (Baum 19782212). Other federal judges agree: “The law as enumerated by the Supreme Court of the United States must be obeyed by all of us. If we do otherwise we will 7"Brown v. Board of Education, 347 US. 483 (1954). 111 destroy our present form of Constitutional government” (Evans v. Buchanan, 256 F.2d 688 at 695). Reputedly, lower court judges try to interpret the Supreme Court’s will as closely as possible, leaving open the possibility that they sometimes err: “If the view we have expressed is incorrect there will be opportunity for its correction when the Supreme Court has spoken” (Federated Mutual Implement and Hardware Insurance C a. v. Fairfax Equipment Company, Inc., 261 F.2d 207 at 207). They may also feign an inability to comprehend the current doctrine: “Federal laws? Nobody understands federal laws” (Johnson and Canon 1984:45; see also Murphy 1959). Further, some judges even attempt to anticipate what the Court will decide in the filture in order to comply with it prospectively (Wasby 1972, Gruhl 1980). Clement Haynsworth, Nixon’s original choice to fill the F ortas vacancy, illustrates this position in the language cited by the High Court in its overruling of McNalIy v. Hill, 293 US. 131 (1934): “Chief Judge Haynsworth reasoned that this Court would no longer follow McNally, which in his view represented a ‘doctrinaire approach’ based on an ‘old jurisprudential concept’ which has been ‘thoroughly rejected by the Supreme Court in recent cases.”’ (Peyton v. Rowe, 391 US. 54 (1968) at 57). They agreed with him. As Caminker eventually concludes, hierarchical adherence to precedent comports with the Constitution, and is probably necessary to a well-filnctioning judiciary.79 Most studies conclude similarly; that the lower courts do indeed comply with Supreme Court decisions for the most part (Sanders 1995; Sanger and Sheehan 1990; Johnson 1987; Songer 1987b; Stidham and Carp 198280; Gruhl 1980; 7"We, of course, offer no normative argument as to the benefits or detriments of vertical stare decisis. ”And theirs is a study of trial courts. 112 Baum 1978; Murphy 1959). Not many take the “make my day” argument espoused by Akar (1994).“1 However, compliance does, in fact, leave room for substantial influence, and permissive Supreme Court policies may well encourage innovation (Glick 1994). As one scholar put it, “There are many contexts in which the latitude of those charged with carrying out a policy is so substantial that studies of policy implementation should be turned on their heads. In these cases, policy is effectively ‘made’ by the people who implement it” (Johnson and Canon 19843). My analysis will determine whether the Courts of Appeals innovate, comply, or disregard the Supreme Court’s mandate in the confession cases, at least with reference to two of those confession cases — codefendant’s confessions and plenary hearings in habeas corpus. 5.2 COMPLIANCE AND IMPACT 1N JUDICIAL SCHOLARSHIP The literature on judicial impact and compliance is voluminous, albeit somewhat contradictory. Canon contends that “[t]he impact literature is replete with hypotheses stemming from common sense suggestions, case studies or limited comparisons” (1973: 130). Much proliferation has occurred since his writing which has confounded findings even more. Because decision making in the Courts of Appeals is particularly complicated (Songer and Haire 1992), many variables plausibly exert an influence upon a lower court’s decision to dispose of a case in a certain manner. In most cases, the lower court has Supreme Court precedent to apply and interpret; its own policy preferences to pursue; and the relevant law, case facts, litigants, and arguments with which to contend. "Amar, quoting another legal scholar, said that perhaps an unhappy lower court judge should (again, we do not deal in nonnatives here) say to the higher court: “Go ahead, make my day: reverse me. You’re wrong about the Constitution. You’ve taken your oath of office, but that’s no excuse for my violating mine. I’m going to follow my oath of office and decide the Constitution correctly as I understand it. If you don’t like it, take cert” (1994241). While admittedly a bit drastic, it does present an argument seemingly congruent to that of some of the state court justices cited above and elsewhere. 113 The model offered in the last chapter attempted to measure Supreme Court impact by determining the extent to which variables identified as necessarily mattering to the determination of confession voluntariness by the Supreme Court actually matter to the Courts of Appeals. There, we found that the Courts of Appeals do indeed heed the High Court, as several precedential facts were considered by the Court of Appeals in its confession-case decision making. However, a lower court needn’t completely thwart Supreme Court precedent in order to influence its impact. In fact, the lower courts have several options available to them. They can interpret the decision narrowly, basically limiting it to its specific facts (Johnson and Canon 1984). They may cite their own opinions in lieu of citing the offending precedent (Manwaring 1972). They may attempt to distinguish their case from the one for which Supreme Court prescription is available (Caminker 1994; Songer and Sheehan 1990; Baum 1978; Tarr 1977; Manwaring 1972). They may dispose of the case on procedural grounds (Johnson and Canon 1984); they may criticize the Supreme Court while following it (Tarr 1977); or they may simply ignore the offending precedent’s existence. By examining treatment of a couple of Supreme Court cases very closely, we are able to capture this differential compliance and determine whether the lower courts try to avoid the application of Supreme Court precedent. In addition, we can determine whether those lower courts that treat a statistically insignificant precedent (according to Chapter 4) nonetheless comply with the precedent. This adds a depth of analysis that compliance and impact research often overlooks. In addition, it allows for a discussion of differences among the circuits, citations of the circuits to each other, and differences among the panels within a circuit. And does so using non-salient cases, which is clearly the exception in impact and compliance work. In short, this sort of exploration may identify relationships worthy of future study and consideration. 114 5.3 SUPREME COURT CASES UNDER ANALYSIS As mentioned earlier, I have chosen the two precedents under consideration for two reasons. First, neither is a salient Supreme Court decision. Much impact analysis has been directed solely at salient or landmark cases, even though we may expect lower courts to adhere to a highly visible case like Miranda. Indeed, much research has been done on that decision, all of which concludes that the Courts of Appeals exhibit a high level of compliance with the Court’s Miranda prescriptions (Beatty 1972; Cannon 1973; Wasby 1973; Romans 1974; Sanger and Sheehan 1990). By selecting non-salient cases for analysis, I study an environment in which the Courts of Appeals may feel freer to disregard the High Court’s rulings. The second reason for my choice of Supreme Court cases is that one was determined to be significant in the analysis in Chapter 4 and one was not. This allows for potential explanation of the differential treatment by the Courts of Appeals. Each of these cases, while not “salient” by Court scholars’ standards, involves a formal alteration of precedent in an area in which strong views prevail. They involve criminal procedure and the rights afforded to those who would appeal their convictions. Surely judges have attitudes on such a matter, and liberal and conservatives most assuredly differ, if indeed attitudes influence the decision making of Courts of Appeals judges. Further, as we shall see in the next section, both Supreme Court decisions provide the Courts of Appeals with “wriggle room.” Neither is a model of clarity. Reviewing courts, therefore, should be able to find reasons for affirming the petitioners’ convictions, should they so desire. 115 5.3.1 TOWNSEND V. SAIN, 372 US. 293 (1963) Townsend was convicted of murder and sentenced to death. After exhausting state remedies, he sought a writ of habeas corpus, claiming that his conviction violated the Fourteenth Amendment as his confession, obtained while under the influence of drugs - including what some view as a truth serum — was admitted over his objection. The District Court denied his application for habeas corpus without a hearing, and the Court of Appeals affirmed that denial. The Supreme Court reversed, deciding that the lower courts erred in denying the writ without a plenary evidentiary hearing. In its opinion, the Court presents several conditions under which an accused, on habeas corpus, must receive a plenary hearing before the district court. First, “. . .where an applicant for a writ of habeas corpus alleges facts which, if proved, would entitle him to relief, the federal court to which the application is made has the power to receive evidence and try the facts anew” (at 312). Considerations which in certain cases may make exercise of that power mandatory are these: “where the facts are in dispute, the federal court in habeas corpus must hold an evidentiary hearing if the habeas applicant did not receive a full and fair evidentiary hearing in a state court, either at the time of the trial or in a collateral proceeding. In other words, a federal evidentiary hearing is required unless the state-court trier of fact has, after a full hearing, reliably found the relevant facts” (at 312- 313). The Court fiIrther particularizes the controlling criteria. “We hold that a federal court must grant an evidentiary hearing to a habeas applicant under the following circumstances: If ( 1) the merits of the factual dispute were not resolved in the state hearing; (2) the state factual determination is not fairly supported by the record as a whole; (3) the fact-finding procedure employed by the state court was not adequate to 116 afford a filll and fair hearing; (4) there is a substantial allegation of newly discovered evidence; (5) the material facts were not adequately developed at the state-court hearing; or (6) for any reason it appears that the state trier of fact did not afford the habeas applicant a full and fair fact hearing” (at 313-314). A hearing may be avoided if the allegation of newly discovered evidence is “fiivolous or incredible” (at 317), or if the evidence not developed was due to “the inexcusable neglect of the petitioner” (at 317). Given this set of rules, I coded each confession case in the Court of Appeals which treated T awnsend as either followed or not followed. Those that follow will either grant an evidentiary hearing, or provide one of the Supreme Court’s reasons for not doing so (either the claim is fiivolous, or the lack of evidence is due to petitioner culpability). Those coded as not following either simply refirse to grant an evidentiary hearing, or invalidly distinguish their case from Townsend. The Court, however, treated Townsend in two subsequent cases, according to Shepard ’s. In Colorado v. Connelly, 479 US. 157 (1986), the Court distinguishes the facts in Townsend in order to rule on the voluntariness of a confession. So here, the Court is dealing with Townsend as a confession case rather than as a specification of when a plenary hearing in a habeas corpus case should occur.82 The Court’s second subsequent treatment of T awnsend more substantially relates to our current concerns. In Keeney v. Tamaya-Reyes, 504 US. l (1992), the Court 82 In this case, a person, suffering from some sort of mental incapacitation, attempted to render his voluntary statement to a policeman involuntary due Simply to his mental state. The Court says that coercive police activity is a necessary predicate to finding a confession involuntary. It distinguishes Townsend, wherein the petitioner was given a truth serum, from the case at bar because in Townsend, there was evidence of police coercion. Here, there was none. 117 overruled a portion of the Townsend precedent. Respondent, a Cuban immigrant with almost no knowledge of English, collaterally attacked the plea he made in a state proceeding, arguing that his interpreter did not adequately interpret the mens rea element of manslaughter. Because of this, he contended that he did not know he was pleading no contest, but rather thought he agreed to be tried for manslaughter. The State Court of Appeals affirmed, and the State Supreme Court denied review. Respondent then sought a writ of habeas corpus from the District Court, contending that material facts were not adequately developed, implicating the fifth circumstance of T awnsend, and thereby entitling him to a federal evidentiary hearing on the constitutionality of his nala cantendre plea. The District Court ruled the Situation a matter of petitioner neglect and denied the petition. The Court of Appeals reversed, ruling that Townsend and Fay v. Naia, 372 US. 391, required an evidentiary hearing “unless respondent had deliberately bypassed the orderly procedure of the state courts” (at 4). The Supreme Court overruled those portions of Townsend and Fay that prescribed the “deliberate bypass standard [as] the correct standard for excusing a habeas petitioner’s failure to develop a material fact in state-court proceedings” (at 5). Instead, the Court required use of the “cause-and-prejudice” standard: “As in cases of state procedural default, application of the cause-and-prejudice standard to excuse a state prisoner’s failure to develop material facts in state court will appropriately accommodate concerns of finality, comity, judicial economy, and channeling the resolution of claims into the most appropriate form” (at 8). Thus, I expect that the Courts of Appeals will behave differently after the Supreme Court announced this change in precedent. 118 5.3.2 BRUTONV. UNITED STATES, 391 US. 123 (1968) In this case, the joint trial of Bruton and one Evans resulted in their convictions for armed postal robbery. Evans did not take the stand, but a postal inspector testified as to his confession, in which Evans implicated Brutan. The trial judge instructed the jury that the confession could be used only against Evans, as per the Supreme Court’s decision in Delli Paoli v. United States, 352 US. 232. Both appealed the decision and, on appeal, the Court of Appeals set aside Evans’ conviction on the basis that the confession should not have been used against him. It affirmed Bruton’s conviction, however. The Supreme Court, overruling Delli Paoli, held that the admission of Evans’ confession without his testifying robbed Brutan of the opportunity to cross—examine him, thereby violating the Confrontation Clause of the Sixth Amendment. “Plainly, the introduction of Evans’ confession added substantial, perhaps even critical, weight to the Govemment’s case in a form not subject to cross-examination, since Evans did not take the stand. Petitioner thus was denied his constitutional right of confrontation” (at 127-128). “Delli Paoli assumed that this encroachment on the right to confrontation could be avoided by the instruction to the jury to disregard the inadmissible hearsay evidence. But, as we have said, that assumption has since been effectively repudiated. . .We. . .held [in Jackson v. Denna, 378 US. 368 (1964)] that a defendant is constitutionally entitled at least to have the trial judge first determine whether a confession was made voluntarily before submitting it to the jury for an assessment of its credibility. More specifically, we expressly rejected the proposition that a jury, when determining the confessor’s guilt, could be relied on to ignore his confession of guilt should it find the confession involuntary” (at 128-129). “The fact of the matter is that too often such admonition against misuse is intrinsically 119 ineffective in that the effect of such a nonadmissible declaration cannot be wiped fiom the brains of the jurors. The admonition therefore becomes a filtile collection of words and fails of its purpose as a legal protection to defendants against whom such a declaration should not tell” (at 129 citing the dissent in Delli Paali, Ibid at 247). “The naive assumption that prejudicial effects can be overcome by instructions to the jury. . .all practicing lawyers know to be unmitigated fiction” (at 129, citing the concurrence in Krulewitch v. US. 336 US. 440 at 453). Given the foregoing, the lower courts are compliant with Brutan only if they rule that the confession of a codefendant may not be introduced against the other even though the jury has been instructed not to consider it. However, as we shall see, there are several aspects of such confessions that the Court’s decision does not cover; hence valid distinguishings arguably abound. In fact, the Court itself has distinguished Brutan several times, in effect, limiting its applicability.83 In Frazier v. Cupp, 394 US. 731 (1969), the Court found that an opening statement relating to a codefendant’s confession did not violate the defendant’s confrontation rights. The Court said that, in this case, because the prosecutor expected to produce the evidence as the codefendant was scheduled to testify, and because the prosecutor did not assert that the evidence was essential to his case, and because the court did give limiting instructions, no constitutional rights were infringed. “It may be that some remarks included in an opening or closing statement could be so prejudicial that a finding of error, or even constitutional error, would be unavoidable. But here we have no 83 Shepard 's lists Idaho v. Wright, 497 US. 805 as distinguishing Brutan, but I see it as following the holding in Brutan. 120 more than an objective summary of evidence which the prosecutor reasonably expected to produce. . .At least where the anticipated, and unproduced, evidence is not touted to the jury as a crucial part of the prosecution’s case, it is hard for us to imagine that the minds of the jurors would be so influenced by such incidental statements during this long trial that they would not appraise the evidence objectively and dispassionately” (at 73 6, references omitted). In Duttan v. Evans, 400 US. 74, (1970), the Court held that where a witness testifies, no confrontation question exists. “This case does not involve evidence in any sense “crucial” or “devastating,” as did all the cases just discussed. It does not involve the use, or misuse, of a confession made in the coercive atmosphere of official interrogation, as did Douglas, Braakhart, Brutan, and Roberts. . .It does not involve a joint trial, as did Brutan and Roberts. . .The witness was vigorously and effectively cross-examined by defense counsel. . .His testimony, which was of peripheral significance at most, was admitted in evidence under a coconspirator exception to the hearsay rule long established under state statutory law. . .From the viewpoint of the Confrontation Clause, a witness under oath, subject to cross-examination, and whose demeanor can be observed by the trier of fact, is a reliable informant not only as to what he has seen but also as to what he has heard” (at 87). In 1985, the Court said that the introduction of a codefendant’s confession for the nonhearsay purpose of rebutting respondent’s testimony that his confession was a coerced copy of his codefendant’s was affirmed (Tennessee v. Street, 471 US. 409 (1985)). The State relied on a confession that respondent made to the Sheriff. Respondent asserted that his confession was coercively derived from his accomplices written confession, as read to 121 him by the Sheriff. In rebuttal, the State called the Sheriff, and in so doing, the Sheriff read the confession of the co-defendant to prove differences between the two statements. The Court held that “the nonhearsay aspect of the accomplice’s confession — not to prove what happened at the murder scene but to prove what happened when respondent confessed — raises no Confrontation Clause concerns. The Clause’s firndamental role in protecting the right of cross-examination was satisfied by the Sheriffs presence on the witness stand” (at 413-414). The trial judge did give instructions on the use of the accomplice’s confession, and these were deemed adequate. In 1987, the Court decided that it is acceptable to admit the confession of a codefendant so long as it is redacted as to the other defendant (Richardson v. Marsh, 481 US. 200, (1987)). In this case, a codefendant’s confession was admitted only as to said codefendant, and all reference to petitioner was deleted. In fact, the confession was redacted such that it appeared that only the confessor and another person were involved in any criminal activity. Petitioner contends that, in addition to other admissible evidence, the jury could infer the presence of the petitioner in her codefendant’s statement. Over a Brennan/Marshall dissent, the Court “hold[s] that the Confrontation Clause is not violated by the admission of a nontestifying codefendant’s confession with a proper limiting instruction when, as here, the confession is redacted to eliminate not only the defendant’s name, but any reference to his or her existence. We express no opinion on the admissibility of a confession in which the defendant’s name has been replaced with a symbol or neutral pronoun” (at 211). Incidentally, the Court decided that Brutan was to be applied retroactively (Roberts v. Russell, 392 US. 293 (1968)), and that errors under Brutan were subject to 122 harmless error analysis (Harrington v. California, 395 US. 250 (1969)). The lower courts also cite other Supreme Court cases in support of their distinguishing a given case from the Supreme Court’s policy prescription in Brutan (Nelson v. 0 ’Neil, 402 US. 622 (1971), Cruz v. New York 481 US. 186 (1987), Parker v. Randolph, 442 US. 62 (1979), and California v. Green, 399 US. 149 (1970)). These will be discussed as we discuss the treatment of Brutan by the various circuits. While these decisions may have affected the lower courts’ treatment of Brutan, they did not formally treat Brutan, at least according to Shepard ’s. 5.4 IDENTIFICATION OF COURTS OF APPEALS CASES Tarr (1977) offers usefirl definitions of compliance and noncompliance, which I employ in this study in order to identify relevant Courts of Appeals decisions for inclusion in my analysis. Compliance involves “. . . proper application of standards enunciated by the Supreme Court in deciding all cases raising similar or related questions” (1977:35), while noncompliance “. . .involves a failure to apply — or properly apply — those standards” ( 1977 :35). Using these definitions, I selected cases via Shepard ’s United States Supreme Court Citations. 8‘ Songer cautions against the use of Shepard ’s in impact case selection, for he says that “The impact of a given Supreme Court decision should be assessed by examining all decisions below in which the principles announced by the Court are relevant to the “I have necessarily excluded unpublished decisions (for problems inherent in this exclusion see Sanger and Davis 1989; Pacelle and Baum 1992). Upon a quick reading of such cases, there appears to be no systematic bias resulting from this exclusion. (A LEXIS search was conducted using the name of the Supreme Court decision and the word “unpublished.”) 123 resolution of the issues presented to the lower court” (198732570). He cites the possibility that lower courts merely ignore Supreme Court precedent as a means of noncompliance. While I agree theoretically, several scholars have found evidence to mitigate the seriousness of this problem for studying the Courts of Appeals. In order to determine whether lower courts were employing this strategy, Reddick and I identified common West Key numbers between the overruled and overruling decisions, ascertaining the issues that were the basis of the overruling,85 this being a defensible set of cases that should cite Supreme Court precedent. Thus, we obtained every lower court decision under those keys from the year of the overruling decision to the present. This generated several thousand cases. We examined a sample of these lower court decisions to determine if in fact they ignored the change in precedent and thus were noncompliant -— behavior which Shepard ’s Citations would fail to identify. We found not a single opinion that overtly ignored an overruling decision. On the basis of this search, we concluded that the Courts of Appeals do not simply disregard an objectionable precedent, but rather find more subtle ways to avoid it. These courts may well differ from the state supreme courts Songer used in his illustration, and, as the Benesh-Reddick analysis and other studies have shown (Songer 1987b; Sanger and Sheehan 1990), the intermediate federal courts do not openly avert Supreme Court precedent. In addition, a recent study has found the coding done by Shepard ’s to be quite valid and reliable, especially for the very positive (followed) and very negative treatments ”Some of the overruling and overruled decisions had no key numbers in common, as the Key Number System has evolved over the years, adding some keys and removing others. In these cases, we chose key numbers from the overruling decision which pertained to the basis upon which the Court was altering its prior decision. 124 (overruled, limited) (Spriggs and Hansford 1999). Although they find that those multifaceted or neutral treatments should probably be coded for their validity (distinguished, explained, harmonized) that they found the others to be both valid and reliable is a good indication that our fears in using Shepard ’s have been largely unfounded. Thus my use of Shepard 's Citations is appropriate. I utilized Shepard ’s to identify the Courts of Appeals’ treatment of Supreme Court decisions. In order to assess a circuit court’s behavior of either Brutan or T awnsend (either followed, limited, criticized, questioned, or distinguished), I carefully evaluated the courts’ opinions. My inquiry did not address the propriety of whether the Court of Appeals should have provided the specified Shepard ’s treatment to either Brutan or Townsend. But rather, whether the Shepard ’s treatment accurately characterized the precedent in light of the facts of the Court of Appeals case. I caded all of the circuit court cases as having either “followed” or “not followed” an overruling decision, or eliminated the citation from the analysis.86 5.5 COURTS OF APPEALs’ COMPLIANCE WITH BRUTON AND TOWNSEND I proceed on a circuit-by-circuit basis, beginning with Brutan. Tables 5.1 and 5.2 contain a summary of circuit treatment of the two precedents. In the following two sections, I assess the way the individual circuits treated the Supreme Court’s decisions in Brutan and Townsend. I will not discuss every case heard 86 I eliminated all cases containing a “neutral” treatment from Shepard’s (e. g., explained, harmonized, or cited), as I am not concerned with said cases. I also eliminated all cases with a reference to the Supreme Court precedent only within either a different opinion (e. g., a concurrence) or a dissenting opinion. Finally, I eliminated any case that was not based upon a coerced confession. 125 by each of the circuits; rather, I will briefly summarize the “valid” distinguishings made by each circuit; i.e., those I have coded as “followed, ” and then discuss those cases which I have deemed to “not follow” the relevant Supreme Court decision. Patterns should be discovered within the circuits and conclusions duly drawn. Do note that I accept Shepard ’s decision when it codes a case as “following;” therefore, I do not discuss these cases.87 BRUTON v. UNITED STATES 5.5.1 THE FIRST CIRCUIT The first circuit treated Brutan in five confession cases. In four of those cases, Shepard ’s coded the circuit as having distinguished Brutan, and in one, it coded it as following the Supreme Court precedent. (Again, I accept Shepard ’s determination of followed and do not discuss said cases here.) I find the first few cases to dutifully follow Brutan, while the most recent seemingly avoids the Court’s precedent. In the distinguishing cases that I have coded as following Brutan, the circuit differentiated between its decision and the Court’s for the following reasons: lack of a jury (US. v. Castro, 413 F.2d. 891 (1969)); a codefendant who was indeed available for cross- examination (LaFrance v. Bahlinger, 499 F.2d. 29 (1974)); and the presence of a conspiracy, thereby invoking an exception to the hearsay rule (Grieco v. Meachum, 533 F.2d. 713 (1976)). However, I judge the last of the First Circuit’s cases not to follow Brutan (US. v. Smith, 46 F.3d. 1223 (1995)). In this case, a codefendant called an authority on 87 As mentioned in previously, Benesh and Reddick (1998) tested for the validity of Shepard ’s “followed” coding, as did Spriggs and Hansford (1999). Both found that 126 professional responsibility to the stand. Before he did so, the court indicated that his testimony applied only to Cohen, one of the defendants. This authority testified that Cohen had come to him with a professional responsibility problem. “Documentation which he had prepared and which was complete and on file had been changed by three people [identified in the testimony as a former ofiicer, a former director, and a manager of a bank] in their indication to him when they met with him” (at 1228). In other words, they told Cohen that they had changed the documents that he had dutifirlly finished and filed. Cohen’s codefendants moved for a mistrial citing Brutan. “The court submitted the case to the jury because the Brutan error (if any) occurred during the last day of testimony in a lengthy trial, and might be mooted by an acquittal. In addition, the harmfirlness of the error would be more apparent in light of the verdicts” (at 1228-1229). All of the defendants were convicted and the district court found that there had been an error under Brutan, but that said error was harmless beyond a reasonable doubt. The first circuit agreed, saying that “Admittedly, Cohen’s statement might tend to incriminate Smith and Devaney [the codefendants] by showing that the co- conspirators met to discuss damage control. In this sense, however, the statement falls far outside the pale of the ‘powerfiilly incriminating’ evidence that produces Brutan errors” (at 1229). The court concluded that it was “merely cumulative of the govemment’s case” (at 1229). Since the jury convicted even Cohen, it must not have believed him that the others admitted their conspiracy to him. Therefore, the district court was affirmed. While the Court has advocated the use of redacted confessions, it seems that in this case the jury Shepard ’s designation comported with standard definitions of “following.” 127 could easily infer that the codefendants were those named by the Professor. Therefore, I consider this case not to have followed Brutan. Thus, in four of its five cases the First Circuit conformed to Brutan. 5.5.2 THE SECOND CIRCUIT The second circuit treated Brutan 18 times, 12 of which Shepard ’s identified as distinguished and 6 followed. I have coded all 18 as following Brutan, although I do discuss one that is on the edge of compliance. In the followed cases that Shepard ’s distinguishes, the lower court draws distinctions between its case and the Supreme Court’s, such as the following: though the complaining codefedant confessed, independent evidence of his guilt was overwhelming (US. v. Wallack, 414 F .2d. 246 ( 1969)); the error was harmless due to the wealth of evidence against the accused (US. v. LaVallee, 415 F .2d. 150 (1969)); the well- established co-conspirator hearsay exception (US. v. Littman, 421 F.2d. 981 (1970)); the confessing codefendant took the stand (US. v. Mancusi, 441 F .2d. 1073 (1971)); and the redacted statement in no way inculpated the co-defendants, in fact not referencing other persons at all (US. v. Truda, 449 F.2d. 649 (1971). Ifind US. v. Fallette, 430 F.2d. 1055 (1970), to be a close call. Nelson (appellant) and Biggins were jointly tried and convicted of murder. Nelson claimed a Brutan violation because two confessions made by Biggins were introduced at the joint trial, and Biggins did not take the stand. Biggins’ statement did not name Nelson as his colleague, but referred to a man known as “Oliver.” Biggins, however, gave a physical description of Oliver that more or less fit Nelson’s description. The circuit relied on the presence or absence of a goatee as evidence that Nelson was not “clearly inculpated by his 128 general resemblance to ‘Oliver’ at the time of trial” (at 1057). The circuit reasoned that these statements “were not the type of ‘powerfully incriminating’ statements to which the Court had reference in Brutan” (at 1058). Biggins’ said that he had run into Oliver earlier in the evening and that they then proceeded to the Uptown Bar where the robbery and murder occurred. If the jury believed that Nelson was Oliver, surely this is incriminating evidence. Nonetheless, the circuit said it was not a vital part of the government’s case, and that there was other evidence linking Nelson to the crime, including the identification of eyewitnesses. In addition, Biggins’ statement that Oliver was the principal actor in the murder was directly contradicted by the eyewitnesses, who all agreed that Biggins was primary robber and did the actual killing. The judge did “carefirlly instruct. . .the jury that these statements could be considered only with respect to the defendant Biggins and that they were not to be considered in any way with respect to the innocence or guilt of the defendant Nelson” (at 1059). “We find the possibility that the jury made an inference from [the] testimony that it would not have made in any event, and that this inference made the difference between acquittal and conviction for [appellant], far too remote to meet the Brutan tests” (at 1059, citations deleted). While I code this decision as following, some may disagree. Because the circuit made valid distinctions, and in some cases even made stronger prescriptions than the Supreme Court (US. v. Danzey, 594 F.2d. 905 (1979)), I consider the Second Circuit compliant with Brutan. 129 5.5.3 THE FOURTH CIRCUIT“ The fourth circuit treated Brutan in 7 confession cases, four of which distinguished and three of which followed, according to Shepard ’s. I judge them all compatible with Brutan. The circuit distinguished for various reasons. (1) There was an adequate redaction and so the confession does not refer specifically to the codefendants (Close v. US, 450 F.2d. 152 (1971), and US. v. Grant, 549 F.2d. 942 (1977)); (2) the admission of the fact that the codefendant remained silent in the face of accusations by his accomplice was harmless due to the overwhelming evidence of guilt (Miller v. Cox, 457 F .2d. 700 (1972), and Falstan v. Allsbraak, 691 F.2d. 184 (1982)); (3) the confessions were interlocking, as discussed by a plurality of the Supreme Court in Parker v. Randolph, 442 US. 62 (1979) (US. v. Smith, 792 F.2d. 441 (1986)); and (4) the testimony was not incriminating against the codefendant (US. v. Crockett, 813 F.2d. 1310 (1987)). In US. v. T ruslow, 530 F.2d 257 (1975), the Fourth Circuit scrupulously guarded the codefendant’s right to confront witnesses against him or her. While they do not find Brutan violations in all of the statements admitted at trial, they deem the cumulative effect of all of the admissions to be such that the defendants were deprived of their constitutional right to confrontation. They deem the failure of the trial court to order severance or take other preventive action to be enough to reverse its judgment. The Fourth Circuit correctly distinguished its cases from Brutan, and in some cases, went even firrther than the Supreme Court decreed in protecting an accused’s confrontation rights. 88 The Third Circuit did not treat Brutan in any of its confession cases. 130 5.5.4 THE FIFTH CIRCUIT The Fifth Circuit treated Brutan in more confession cases of any of the others — 37. Shepard ’s coded 26 as distinguishing Brutan and 11 as following it. I have coded two of the Circuit’s decisions as failing to follow Brutan, in that they avoided the application of the Brutan precedent Valid distinctions made by this Circuit include the following: the statement did not refer to the codefendant (White v. US, 415 F .2d. 292 (1969)); the codefendants all testified (Roberts v. US, 416 F.2d 1216 (1969)); the government did not place the statements into evidence, nor did they incriminate appellant (US. v. Venere, 416 F .2d. 144 (1969)); the statement did not incriminate the defendant (US. v. Gibson, 416 F.2d. 326 (1969), US. v. Pollard, 509 F.2d. 601 (1975), US. v. Grilla, 527 F.2d. 1344 (1979), and US. v. Castro, 596 F.2d. 674 (1979)); the codefendant was available for and was cross- examined (US. v. Johnson, 478 F.2d. 1129 (1973), and US. v. Maddox, 492 F.2d. 104 (1974); the court both redacted the confessions and excluded the statements at issue so that they were not mentioned before the jury (US. v. Wilson, 500 F .2d. 715 (1974)); the codefendant made the statement while in the course of a conspiracy (US. v. Burroughs, 650 F .2d. 595 (1981)); the statements do not reflect guilt on the codefendant (US. v. Coppola, 788 F.2d. 303 (1986)) The first of this Circuit’s noncompliant decisions was Migliare v. US., 409 F.2d. 786 (1969). Because the evidence presented from a statement by Migliare about Mingo (the two codefendants to the case) was made during the commission of a crime, they were admissible. This conforms to Brutan. However, a statement made by Mingo about 131 Migliore was not admitted because “there Migliore made no objection at the trial to the admission of this testimony” (at 788). This arguably violates the spirit of Brutan. The second questionable decision was one in which the circuit, en banc and in a lengthy opinion, determined that Brutan was not applicable (Hoover v. Beta, 467 F.2d. 516 (1972)). At issue was a Texas law that “permits inculpatory references against an accomplice to be heard by the jury, subject to limiting instructions by the court to regard the whole confession as evidence only against the declarant confessor” (at 529). This seems to directly violate the Court’s holding in Brutan. However, the Circuit does have a plausible reason for excepting this case from Brutan ’s holding: the trial was not joint, but rather that of an accomplice, which was severed from the trial of the principals. Because the state makes the guilt of the principals prerequisite to the conviction of an accomplice, the Circuit argued that the admission of the confession was justified and did not violate the accomplice’s confrontation rights. Further, the majority pointed to the examination of the prospective jurors as to their ability to consider a confession in this way protective of the accused’s rights. Therefore, it did not consider the trial court’s conclusion that a limiting instruction could protect the accomplice erroneous. The dissenters, however, asserted that references to the accomplice could surely have been taken out of the confession and the court could have still proved the guilt of the principals while protecting the confrontation rights of the defendant. It is my detemlination that the Supreme Court may well agree with the dissenters and so I coded this decision as not following Brutan. The Court went to great lengths in Brutan to discuss the meaninglessness of limiting instructions, and this Circuit relied upon the curative effect of them in admitting the statements. This is incompatible with the Court’s 132 decision. I think this Circuit, while evidencing some division en banc, is apt to accept reasoning which in effect thwarts Supreme Court precedent. 5.5.5 THE SIXTH CIRCUIT The Sixth Circuit treated Brutan in 18 confession cases, distinguishing 8 cases, following it in 9, and limiting it once. I consider all 18 cases compliant with Brutan, although two arguably are not. The Circuit, however, did assert “This court has sought faithfirlly to follow the teachings of Brutan. . .” (Randolph v. Parker, 575 F.2d 1178 at 1182). It seemingly has done so. Valid distinguishings include the following: the confessing codefendant took the stand (although he did deny making the statement) (US. v. Sims, 430 F.2d. 1089 (1970)); the appellant himself confessed, and the codefendant making incriminating statements also took the stand — there was also much evidence against him apart from the confessions (Farmer v. Krapp, 445 F.2d. 5 (1971)); codefendant was a co—conspirator and the statement was made in fiIrtherance of the conspiracy (US. v. McManus, 560 F.2d. 747 (1977), and US. v. Kendricks, 623 F .2d. 1165 (1980)); and the trial was before a judge not a jury (Rogers v. McMackin, 884 F.2d. 252 (1989)). The two questionable cases are Yates v. US., 418 F.2d. 1228 (1969), and Childs v. Cardwell, 452 F.2d. 541 (1971). In the first, the Court made a redaction that I am not certain would pass scrutiny had the Supreme Court reviewed the case. The trial court had “masked” the written confession so that the other codefendants’ names could not be read by the jury. The clerk then photographed the confession and entered in as evidence. One of the codefendants claimed that his identity was not sufficiently concealed. 133 In the second problematic case, the Circuit did not find for the appellant because he did not object to the admission of the testimony of a police officer as to the statements of a codefendant. The Court argued that, because the codefendant can cross-examine the officer, and additionally, since the codefendant was already convicted, the codefendant could have been called as a hostile witness, the appellant’s confrontation rights were not violated. The Circuit found the error harmless. Nonetheless, the Circuit arguably did adhere to the Brutan holding, and so we again find very little evidence of any noncompliance. 5.5.6 THE SEVENTH CIRCUIT The Seventh Circuit treated Brutan in 20 confession cases, 15 of which Shepard ’3 labeled “distinguished” and 5 “followed.” I deem all of these cases compliant with Brutan. Examples of this Circuit’s distinguishings include: the statements were not incriminating and, further, the confessing codefendant was a witness at the trial and thereby subject to cross-examination (US. v. Hoffa, 402 F.2d. 380 (1968)); the appellant did not move for a severance, an additional codefendant testified (and was cross-examined) to the same things in the admitted statement, the statement was redacted, and the defendant himself confessed (US. v. Pate, 418 F.2d. 1028 (1970)); the agent did not identify a codefendant as his source of the information until the appellant cross-examined him, hence there was no reversible error (454 F.2d. 1208 (1972)); the defendant himself confessed, substantially in the same words as his codefendant (US. v. Spinks, 470 F.2d. 64 (1972)); the co- conspirator exception to the hearsay rule occurred (US. v. Donner et al., 497 F .2d. 184 (1974)); and the statement was redacted and the jury warned twice to use it only against 134 the confessing codefendant (US. v. Kreiser, 15 F.3d. 635 (1994)). This Circuit clearly complied with Brutan. 5.5.7 THE EIGHTH CIRCUIT The Eighth Circuit treated Brutan in 12 cases, 6 of which Shepard ’s classified as distinguished, 5 followed, and 1 limited. One decision is a bit questionable, but the others all validly distinguish themselves from Brutan. Some reasons for the distinctions are the following: both defendants confessed, and the codefendant in question took the stand (US. v. Fountain, 449 F .2d. 629 (1971)); the codefendant took the stand and denied making the statement (US. v. Valenzuela, 521 F.2d. 414 (1975)); and the co-conspirator exception to the hearsay rule (US. v. Kelley et al., 526 F.2d. 615 (1975)). The one case that gives pause is T asby v. US, 451 F .2d. 394 (1971). The Circuit deemed harmless the introduction of in-custody confessions made by each of the defendants. The testifying witnesses avoided any specific reference of one codefendant to the others, except that reference was made to the fact that one of the accomplices had a noticeable eye defect. Only the appellant had such a defect and so was easily identifiable. However, the court concluded this to be harmless error because law enforcement officers had arrested him during his actual participation in the crime. It seems that the Circuit could have remanded the case back to try the appellant without the indirect identification, but chose not to. While I see this as following Brutan for the most part, the opposite argument could be made. Overall then, the Eighth Circuit, like the others, implemented the Brutan decision. 135 5.5.8 THE NINTH CIRCUIT The Ninth Circuit treated Brutan in 15 cases, ten of which distinguished it and five of which followed it. I have determined that all but one of this court’s decisions adhered to Brutan, while two are seemingly indeterrninable. Those that distinguish it do so validly for reasons that include the following: the codefendants took the stand (Santaro v. US., 402 F .2d. 920 (1968), and Rios-Ramirez v. US., 403 F.2d. 1016)); the appellant was tried by the court not by a jury (Cockrell v. Oberhauser, 413 F.2d. 256 (1969)); the codefendant’s statement was used only in the opening statement (US. v. Maurice, 416 F.2d. 234 (1969)); the statement only mentioned “other participants” (US. v. Akers et al., 542 F.2d. 770 (1976)); admission was harmless error because appellant made the same statement rendering the codefendant’s statement merely corroborative (US. v. Espericueta-Reyes, 631 F.2d. 616 (1980)); the challenged statements were made by the defendant himself (US. v. Gonzales, 749 F.2d. 1329 (1984)); and the statement was not hearsay (White v. Lewis et al., 874 F.2d. 599 (1989)). In US. v. Barragan, 453 F.2d. 386 (1971), the Circuit Court completely denies and does not discuss the alleged Brutan violation because “no proper objection was made to lay the basis for the application of Brutan. We think the statements of the two (husband and wife) were admissible on the government’s case in chief’ (at 387). Completely rejecting the claim of a husband and wife that one statement implicated the other is questionable indeed. It seems that this is what circuits will do to avoid the application of precedent to a given situation. I deem this case evidence of the Ninth Circuit’s imperfect compliance. 136 The other two cases with which I take issue are US. v. Vissars, 596 F.2d. 400 (1979), and US. v. Arambula-Ruiz, 987 F.2d. 599 (1993). In the first, I question the court’s holding because the dissent disagrees with a factual statement of the majority. Here, the majority finds that if there were error in the admission of the codefendant’s statement it was harmless, due to the court’s belief that the “evidence of guilt is. . .overwhelming” (at 404). The dissenter disagrees, saying “I do not believe that the Brutan error was harmless beyond a reasonable doubt. The case against Keenberg was strong, but I do not believe that it was overwhelming” (at 404). Because there is no way to determine the accuracy of the majority’s argument, I deem the Circuit to have followed here, noting my reservations. The other debatable case deals the incriminating nature of the statements admitted at trial. The statement was redacted, but both parties agreed that it was inadmissible. “However, they dispute its prejudicial effect” (at 605). “We have found no cases which address the precise issue presented here: whether an incriminating question posed by defense counsel and based upon a defendant’s confession creates reversible Brutan error” (at 605). The Circuit did indeed confront a matter not directly addressed by Brutan. It decided that no error occurred because the question was stricken before it was answered and was followed immediately by a limiting instruction from the judge. This may well be true, but it is a case that could have plausibly gone either way. Merely asking a question may influence a jury’s decision. Here, the Circuit gives the jury the benefit of the doubt; a benefit I am no altogether convinced would also be given the jury by the High Court. All in all, however, the Ninth Circuit follows the others and is largely compliant with the Supreme Court. 137 5.5.9 THE TENTH CIRCUIT The Tenth Circuit treated Brutan in 11 confession cases, ten of which distinguished it, and one that followed it. I consider all to have followed the High Court prescription, although one case may be deemed questionable. Reasons for the Tenth Circuit to distinguish Brutan mostly fall into one of two categories: the codefendant testified (Thomas v. US., 409 F.2d. 730 (1969), McHenry v. US, 420 F.2d. 927 (1970), Duggar v. US., 434 F.2d. 345 (1970), US. v. Gurule, 437 F.2d. 239 (1970), US. v. Steel, 458 F.2d. 1164 (1972), and US. v. Trautman, 458 F.2d. 217 (1972)); or the conspiracy exception to the hearsay rule applied (US. v. Cox, 449 F.2d. 679 (1971), US. v. Montgomery, 582 F.2d. 514 (1978), and US. v. Rogers et al., 652 F.2d. 972 (1981)). The one questionable case is US. v. Jackson et al., 482 F.2d. 1167 (1973). The circuit panel found that the admission of several exhibits did not violate petitioner’s Brutan protections. The appellant argued that the admission of the exhibits improperly stigmatized him without the opportunity for cross-examination. The court found that the exhibits implicated only the one defendant and did not harm the appellant. It said that the exhibits “can in no way be considered so powerfully incriminating to Jackson that limiting instructions would not sufficiently protect his rights” (at 1175). The panel, therefore, found that the limiting instructions were adequately protective of appellant’s constitutional rights. This holding, while seemingly consistent with Brutan, is arguable. In all, however, the Tenth Circuit was quite cognizant of the Court’s holding in Brutan and complied with it. 138 5.5.10 THE ELEVENTH CIRCUIT The Eleventh Circuit treated Brutan in 10 confession cases. In five, Shepard ’s deemed the Circuit to have distinguished its case from Brutan, and in the other half, it asserted that Brutan was followed. 1 find that the Circuit followed Brutan in all but one case, and that another is questionable. In distinguishing Brutan, the Circuit noted the following: the codefendant merely agreed with the appellant’s version of the facts (Westbrook v. Zant, 704 F.2d. 1487 (1983)); the statement was not directly incriminating and it was made in the prosecution’s opening statement which is not admitted as evidence (US. v. Jimenez, 780 F.2d. 975 (1986)); and that the statement was admitted only toward the voluntariness of the confession and so was not hearsay (Cargill v. Turpin, 120 F.3d. 1366 (1997)). One case does not seem to comply with Brutan (Bawden v. Francis, 733 F.2d. 740 (1984)). In this case, the panel decided that the admission of a codefendant’s confession was harmless because the appellant “suggests nothing that he could have gained by cross- examining Hardaway on the subject” (at 757). Since the Supreme Court has found that all have a right to confront witnesses against them, this justification is seemingly inapposite to the Court’s preferred policy. The Circuit claims that the evidence was overwhelming and that the appellant did make the jury aware of the inculpatory statement before the testimony in question. In addition, the prosecution did not mention the problem statement again in his closing. While the Circuit seemed to avoid Brutan in this case, it did follow it in the others, except for one other potentially problematic case (US. v. Arias, 984 F.2d. 1139 (1993)). Here, the Court relied upon a redaction that merely substituted the codefendant’s names 139 with “wife” or “husband.” This redaction hardly seems to maintain the confi’ontation rights of the two codefendants. The Court maintained that the statements were not confessions, but rather “attempts by the parties to exculpate themselves” (at 1142) and therefore, were not essential to the prosecution of their cases. However, the use of confessions redacted only to replace a codefendant’s name with “wife” or “husband” is highly suspect. 5.5.11 THE D.C. CIRCUIT The District of Columbia Circuit treated Brutan in six confession cases. In half it followed Brutan, and in the other half, according to Shepard ’s, it distinguished it. I find that five of the decisions follow Brutan and one does not. In two of the valid distinguishings, evidence was excluded as hearsay and the jury admonished not to consider it (US. v. Burroughs, 935 F.2d. 292 (1991)) and the confession was redacted and a neutral pronoun inserted (US. v. Washington, 952 F.2d. 1402 (1991)); therefore, no Brutan violation occurred. In the invalid distinguishing, the Court admitted the statements of a codefendant merely because they were “reliable.” The panel says “Judd’s statements to Young and Thomas were reliable. Judd’s statements were not contained in a confession to law enforcement officials. He made the statements to lay persons with whom he had no motive or incentive to diminish his role by shifting blame; quite the contrary, Judd’s statements revealed conduct beyond his mere presence in the area where Copeland was seen and shot” (at 741). The Court added that even if these statements were admitted in error, that error was harmless due to the overwhelming evidence against both defendants. It seems that a reliance on the reliability of the 140 statements was not the aim of the High Court, and therefore, I think the D.C. Circuit strayed from the Supreme Court’s policy prescription. TOWNSEND v. SAIN The Circuits’ treatment of Townsend admits of a presentation different hour that I accorded Brutan. In Townsend, the Supreme Court laid out a detailed set of six conditions (as mentioned above), any one of which required a district court judge to provide an accused, on habeas corpus, with an evidentiary hearing. Reading these cases that involve a confession provides a reasonably objective indication of circuit court compliance, or the lack thereof: i.e., the degree of detail the circuit panel’s opinion provides in determining whether or not a T awnsend-type evidentiary hearing is required. Although the High Court specified in detail the conditions that require a hearing, many of those details contain vague and undefined terms; e. g., “firll and fair,” “substantial allegation,” “fairly supported,” “adequately developed.” Thus, I consider Townsend complied with if the circuit panel either returned the case to the federal district court to apply T awnsend or to evaluate its applicability (most always the former) or, if in denying the habeas petititoner’s plea, it carefiIlIy scrutinized lower court records and presented reasons — based on Townsend — why the lower court adhered to the High Court’s strictures. 5.5.12 THE FIRST CIRCUIT The First Circuit treated only one confession case with reference to Townsend. Shepard ’s coded that treatment as distinguishing Townsend, but upon reading, the lower court seems to follow the Supreme Court’s prescription. The court exhaustively recounted the facts in the case, finding that the influence of alcohol was not considered by 141 the state court and so the defendant deserved a hearing on the matter. The First Circuit, then, in its single treatment afforded T awnsend due deference. 5.5.13 THE SECOND CIRCUIT The second circuit treated Townsend in three cases, all of which Shepard ’s coded as following the Supreme Court’s treatment of the matter. All of them reverse the conviction, remanding the case to the lower court for evidentiary hearings. Hence, compliance. 5.5.14 THE THIRD CIRCUIT The third circuit treated Townsend in four confession cases — again all of which are coded by Shepard 's as having followed the Supreme Court precedent. In one case (US. v. Rundle, 389 F.2d. 47 (1968), one of the judges on the panel dissented. In another, (US. v. Yeager, 453 F.2d. 1001 (1971), a justice specially concurred reluctantly agreeing that the Townsend principle was applicable to the case at hand. Given the dissent and the special concurrence, one may conclude that the Third Circuit is somewhat less willing to comply with Townsend than was the Second Circuit, even though it reported no such non- compliant decisions 5.5.15 THE FOURTH CIRCUIT The Fourth Circuit considered three confession cases, with all three following the precedent set in Townsend. In this Circuit, compliance occurred even though two of the cases affirmed lower court decisions. The case that was reversed and remanded construed T awnsend in a manner definitely sympathetic to the petitioner (Edwards v. Garrison, 529 F.2d. 1374 (1975)). 142 5.5.16 THE FIFTH CIRCUIT The Fifth Circuit treated Townsend in ten confession cases, all of which Shepard ’s deemed to follow Townsend. However, in carefully reading even those that Shepard ’s coded as followed, there is evidence that this Circuit was reluctant to comply with the High Court’s prescription. In the first of the confession cases, Allison v. Holman, 326 F.2d. 294 (1963 ), the panel treated Townsend in a rather offhand fashion, doing little more than simply saying — in a conclusory fashion, without elaboration or explanation - that the lower courts had complied with Townsend. I read this as an indication that the Circuit did in fact avoid application of the precedent, by merely affinning summarily the lower court. Perhaps this is a common tactic employed by the Courts of Appeals when it wishes to avoid application of a disapproved Supreme Court precedent. Circuit panels behaved similarly in two other cases: Gatcher v. Beta, 444 F.2d. 696 (1971) and in McMillin v. Beta, 447 F.2d. 453 (1971). While I’ve accepted the panel’s decision in Dempsey v. Wainwright, 471 F.2d. 604 (1973), as followed, it nonetheless applied Townsend in a slipshod fashion. The judges did give reasons why the lower court complied with T awnsend, but they did so in a hair splitting fashion, rather than one consistent with the spirit of T awnsend. In Burns v. Beta, 371 F.2d. 598 (1967), the panel did follow Townsend, however, granting the appellant an evidentiary hearing. But a district court judge on the panel dissented, rather clearly rejecting T awnsend. Overall then, It is rather apparent that the Fifth Circuit only marginally complied with Townsend. 143 5.5.17 THE SIXTH CIRCUIT The Sixth Circuit treated T awnsend in three confession cases, two of which Shepard ’s considers to have distinguished the High Court’s decision, and one to have followed it. Given the care with which the panels evaluated Townsend, I deem all three to adhere to the precedent. 5.5.18 THE SEVENTH CIRCUIT The seventh circuit considered Townsend in four confession cases. Shepard ’s coded three of these treatments as distinguishing it, while the other followed it. I disagree and regard them all as dismissive of Townsend; hence, this Circuit is not in compliance. In the first three cases (US. v. Ogilvie, 337 F.2d. 427 (1964), VanErmen v. Burke, 398 F.2d. 329 (1968), and US. v. Pate, 405 F.2d. 449 (1968)) the court distinguished as a means of avoiding adherence. These panels addressed Townsend only obliquely, viewing it as peripheral to the resolution of the controversies. In the final Seventh Circuit case, Weidner v. Ihieret, 932 F.2d. 626 (1991), a very lengthy opinion affinned the district court judge while again skirting the mandate of T awnsend. The panel had initially remanded the case to the district court to determine the voluntariness of the petitioner’s confession because the “sparse factual findings failed to elucidate the facts underlying Weidner’s [the petitioner’s] confession” (at 627). The panel had suggested that the district judge secure an affidavit from the state trial judge. This he did. Weidner objected. The panel then held that “the law of the case” precluded it from revising the procedure it had originally recommended for resolving the matter. While admitting that the “difficult question is whether we erred . . . when we concluded that the hearing we instructed the district judge to hold could begin and end with ‘an affidavit from 144 the state trial judge explaining his cryptic and ambiguous ruling’” (at 631), the panel nevertheless justified its affidavit procedure as compliant with Townsend because “the habeas corpus statute [sicl] suggests we did not” err (at 63 2). 5.5.19 THE EIGHTH CIRCUIT The Eighth Circuit was highly compliant in its eight decisions that treated Townsend in a confession context. Shepard ’s agrees, identifying seven of the cases as following Townsend. Although one of them does so cursorily (Waadson v. Brewer, 437 F.2d 1036 (1971)), care, sensitivity, and thoroughness characterizes the others, especially Grindstaff v. Bennett, 389 F.2d 55 (1968). Note should also be made of the compliance manifested by the panel in Jamison v. Lackhart, 975 F.2d 1377 (1992), which was decided after the High Court overruled Townsend. Shepard ’s properly identifies the treatment accorded Townsend in this case as questioned. 5.5.20 THE NINTH CIRCUIT The Ninth Circuit, in its four T awnsend decisions, is also quite compliant. Shepard 's codes two as distinguishing and two as following, but I deem them all to be in accordance with the Court’s holding in T awnsend. Hart v. Eyman, 458 F.2d. 334, provides a nice example of how well-meaning circuit court judges can construe T awnsend to come to opposite conclusions regarding its applicability. Here, both the majority and the dissent provide fair and detailed evaluations of Townsend in reaching their disparate results. 5.5.21 THE TENTH CIRCUIT The Tenth Circuit treated T awnsend only once in its confession cases. But it was extremely detailed in its consideration and application (Church v. Sullivan, 942 F.2d 1501 145 (1991)). I have read no case that exceeds the thorough, careful, and sensitive treatment afforded the precedent here. 5.5.22 THE ELEVENTH CIRCUIT The Eleventh Circuit also only treated Townsend once in reference to a confession case. In that case, Stinsan v. Wainwright, 710 F.2d. 743 (1983), the court summarizes Townsend very well and indisputably applies it fairly. Of course, there are no cases in the D.C. circuit concerning this case as it involves only federal court treatment of state court decisions. Habeas corpus petitions in the District of Columbia do not involve considerations of federalism. 5.6 CONCLUSIONS The most obvious and significant conclusion, which is immediately marked, is the lack of real noncompliance. While I have coded some of the Brutan and Townsend cases as “not followed,” in reality, their lack of enthusiasm for the Court’s policy prescription is really only that: a lack of enthusiasm. It is certainly not a vendetta — all out, or even partial — against the implementation of them. Additionally, when the Courts of Appeals do somehow avoid application of a given precedent, they justify their decisions by citing reasonably persuasive reasons. Hence, my analysis is a bit artificial when it deems a certain behavior to be noncompliant. These Courts of Appeals are cognizant of the High Court’s ruling, and they follow it for the most part. Their occasional deviation from that behavior, while notable, is clearly the exception even though these may well have been two cases that the lower courts probably did not much like. Brutan and Townsend were chosen with an eye toward the analysis in Chapter 4, wherein the former was insignificant while the latter was statistically significant. 146 Interestingly, the Courts of Appeals are seemingly less compliant with the one exuding a statistical influence on their confession behavior than they are with the decision which was rendered unimportant in that analysis. One reason for this may be the way I treated Townsend in the last chapter. There, I used it as a decision regarding the procedure to be employed by the court in determining admissibility of a confession. The fact interacted with that decision is whether the procedure employed was fair or not.89 However, remanding for an evidentiary hearing is only part of the Courts of Appeals’ finding with respect to fair procedure. The trial court must also follow Jackson v. Denna, supra, and provide a hearing outside the presence of the jury in order to obtain a “yes” answer to the fairness question. This may have driven the result in the last chapter and thereby renders the two findings compatible. In the case of Brutan, I do not find it particularly surprising, given my findings in this chapter, that the decision was not significant in the Courts of Appeals decision-making model presented in Chapter 4. In this chapter, I found that the Supreme Court’s ruling did in fact leave substantial room for the lower courts to interpret and limit the decision. The Courts of Appeals employed a wide range of “valid” distinguishings when treating this case. Because I did not code for such exigencies, due to the fact that these distinguishings were too variegated and each too infrequent to admit of reliable or valid coding, these distinctions could not be included in the statistical model. Therefore, I get a non-finding with respect to the use of a codefendant’s confession, when, in fact, the lower courts have been and continue to be most faithful to the precedent established in Brutan. 8" This is a subjective determination to be sure. However, the two prerequisites for fairness (a Jackson hearing and a Townsend consideration of habeas corpus cases) was 147 Taking a final look at Tables 5.1 and 5.2, I am again struck by the high level of compliance with Supreme Court precedent. Even when I code marginal decisions as being non-compliant, the Courts of Appeals only “disobey” the Supreme Court in less than 5% of all cases treating Brutan and 24% of the few cases treating Townsend. One should also note that noncompliance with Townsend is driven by two circuits. Perhaps the Fifth, because it is in the South, has somewhat divergent Views. Perhaps the Seventh is different because of the ideological makeup of its panels. Either way, the high level of overall compliance cannot be gainsaid. This reaffirms our earlier findings and provides yet another example of the Courts of Appeals remaining attentive to the Supreme Court when clearly it could get away with not being so compliant. There must be something about that High Court! used in this determination. 148 TABLE 5.1 CIRCUIT TREATMENT OF BR UTON V. UNITED STATES Number Shepard 's D Shepard ’s F Followed Not Followed First Circuit 5 4 1 4 1 Second Circuit 18 12 6 18 0 Third Circuit 0 0 0 0 0 Fourth Circuit 7 4 3 7 0 Fifth Circuit 37 26 11 35 2 Sixth Circuit 1890 8 9 18 0 Seventh Circuit 20 15 5 20 0 Eighth Circuit 1291 6 5 12 0 Ninth Circuit 15 10 5 14 1 Tenth Circuit 11 10 1 10 1 Eleventh Circuit 10 5 5 9 l D.C. Circuit 6 3 3 5 1 Totals 159 103 54 152 7 9° Shepard ’s codes one case as limiting Bruton. 9' Shepard ’s codes one case as limiting Bruton. 149 TABLE 5.2 CIRCUIT TREATMENT OF TOWNSEND V. SAIN Number Shepard ’s D Shepard ’s F Followed Not Followed First Circuit 1 0 1 0 Second Circuit 3 3 3 0 Third Circuit 4 4 4 0 Fourth Circuit 3 3 3 0 Fifth Circuit 10 10 6 4 Sixth Circuit 3 1 3 0 Seventh Circuit 4 1 0 4 Eight Circuit 8 7 8 0 Ninth Circuit 4 2 4 0 Tenth Circuit 1 1 1 0 Eleventh Circuit 1 l l 0 D.C. Circuit Not applicable Totals 42 33 34 8 150 CHAPTER 6: CONCLUSION: FAITHFUL AGENTS, IDEOLOGICAL ACTORS This dissertation has attempted to better understand decision making in the United States Courts of Appeals, while accounting for the influence of the Supreme Court in its role as principal. While scholarly attention is beginning to focus on these lower courts, most research in judicial politics remains at the Supreme Court level. However, as can be seen by mere reference to the number of unreviewed Court of Appeals decisions, the decisions of the lower federal appellate courts are increasingly becoming final — the law of the land. Because the Supreme Court hears so few of them, it behooves us to turn our energies to understanding decision making in these increasingly important courts. This dissertation has done that, and I think we gain some valuable insights fiom it. While it is focused on a certain subset of cases, and while it bears direct relevance to only one timespan, I think the conclusions to be drawn from this study are not unlike those that would be drawn had we a universe of cases. Insights found here can be applied to other issue areas and to other times, and in so doing the universality of my contentions can be expressly tested. However, short of doing that, I do think it valid to draw conclusions from what we see here, noting that it is possible that not all lower court behavior is the same as that under study, but also that it is probable that it is not all that different. In order to study decision making in the Court of Appeals I have employed those cases involving confession, coding each as to their facts, in both the Supreme Court and the Court of Appeals. Because attitudes drive decision making in the High Court, I model 151 only the decision making of the lower courts, adding potential influences of the former on the latter. The data were validly and reliably compiled and so the inferences drawn from them retain acceptability. In addition to the statistical analysis, I have formulated a formal model of lower court/ Supreme Court interaction, assigning payoffs and finding equilibria as well as calculating expected utilities. I also took some findings of interest from the quantitative analysis and looked more carefully at the reaction of the Courts of Appeals to Supreme Court decision making, reading lower court treatment of High Court policy prescriptions and deriving some other intriguing findings. Finally, I looked to the individual justices to determine whether they changed their behavior when they were elevated to the High Bench from a Court of Appeals panel. Through all of these approaches, I looked for evidence of the influence of the Supreme Court on the Court of Appeals, as well as reasons for that influence. Most importantly, I looked for explanations of the lower courts’ behavior. I find that, generally, the two institutions are different; that while the Court of Appeals judges are influenced by their attitudes, just like the justices of the Supreme Court, they are also influenced by the Supreme Court itself — its ideological membership and its policy prescriptions — because of the constraints placed upon them by their institutional makeup. In addition, the lower courts hear different types of cases because their docket is mandatory. The two major complexities in the institutional character of the Circuit Courts of Appeal -— its subordinate position and its mandatory docket — do seem to influence the way it makes decisions, as well as the cases it must decide. The remainder of this conclusion will discuss each chapters’ findings in order and will end with some directions for future research by suggesting questions we have not answered here. 152 6.1 WHY COMPLY? In Chapter 2, we found that, although it never maximizes their utility and it is not an equilibrium relationship, the lower courts still maintain loyalty to their principal, the Supreme Court. Using formal theory, I found that the lower court would be better off voting its preferences without regard to Supreme Court prescription on the matter, for that Court rarely supervises the behavior of the Courts of Appeals. However, for some reason, they still comply. This leads me to believe that when it comes to the principal- agency relationship between the lower courts and the Supreme Court, the High Court exerts an unmeasurable influence on the lower courts. That influence, I conclude, is one of moral authority. Scholars have discussed this moral authority before (Baum 1998:242- 3), (Canon and Johnson say the lower courts have an obligation to comply (1999:3 5)) concluding that it certainly plays a role in lower court compliance with Supreme Court policy prescriptions, but the difficulty lies in measuring that impact explicitly. Most certainly, a High Court lacking in authority would gain but little compliance, and so it is logical that the widespread compliance found here and elsewhere can be attributed to said authority. The proof of the existence of such a force is as follows: given that it is rarely in the interest of the lower court to comply with Supreme Court precedents, while nonetheless recognizing that the Supreme Court’s decisions do matter greatly to the lower courts, the inference that moral authority must be driving compliance can be validly drawn. The impetus to comply comes from what the Supreme Court is rather than what it does. We necessarily establish a new form of principal-agency then, which should prove to be a major contribution both to those interested in formal theories of institutional relationships generally, as well as those interested in the specific relationship under study: 153 that of the Courts of Appeals and the Supreme Court. This new principal-agency should aid considerably in understanding why the lower courts would bother with the Supreme Court’s prescriptions at all. No longer does loyalty seem to be irrational. Rather, it is deemed such a part of the culture and socialization of these judges that there is very little question when a choice exists between one’s own policy goals and Supreme Court precedent. However, this does not completely preclude a lower court judge voting his or her own attitudes. It merely means that if they do, they must find a way to make that decision at least look compliant. In Chapter 5, I show some of the ways that the lower courts do that, while still maintaining that compliance is the norm. 6.2 DECISION MAKING IN THE COURTS OF APPEALS The conclusion that the High Court matters because of what it is, not what it does, still leaves the question how the lower court decides cases. Does the lower court merely decide as the Supreme Court would have it decide, no questions asked? In Chapter 4 we found that, while Court policy does matter to the lower courts, they retain their autonomy and still vote at least partially in accordance with their own individual policy preferences. The lower courts pay heed to precedent; they take note of Supreme Court membership; and they vote their attitudes. It has long been argued that, while the Supreme Court decides almost solely according to the policy preferences of its members, perhaps with some strategic interaction as well, lower court judges, finding themselves in a different institutional setting, necessarily must take into account other factors and other actors. In so doing, the lower court judges’ calculus is much more complicated than that of the Supreme Court justices’. While there are undoubtedly judges, even on the lower courts, who vote their attitudes consistently, as a whole the lower court judges need to be less 154 personally preferential than are the High Court justices. They do have an obligation to decide cases in line with the Supreme Court’s jurisprudence, and when in doubt as to the High Court’s preference, it may even be argued that they should attempt to anticipate how the High Court would decide (Gruhl 1980). Either way, the lower court judge is not completely a free agent, and so the Court’s ideological composition as well as many of the Court’s precedents matter to these lower court judges. Also notable is the lack of real difference among the circuits in their decision making, at least insofar as the confession cases are concerned. Only the D.C. Circuit differs considerably from the others, seemingly deciding cases more as the Supreme Court decides cases. That circuit’s jurisdiction and caseload differ fi'om the other circuits, however, and so this seeming incongruity is probably the result of the circuit’s docket rather than a differential decision-making calculus. 6.3 BUT, IS IT COMPLETE COMPLIANCE? While we find much evidence that the lower court does in fact comply with Supreme Court precedent — and other studies have found similarly — it is necessary to note that the lower courts needn’t completely thwart Supreme Court precedent in order to be less than compliant with its decisions. In fact, there are a number of things a lower court can do, short of defiance, in order to limit the effect of an unliked Supreme Court precedent. The lower court can interpret the decision narrowly, limiting it to its specific facts (Johnson and Canon 1984); it may cite its own opinions rather than the offending Supreme Court precedent (Manwaring 1972); they may distinguish their case from the one for which a Supreme Court prescription is available (Caminker 1994; Sanger and Sheehan 1990; Baum 1978; Tarr 1977; Manwaring 1972); they may dispose of it on procedural 155 grounds (Johnson and Canon 1984); they may criticize the Supreme Court while following it (Tarr 1977); or they may simply ignore the offending precedent’s existence. All of these, save perhaps the ignoring, arguably still constitute compliance. However, the range of compliant behavior is striking. In Chapter 5, I attempted to determine whether the lower court, while seemingly compliant, was engaging in this sort of strategic noncompliance. Using one Supreme Court precedent that was determined to be significantly related to lower court decision making in Chapter 4 (Townsend v. Sain) and one that was not (Brutan v. United States) I examined each case in which one of the two precedents was treated by the lower court. I found substantial compliance with both decisions, and possibly more compliance with the one that was not significant. Overall, the Courts of Appeals complied in 95% of the cases treating Brutan and in 76% of those treating T awnsend. While there were occasional strained distinguishings and a couple of cases that seemingly violated the spirit of the Supreme Court pronouncement, the lower courts comply. There are indeed cases in which distinguishing the case from the Court’s precedent is valid; so this type of behavior is not automatically problematic. While the lower courts did distinguish their cases from either Brutan or Townsend quite often, only occasionally do I deem the distinction to be unduly artificial. 6.4 DIFFERENT COURTS = DIFFERENT JUSTICES? Finally, in Appendix A I discuss the possibility that the lower court judge, once elevated to the Supreme Court and given the freedom to make decisions according to his attitudes, will in fact decide cases differently. In order to investigate this possibility, I explore the decision making of the three judges-tumed-justice who served a significant time on both courts: Justices Burger, Blackmun, and Marshall. In that qualitative look at 156 their jurisprudence, I found that, while a few of the judges seemingly change little fi'om the time on the lower court to the time on the higher one, there are some indications for each of the three judges-turned-justice that institutions matter; that is, each is constrained in some way while on the lower court. And, once the ascension to the higher court takes place, those constraints dissipate. 6.5 UNANSWERED QUESTIONS, AND DIRECTIONS FOR FUTURE RESEARCH Overall, we find a highly compliant lower court that, while deciding cases attentive to the policy prescriptions and attitudinal predispositions of the Supreme Court, is still able to incorporate its policy preferences into its decision making. Judges on the lower courts, while more constrained than those on the High Court, may nonetheless decide cases in substantially the same manner on each. However, they are much freer to do so on the High Court as they obviously needn’t fear being overruled. In addition, on the Supreme Court, they obviously have at least some say in the types of cases they and the Court will hear, while those serving on the Courts of Appeals must hear them all. So, while there are definitely differences between the two courts and their decision-making processes, attitudinal theory remains relevant. Though the judges are faithfirl agents, they are also ideological actors. However, questions remain with respect to lower court decision making. First, exactly how much do differences in decisions account for differences in outcome? That is, does the finding that the lower court overturns proportionately far fewer confessions than the Supreme Court mean anything substantively, or does that finding merely suggest that the lower court is hearing claims that are less meritorious than those heard by the Supreme Court? Since every defendant has the opportunity to appeal his or her case to the Court 157 of Appeals and that court must hear that claim, the possibility of fiivolous suits making it to that court are far higher than the possibility that they are heard by the Supreme Court. It would be a relevant and usefill endeavor to examine exactly what the differential docket of the lower court means to its decision making. Do workload considerations constrain the lower court to decide in accordance with the Supreme Court simply because it is easier (i.e., less time-consuming) to do so? Are the cases so often without merit that the appeal to the Court of Appeals actually means very little, as the district court determination is so often upheld there? Once we develop methods for empirically analyzing the substantive merit of the cases appealed to the Courts of Appeals and determine exactly what impact that has on decision making, we will be able to have more confidence in our finding that the lower court does indeed decide cases differently fi'om the Supreme Court. The second unanswered question refers to the actual role of ideology in the lower courts. We find that when we add a variable for ideology into a model of lower court decision making, that variable is significant; nonetheless, exactly what role ideology plays in lower court decision making remains unknown. Is it deterministic as it is in the Supreme Court, or is it merely contributory, as our results would seemingly suggest (since other things are significant as well)? Does it explain separate opinions on the lower courts as it does on the Supreme Court; e.g., do dissenters dissent for ideological reasons, or are there other constraints?92 Do the ideological predispositions of the judges of the Court of 92 One such constraint immediately comes to mind: I wonder, do district court judges, sitting by designation on a certain panel of the circuit court, behave differently than those on the panel full-time? Are there less dissents registered by said justices; less majority opinions written? Does this judge face the additional constraint of being an 158 Appeals change overtime, or are they relatively stable? Finally, are there strategic considerations here, as exist at the Supreme Court level? What is it about the Court of Appeals as an institution that fosters or hinders such strategic interaction? The third inadequately answered question is that contained in the appendix; institutional effects on decision making. An individual-level analysis such as this is highly desirable and emminently useful, provided it be conducted at a more rigorous level. Here, because of the narrow subject I consider, I am only able to glean some vaguely interesting patterns of differential or similar behavior. A larger scale analysis would lend much to the knowledge we have about the effects of institutional constraints on decision making, and this type of study is something that should occupy a high place on firture research agendas. Finally, do my results extend to other subject areas at other points in time? Has the lower court become more ideological overtime? Are differences in decision making dependent upon the subject matter under consideration? Do different parties draw a different type of response from these judges? While some of these matters are being addressed by scholars in the field, it remains important to recognize that this is but one study at one point in time using one set of cases. In order to fully understand the range and depth of influences on lower court decision making, we must continue with this sort of research until we have the knowledge of these lower federal courts (and state supreme courts and trial courts and courts of other nations) that we currently have of the United States Supreme Court. Arguably, these courts are the most important. In the vast majority of cases, they authoritatively prescribe the law of the land. If we are to outsider, or of a lower position? These are interesting questions that the literature has not yet answered. 159 understand the impact of the judiciary on our lives, and on democracy in general, we necessarily must turn away from our one unique (yet important) United States Supreme Court, and begin to come to a theory and an understanding of courts in general. In so doing, our knowledge will become as broad and deep and influential as it possibly can be, thereby maximizing our contribution to society. 160 APPENDIX A 161 APPENDIX A: FROM JUDGE To JUSTICE — INSTITUTIONAL EFFECTS ON JUDICIAL DECISION MAKING PROCESSES A final method by which to measure the institutional differences between the Supreme Court and the Court of Appeals is to examine the very same actors in the two different institutional settings. In this appendix, I explore the possibility that the same judge finding himself in a different institutional context, will decide cases differently. It should be the case that the lower court judge, finding more constraints on his behavior, will decide cases with due deference to precedent and to the High Court. However, once let fiee fiom the bonds of the judicial hierarchy, this same judge, now a justice, should decide more in accordance with his own policy preferences, never mind any additional influences. In other words, voting on the Supreme Court should be more “sincere.” In this appendix, 1 will explore the possibility that the same judges, in different institutional settings, decide cases differently. In order to do so, I examine the voting and opinion writing behavior of three judges later incamated as justices: the conservative Warren Earl Burger, the liberal Thurgood Marshall, and the hybrid Harry Andrew Blackmun. Although the cases in which these three participated are somewhat scarce, I think an exploration, from which we can draw some tentative conclusions, is warranted and defensible.93 93 Because of the lack of real numbers here I could not justify an additional chapter. Future research will filrther explore the potential of this research question, but for now, we are left only with tentative conclusions. 162 A.l THE JUDGE TURNED JUSTICE Warren Earl Burger served on the Court of Appeals for the District of Columbia circuit from 1956 to 1969. During that time, as can be seen by Table A. 1, he participated in the decision of 22 confession cases, 7 of which were heard by the circuit en banc. “As a circuit judge he attracted national attention from bench and bar for his judicial administration and for his jousts with the civil libertarians on that talented but badly divided court” (Morris 19922104). He led the oft-dissenters on that panel opposing the extension of rights to the criminally accused, giving preference instead to police, prosecution, and trial judges. As Chief Justice of the Supreme Court, he was the least enthusiastic on that bench to extend the civil rights decisions of the liberal and activist Warren Court. It does seem, from the data presented in Table A. 1, that this particular justice was never constrained from voting his attitudes. He certainly had no qualms about dissenting while seated on the D.C. circuit, and in fact, his dissents speak quite clearly of an ideologue in this set of cases. Harry Andrew Blackmun served on the Eighth Circuit of the Court of Appeals from 1959 to 1971 when he was appointed to the High Court. While his tenure on the Court of Appeals is not often remarked upon, his voting on the Supreme Court was quite interesting. As a freshman justice, Blackmun nearly always aligned himself with the Chief Justice. So much so that the pair, Burger and Blackmun, came to be known as the “Minnesota Twins.” However, as he became more comfortable on the Bench, he moved away from Burger, “becoming outspoken and explicit in his efforts to keep an increasingly conservative Court on center” (Wasby 1992276). Even on criminal procedure, he moved away from his strongly held conservative views to become more supportive of the rights 163 of the accused. Because I have so few of his participations in my dataset, I cannot make many broad comparisons between his tenure on the Court of Appeals and his tenure on the High Court. Nonetheless, his voting appears again to be quite consistent. Thurgood Marshall served on the Second Circuit Court of Appeals fiom 1961 to 1965, writing several important decisions while there. In 1965 he was named the first Afiican-American solicitor general by President Lyndon Johnson, and in 1967, he became the first Afiican-American to occupy a seat on the Supreme Court. Although we have few of his lower court opinions to examine, his voting on the Supreme Court was overwhelmingly liberal in this issue area. What is striking when comparing his tenure on the lower court with that on the Supreme Court is the huge increase in his dissent rate. In my set of confession cases, Marshall never dissented while on the lower court, even though he was often accompanied by two Republican panelists, while on the Supreme Court he dissented in fully 30% of all confession cases in which he participated. Marshall’s jurisprudence, in this small set of cases, seems to indicate some level of constraint on the lower court and warrants closer examination. A.2 EXPLORING DECISION MAKING AND OPINION WRITING FROM JUDGE TO JUSTICE Because of the lack of opinions in the dataset when discussing judges of the Court of Appeals who are later appointed to the Supreme Court, looking directly and qualitatively at the opinions written by the actors in their differential roles could be quite telling. In this section, then, I will examine the language used by the judge-turned-justice to see if there are any words of reticence and then later of grandeur — that is, on the lower court, language deferential to the High Court juxtaposed with, on the Supreme Court, broadly sweeping proclamations without regard to influences other than personal policy 164 preferences. While I have the most information for Judge/Justice Burger, I will examine all three judges, attempting to draw some conclusions about how different ideological affiliations seemingly affect role perceptions of the actor in the two institutional settings. First, perhaps it would be useful to discuss the differences between the two institutions. What is it about the Court of Appeals that makes us expect its judges to behave differently from the justices of the Supreme Court? As we have found in the earlier chapters of the dissertation, the lower court is seemingly aware both of the ideological makeup of the High Court, and the precedents it has espoused. In the model tested in Chapter 4, I find that the lower court jurist incorporates his or her own ideology into his or her decision making, while also paying heed both to Supreme Court ideology (in order to avoid being reversed, perhaps) and also Supreme Court precedent, duly complying with the policy prescriptions of the High Court in this area of the law. Because other scholars have determined that the sole motivating factor for decision making on the Supreme Court is policy preferences of the justices, including the desire to have their preferences translated into law (Segal and Spaeth 1993), I contend that the lower court jurist is necessarily “other-involved” in her decision making. That is, the Courts of Appeals judge cannot merely vote her preferences. I look at the judges in my dataset for indications of this type of behavior, examining both the judge-tumed-justices’ opinions and the opinions that they join during their tenure on the Court of Appeals and later on the Supreme Court. By no stretch of the imagination do I have impressive numbers of opinions for any of the judges in any of their capacities; still, some tentative conclusions can be drawn, or at least some notes for further research. 165 A.3 JUDGE/JUSTICE BURGER Warren Burger participated in 22 confession cases in his tenure on the D.C. circuit. Of those 22 decisions, he only voted to exclude confessions in four cases. And notably, these four votes to exclude confessions can be readily explained while keeping to the notion that Burger was an ideological actor fiom the beginning. In Naples v. US, 307 F .2d. 618, Burger concurred in the opinion of the court except for the court’s mandate that the new trial be conducted by a different judge. This decision was heard en banc, before 6 Democratic-appointees and 3 Republican appointees. The decision, as regards the confession, was unanimous, although the Republican-appointed Bastian and his Democratic counterpart Miller (who behaves as a Republican) did dissent on another aspect of the case. The fact that Burger concurred with a remand to the lower court to determine the voluntariness of the confession, then, does not seem so out of character.“ In Lackley v. US, 270 F.2d. 915, Burger dissented from a decision admitting a confession. However, in his dissent he did not really advocate overturning the confession. Rather, he quarrels with the court’s review of the case, holding that there was no adequate basis for appellate review. Again, then, an easily understandable deviation. In T rilling v. US, 260 F.2d. 677, we see a somewhat constrained Warren Burger. Here, he concurs but says: 9‘ When an ideologue joins a unanimous court in the resolution of some issue, it seems safe to assume that the case was “easy” or straightforward. Even the attitudinal model does not firlly explain unanimous decisions. [While this common-sense explanation is probably persuasive, there is seemingly a research question to explore: why are there unanimous decisions?] I think that viewing them as noncontroversial cases explains the seemingly incongruity with a model that emphasizes preferences over all other considerations. 166 I think the result reached by Judge DAN AHER is compelled by the Mallory case and therefore concur with him, but only because I conclude we are not free to do otherwise. I do this reluctantly because what Judge Prettyman [in dissent] has said makes sense and ought to be the law. The steady expansion of the meaning of ‘unnecessary delay’ by the courts since Rule 5(a) has been in effect suggests to me that this area of the law cannot be developed properly on a case-by-case basis, even though that is sometimes appropriate; Rule 5(a) should be re-examined by the rule making process or by Congress. Judge Prettyman’s dissent, joined by fellow-Democrat Miller” and the Republican- appointed Bastian, goes on at great length about what is and is not unnecessary delay, highlighting the fact that delay is not per se problematic. I would not categorize the opinion as being defiant of Supreme Court precedent. However, the majority here exhibits much more automatic compliance. Either way, this is a striking example of the institutional constraints placed on the Court of Appeals judge. Clearly Burger would not decide this case in the same manner were he on the High Court, free to vote as he saw fit. In every one of his other lower court opinions, Burger voted to allow the confession into evidence. In many of these cases, he is in the minority, as was usually the case for him on a Democratic-appointee-dominated Circuit court. In Rettig v. United States, 239 F .2d. 916, he argues that a suspect may not both claim she did not make the confession in question and claim that it was involuntary due to delay in arraignment. He cites Supreme Court cases and interprets them in doing so. In Killaugh v. United States, 315 F.2d. 241, Burger makes some bold statements in dissent. The majority in this case, again dominated by Democratic appointees, determines that a confession made a day after arraignment was the fruit of the pre-arraignment confession and thereby inadmissible. 9’ While Wilbur K. Miller was indeed appointed by a Democrat, his voting behavior is much more consistent with that of a conservative. 167 Burger is quite agitated with this ruling, saying that “The majority holding constructs an entirely new ‘statute’ and takes a step neither contemplated by Congress nor remotely warranted by the Mallory case” (at 257). He goes on, I find it difficult to characterize what the court does in this case. To me it is an abuse of judicial power to write what is, in effect, an amendment to Rule 5(a) because some think that Congress did not go far enough. More than that the arrogated power is exercised in a way which offends common sense. Some of the members of the court might remember that there are other branches of government at least equally qualified to flame the laws, explicitly ordained to do just that, and no less concerned than we are with individual liberty. Our task as judges, properly exercised, is a narrow one: to interpret the laws faithfully as Congress wrote them, not as we think Congress ought to have provided (at 260). He becomes quite animated at the end of his opinion: “If it is to be the law that the courts may not use a voluntary confession, made nearly 24 hours after the judicial warning in a preliminary hearing, after time for ‘deliberate reflection,’ after firll opportunity to secure counsel and after rejecting offers of counsel then indeed this court will have converted the shield of Rule 5(a) into a sword. What Congress clearly intended as a protection will have become a weapon of special advantage exclusively for the guilty” (at 260). As we shall see, he echoes this eloquent and passionate exposition of the duties of a judge on the High Court as well. In Spriggs v. United States, 335 F.2d. 283, he dissents fi'om a holding of unnecessary delay in arraignment because he views the delay as reasonable. In Pea v. United States, 397 F.2d. 627, he dissents from the majority’s creation of the reasonable doubt standard in a trial judge’s review of the voluntariness of a confession outside the hearing of the jury. Finally, in Frazier v. United States, 419 F.2d. 1161, we again find a less than deferential lower court judge, as he dissents from the majority’s determination 168 that a request by an accused not to take notes indicated a less-than-knowledgeable waiver of his rights. Burger says, first, that the Mallory decision “has been drawn into serious doubt by recent Congressional enactments” (at 1171). Later, discussing Miranda, he says that “The articulation of a stringent waiver requirement was merely a device through which the Court sought to ensure that Fifth Amendment guarantees were not unduly impaired at pre-trial interrogations. The guidelines set forth in Miranda were means servicing constitutionally prescribed ends; as artifices of implementation they are subordinate and only incidental to the rights they were designed to secure” (at 1172). I think this statement remarkable indeed, as a lower court judge seemingly determines for himself which portions of relevant Supreme Court precedent he will regard highly and which he will not. He goes on to chastise the rest of his panel (two Democrats) for “engag[ing] in sheer speculation of Appellant’s thought processes which places a premium on the capacity of judges to probe Appellant’s mind” (at 1174). I find the closing of this opinion amusing: Each time judges add nuances to these ‘rules’ we make it less likely that any police officer will be able to follow the guidelines we lay down. We are approaching the predicament of the centipede on the flypaper — each time one leg is placed to give support for relief of a leg already ‘stuck,’ another becomes captive and soon all are securely immobilized. Like the hapless centipede on the flypaper, our efforts to extricate ourselves from this self-imposed dilemma will, if we keep it up, soon have all of us immobilized (at 1176). It seems that, in this case, Burger is driven purely by attitudes, and openly criticizes Supreme Court policy. As his is not the majority view, we would probably still deem this circuit as being compliant with Miranda and/or Mallory were we conducting a compliance study. However, when we switch the unit of analysis to the individual level, it may well be 169 that we find instances of judicial disapproval of Supreme Court policy and the interpretation thereof. Comparing the above-described behavior to Burger’s behavior on the Supreme Court, we find little to denote a change in view due to the change in status. Granted, he did not participate in many cases in my dataset while on the Supreme Court. However, he did have a fiiendlier group of colleagues there and, therefore, was forced to dissent far less often. On the High Court, he again only voted to exclude a confession in two of ten instances. And, like those instances in which he similarly behaved on the D.C. Circuit Court of Appeal, they were cases which approximated unanimity. In Brown v. Illinois, 422 US. 590, every justice agreed that the confession be excluded from evidence, including Burger. In Mincey v. Arizona, 437 US. 385, all but Rehnquist concluded similarly. It seems that Burger is a pure ideologue. He was an ideologue on the Courts of Appeals, and he remained so on the Supreme Court. Perhaps his conservative ideology accounts for this behavior. Perhaps the fact that only two others in his entire circuit consistently agreed with him (in this set of cases) might explain his exasperation with the decisions to which he was a party, and he couldn’t help but dissent and dissent loudly. His circumstances may well be unique. It might be that another judge, in a more amicable environment, would play the role of faithful agent more convincingly. A.4 JUDGE/JUSTICE BLACKMUN While Blackmun, like Burger, displays little behavior shift from one court to the other, as shown on Table A. 1, there is something different at work in his case. Blackmun heard only four confession cases while on the Eighth Circuit Court of Appeal. In those four cases, he was on panels with a variety of ideological compositions. He sat with a 170 Democrat and a Republican in Young v. United States, 344 F.2d. 1006, and the unanimous decision, upholding use of the confession, was written by the Democrat. In another, he again sat with one Democrat and one Republican, but this time he wrote the opinion for the unanimous panel, drafting a lengthy defense of the lower court’s finding of voluntariness (F eguer v. United States, 302 F.2d. 214). He authored another unanimous opinion for a panel consisting of himself and two Democrats, allowing another confession into evidence (Cox v. United States, 373 F.2d. 500). In only one decision did Blackmun vote to exclude a confession, and he wrote the opinion for a unanimous court composed of one fellow Republican and one Democrat (Mitchell v. Stephens, 353 F.2d. 129). He seems downright deferential to the High Court in his opinion, attempting to predict what they would do if faced with the situation therein. He opines, “But, taking into consideration all the facts, the absence of a personal expression of waiver by Mitchell, and the tremendous stake so far as he is concerned, we are not at all convinced that the Supreme Court, on this record, would uphold the district court in its determination that Mitchell had waived the coercion issue” (at 141). Perhaps none of these judges voted ideologically, or more likely, they did not need to in this set of cases because they were “cut and dry,” so to speak. When Blackmun was elevated to the High Bench, as mentioned before, he aligned himself most often with Burger until he became a bit more independent and began voting more liberally. In my dataset, l have only 9 decisions in which Blackmun participated once on the High Bench, and only one in which he actually wrote an opinion (Brown v. Illinois, supra). This provides very little information as to any changes from one bench to the other, just as the total lack of dissent from him decreases the utility we gain in 171 examining his voting behavior. It is interesting to note that he does indeed appear to align himself first with Burger, and then, as time goes on, with the more liberal wing of the Court; he is most definitely voting ideologically on the High Bench, although his ideology seemingly changed overtime. But because we’ve so little information as to his work on the lower court, we cannot draw many conclusions. His lack of dissent, even when faced with some liberal outcomes seems to point to a seeming lack of ideological behavior. But, the cases he decided on the lower court may have just been “easier” and more clear-cut, hence avoiding the necessity for separate opinion writing. He was clearly a policy-seeker on the High Court, but his role perception while on the lower court is somewhat elusive. A.5 JUDGE/JUSTICE MARSHALL The final judge-tumed-justice under study is Thurgood Marshall. Again, we face data problems, having only four participations while on the Second Circuit and 15 once elevated to the Supreme Court. Compounding matters is his failure to write even in one of the four lower court decisions, while writing in only five of his 15 High Court participations. However, there remain conclusions/speculations to be drawn from his behavior on the High Bench versus his behavior on the Court of Appeals. First, while he joins each majority opinion on the lower court, whether serving with all Republicans or a Republican and a Democrat, he dissents quite often on the Supreme Court. In fact, only a few justices wrote more dissents in their tenure than he (although some wrote more in proportion to the number of their participations) (Epstein et al. 1996). In addition, he wrote Opinions in one-third of his Supreme Court confession cases, although he wrote in none of the decisions while on the lower court. This may be due to the fact that some circuits of the Court of Appeals seem to assign opinions randomly (Wasby 1999) while the 172 Supreme Court most assuredly does not (Benesh, Lowry, Sheehan, and Spaeth 1999). But nonetheless the lack of separate opinions in somewhat telling. It should be noted that separate Opinions are far less frequent on the Court of Appeals than they are on the Supreme Court, however, so the lack thereof may be more a firnction of behavioral norms than a lack of ideology. The possibility nonetheless exists that Marshall simply did not deem it proper to assert his ideological bent while on the lower court, while he felt it acceptable, maybe even necessary, to do so while on the Supreme Court. In fact, in one case, he even advocated the overruling of a two-year-old precedent in concurrence while on the High Court (Mincey v. Arizona, supra) thereby showing virtually no constraint whatsoever. A.6 CONCLUSIONS AND SPECULATIONS So, what, if anything, have we learned from this? First, we have learned that, in order to conduct the sort of analysis proposed here, one must either look over longer time periods or across many different issues. I think the former more plausible than the latter, for in comparing decisions among courts, one must devise some way to render those decisions comparable. In this dissertation, I use fact patterns. There are probably other ways to do this as well, but most are not conducive to more than one issue area at one time. So, in the firture, I plan to backdate and update my dataset in order to look more systematically at voting changes in a single justice as he makes the transition fiom the Court of Appeals to the Supreme Court. Some current research is beginning to focus upon this particular test of institutional constraints on decision making, and I think it is a very fi‘uitfirl avenue to pursue (Ditslear and Scott 1999). 173 Second, we do learn some things of substance. First, not all lower court judges are constrained by the Supreme Court or by precedent. Some of the lower court judges are indeed policy driven even within the institutional constraints imposed. The question of what drives such behavior is left unanswered here, but perhaps it is ideology or perhaps it is the ideological composition of the rest of the circuit. Maybe ideological behavior is oftentimes masked because there are just too many judges who think alike on the lower bench, or who perhaps share a role perception of deciding cases solely according to Supreme Court pronouncement. Both behaviors would look the same even though motivations there are extremely different. But, if there is some consensus on either, we would find in our statistical analyses that ideology does not seem to matter quite as much as it does on the Supreme Court. Another substantive conclusion we can draw from this analysis is that opinion writing behavior seemingly differs between the two courts. Now, this is noticeable in much of the analysis in this dissertation, but at the individual level it becomes more prominent. Judges who write often on the Supreme Court did not behave in the same manner when they were on the Court of Appeals. There is any number of possible explanations for this phenomenon, and this too would be a fiuitfill course for firture research. Is there a norm of consensus in the lower courts, or, as I have suggested several times in this dissertation, are the cases they hear merely “easier”? Are Courts of Appeals judges too overworked to spend precious time writing concurrences and dissents when their energies are more fruitfully spent elsewhere? Do district court judges dissent less often than Circuit court judges when they sit on Courts of Appeals panels, and is this due to their inferior status among their colleagues on the panel? Opinion writing questions 174 abound in these lower courts for, while we have studied such questions on the Supreme Court, we have just scratched the surface in terms of the norms and reasons for behaviors on the Courts of Appeals. A fourth and final substantive conclusion drawn from this appendix is that judges differ. This may seem obvious, but it is worth discussing nonetheless. They differ from each other even when serving on collegial courts together. They differ in their behaviors depending upon the court on which they are currently sitting (or at least we think they do). They differ from other political actors in that their constraints are different. They differ in their perceptions of the jobs they do and the ways in which that job should be done. They differ ideologically, some finding truth in the conservative View of a given issue and others finding solace in the liberal view. They differ from themselves at a given point of time, as they mature or perhaps become more comfortable with their role. Regardless, they differ, and if we continue to do research at the aggregate level, we miss this fascinating aspect of judicial behavior. We miss the actual behavior of the judges. Therefore, let me join Gibson (1998) in advocating more research at the individual level so that we may better understand judges and better explain why they differ. For only in such an explanation and such an understanding can we get to the universal theories of judicial behavior that we all hope exist. 175 TABLE A. 1: JUDGE TO JUSTICE VOTING BEHAVIOR Judge Burger Justice Burger Number of Confession Cases: 22 10 Votes to Exclude Confessions: 4 2 Percent Liberal Voting: 18% 20% Number of Dissents: 8 1 Percent Dissenting: 36% 10% Judge Blackmun Justice Blackmun Number of Confession Cases: 4 9 Votes to Exclude Confessions: l 3 Percent Liberal Voting: 25% 33% Number of Dissents: 0 0 Percent Dissenting: 0% 0% Judge Marshall Justice Marshall Number of Confession Cases: 4 15 Votes to Exclude Confessions: 2 12 Percent Liberal Voting: 50% 80% Number of Dissents: 0 5 Percent Dissenting: 0% 33% 176 TABLE OF CASES 177 TABLE OF CASES COURTS OF APPEALS CASES Allison v. Holman 326 F.2d. 294 (1963) Bowden v. Francis 733 F.2d. 740 (1984) Burns v. Beta 371 F.2d 598 (1967) Cargill v. Turpin 120 F.3d. 1366 (1997) Childs v. Cardwell 452 F.2d 541 (1971) Church v. Sullivan 942 F.2d 1501 (1991) Close v. US. 450 F.2d. 152 (1971) Cockrell v. Oberhauser 413 F.2d. 25 6 (1969) Cox v. US. 3731726! 500 (1967) Dempsey v. Wainwright 4 71 F.2d. 604 (1973) Duggar v. US. 434F261. 345 (1970) Edwards v. Garrison 529 F.2d. 1374 (1975) Evans v. Buchanan 256 F.2d. 688 (1958) Farmer v. Kropp 445 F.2d. 5 (1971) Federated Mutual Implement and Hardware Insurance C a. v. Fairfax Equipment Company, Inc. 261 F.2d 207 (1958) Feguer v. US. 302F2d. 214 (1962) Falstan v. Allsbraak 691 F.2d. 184 (1982) Frazier v. US. 419F261. 1161 (1969) Gatcher v. Beta 444 F.2d. 696 (19 71) Grieco v. Meachum 533 F.2d. 713 (1976) Grindstaff v. Bennett 389 F.2d. 55 (1968) Hart v. Eyman 458 F.2d. 334 (1972) Hoover v. Beta 467F2d. 516 (1972) Jamison v. Lackhart 975 F.2d. 1377 (1992) Killaugh v. US. 315 F.2d. 241 (1962) LaFrance v. Bahlinger 499 F.2d. 29 (1974) Lackley v. US. 270 F.2d. 915 (1959) McHenry v. US. 420 F.2d. 927 (1970) McMilIin v. Beta 44 7 F.2d. 453 (1971) Migliare v. US. 409F261. 786 (1969) Miller v. Cox 45 7 F.2d. 700 (1972) Mitchell v. Stephens 353 F.2d. 129 (1965) Naples v. US. 307F261. 618 (1962) Pea v. US. 397F261. 627 (1967) Randolph v. Parker 5 75 F.2d. 11 78 (1978) Rettig v. US. 239F2d. 916 (1956) Rios-Ramirez v. US. 403F261. 1016 (1968) 178 Roberts v. US. 416F261. 1216 (1969) Rogers v. McMackin 884 F.2d 25 2 (1989) Santara v. US. 402 F.2d 920 (1968) Spriggs v. US. 335 F.2d. 283 (1964) Stinsan v. Wainwright 710 F.2d 743 (1983) Tasby v. US. 451F261. 394 (1971) Thomas v. US. 409F261. 730 (1969) Trilling v. US. 260 F.2d 677 (1958) US. v. Akers et al. 542 F.2d. 770 (1976) US. v. Arambula-Ruiz 987 F.2d. 599 (1993) US. v. Arias 98417.2d 1139 (1993) US. v. Barragan 453 F.2d. 386 (1971) US. v. Burroughs 650 F.2d 595 (1981) US. v. Burroughs 935 F.2d 292 (1991) US. v. Castro 413 F.2d 891 (1969) US. v. Castro 596 F.2d. 674 (1979) US. v. Coppola 788 F.2d. 303 (1986) US. v. Cox 449F.Zd. 679 (1971) US. v. Crockett 813F261. 1310 (1987) US. v. Danzey 594 F.2d. 905 (1979) US. v. Donner et al. 497F.Zd. 184 (1974) US. v. Espericueta-Reyes 631 F.2d. 616 (1980) US. v. Follette 430 F.2d. 1055 (1970) US. v. Fountain 449F261. 629 (1971) US. v. Gibson 416F261. 326 (1969) US. v. Gonzales 749F261. 1329 (1984) US. v. Grant 549 F.2d 942 (1977) US. v. Grilla 527F2d. 1344 (1979) US. v. Gurule 437F261. 239 (1970) US. v. Hoffa 402 F.2d. 380 (1968) US v. Jackson et al. 482 F.2d. 1167 (1973) US. v. Jimenez 780 F.2d. 975 (1986) US. v. Johnson 478F.Zd. 1129 (1973) US. v. Kelley et al. 526F.Zd. 615 (1975) US. v. Kendricks 623 F.2d. 1165 (1980) US. v. Kreiser 15 F.3d. 635 (1994) US. v. LaVallee 415 F.2d. 150 (1969) US. v. Littman 421 F.2d. 981 (1970) US. v. Maddox 492 F.2d. 104 (1974) US. v. Mancusi 441 F.2d 1073 (1971) US. v. Maurice 416F261 234 (1969) US. v. McManus 560 F.2d. 747 (1977) US. v. Montgomery 582 F.2d. 514 (1978) US. v. Ogilvie 337F261. 427 (1964) US. v. Pate 405 F.2d. 449 (1968) 179 US. v. Pate 418F261 1028 (1970) US. v. Pollard 509F2d 601 (1975) US. v. Rogers et al. 652 F.2d 972 (1981) US. v. Rundle 389F261. 47 (1968) US. v. Sims 430 F.2d 1089 (1970) US. v. Smith 46 F.3d 1223 (1995) US. v. Smith 792F.Zd 441 (1986) US. v. Spinks 470 F.2d 64 (1972) US. v. Steel 458F261 1164 (1972) US. v. Trautman 458F261. 217 (1972) US. v. Trudo 44917.2d 649 (1971) US. v. Truslaw 530 F.2d 257 (1975) US. v. Valenzuela 521 F.2d 414 (1975) US. v. Venere 416F261 144 (1969) US. v. Vissars 596 F.2d. 400 (1979) US. v. Wallack 414F261. 246 (1969) US. v. Washington 952 F.2d. 1402 (1991) US. v. Wilson 500 F.2d. 715 (1974) US. v. Yeager 453F2d. 1001 (1971) VanErmen v. Burke 398 F.2d. 329 (1968) Weidner v. Thieret 932 F.2d. 626 (1991) Westbrook v. Zant 704 F.2d. 1487 (1983) White v. Lewis et al. 874 F.2d. 599 (1989) White v. US. 41517.2d. 292 (1969) Woodsan v. Brewer 437F2d 1036 (1971) Yates v. US. 418F.Zd 1228 (1969) Young v. US. 344E261. 1006 (1965) SUPREME COURT CASES Aschcraft v. Tennessee 322 US. 143 (I944) Boulden v. Holman 394 US. 4 78 (1969) Brown v. Board afEducatian 34 7 US. 483 (1954) Brown v. Illinois 422 US. 590 (1975) Brown v. Mississippi 297 US. 278 (1936) Brutan v. US. 391 US. 123 (1968) Bureau v. McDowell 156 US. 465 (1921) California v. Green 399 US. 149 (1970) Chambers v. Florida 309 US. 241 (1940) Colorado v. Cannelly 479 US. 15 7 (1986) Craaker v. California 357 US. 433 (1958) Cruz v. New York 481 US. 186 (1987) Culambe v. Connecticut 367 US. 568 (1961) Davis v. North Carolina 384 US. 737 (1966) Delli Paali v. US. 352 US. 232 (195 7) 180 Duttan v. Evans 400 US. 74 (1970) Escobeda v. Illinois 378 US. 4 78 (1964) Fay v. Naia 372 US. 391 (1963) Frazier v. Cupp 394 US. 731 (1969) Gallegos v. Colorado 370 US. 49 (1962) Harrington v. California 395 US. 250 (1969) Harris v. New York 401 US. 222 (1971) Haynes v. Washington 3 73 US. 503 (1963) Idaho v. Wright 497 US. 805 (1990) Jackson v. Denna 3 78 US. 368 (1964) Keeney v. Tamaya-Reyes 504 US. 1 (1992) Krulewitch v. US. 336 US. 440 (1949) Lee v. Mississippi 332 US. 742 (1948) Lisenba v. California 314 US. 219 (1941) Lynumn v. Illinois 3 72 US. 528 (1963) Mallory v. US. 354 US. 449 (1957) Mapp v. Ohio 367 US. 643 (1961) McNabb v. US. 318 US. 332 (1943) McNally v. Hill 293 US. 131 (1934) Milton v. Wainwright 407 US. 371 (1972) Mincey v. Arizona 43 7 US. 385 (1978) Miranda v. Arizona 384 US. 436 (1966) Nelson v. 0 ’Neil 402 US. 622 (1971) Parker v. Randolph 442 US. 62 (1979) Payne v. Arkansas 356 US. 560 (1958) Peyton v. Rowe 391 US. 54 (1968) Richardson v. Marsh 48] US. 200 (1987) Roberts v. Russell 392 US. 293 (1968) Rogers v. Richmond 365 US. 534 (1961) Silverthorne Lumber Co. v. US. 251 US. 385 (1920) Spano v. New York 360 US. 315 (1959) Stein v. New York 346 US. 156 (1953) Tennessee v. Street 471 US. 409 (1985) T awnsend v. Sain 3 72 US. 293 (1963) T urner v. Pennsylvania 338 US. 62 (1949) Wang Sun v. US. 371 US. 471 (1963) OTHER MISCELLANEOUS People v. Defare 150 NE. 585 (1927) State v. Nusbaum 198 N. W.2d. 650 (1972) 181 REFERENCES 182 REFERENCES Achen, Christopher H. 1982. Interpreting and Using Regression. Newbury Park: Sage Publications. Aldrich, John H. and Forrest D. Nelson. 1984. Linear Probability, Lagit, and Prabit Models. Newbury Park: Sage. Amar, Akhil Reed. 1994. “On Lawson On Precedent.” Harvard Journal of Law and Public Policy 17:29-44. Arrow, Kenneth J. 1985. “The Economics of Agency.” In John W. Pratt and Richard J. Zeckhauser, eds. 1985. Principals and Agents: The Structure Of Business. Boston: Harvard Business School Press. Arrow, Kenneth J. 1986. “Agency and the Market.” In Kenneth J. Arrow and Michael D. Intrili gator. Eds. Handbook of Mathematical Economics. New York: North- Holland. Arrow, Kenneth J. and Michael D. Intriligator. Eds. 1986. Handbook of Mathematical Economics. New York: North-Holland. Baum, Lawrence. 1978. “Lower-Court Response to Supreme Court Decisions: Reconsidering a Negative Picture.” The Justice System Journal 3 :208-219. Baum, Lawrence. 1998. The Supreme Court. 66‘ Ed. Washington, DC: CQ Press. Beatty, Jerry K. 1972. “State Court Evasion of United States Supreme Court Mandates During the Last Decade of the Warren Court.” Valparaiso Law Review 6:260-85. Belcaster, John F. 1992. “The D.C. Circuit’s Use of the Chevron Test: Constructing a Positive Theory of Judicial Obedience and Disobedience.” Administrative Law Review 44:745-765. Benesh, Sara C. 1997. “Confession Cases in American Courts: Perspectives on the Hierarchy of Justice.” Presented at the Annual Meeting of the American Political Science Association, Washington, DC. 183 Benesh, Sara C., Robert Lowry, Reginald S. Sheehan and Harold J. Spaeth. 1999. “Strategic Equity: Opinion Assignment on the United States Supreme Court.” The Program for Law and Judicial Politics Working Paper 98-03. East Lansing: Michigan State University. Benesh, Sara C. and Malia Reddick. 1998. “Overruled: An Event History Analysis of Lower Court Reaction to Supreme Court Alteration of Precedent.” Presented at the Annual Meeting of the Midwest Political Science Association. Chicago, IL. Bloch, Susan Low. 1992. “Marshall, Thurgood.” In The Oxford Companion to the Supreme Court of the United States. Kermit L. Hall, Editor. New York: Oxford University Press. Brace, Paul, Laura Langer, and Melinda Gann Hall. 1998. “Measuring the Preferences of State Supreme Court Judges.” Presented at the Annual Meeting of the Midwest Political Science Association. Brehm, John and Scott Gates. 1997. Working, Shirking, and Sabotage: Bureaucratic Response to a Democratic Public. Ann Arbor: University of Michigan Press. Caminker, Evan H.. 1994. “Why Must Inferior Courts Obey Superior Court Precedent?” Stanford Law Review 46:817-873. Canon, Bradley C. 1973. “Reactions of State Supreme Courts to a US. Supreme Court Civil Liberties Decision.” Law and Society Review 8: 109-34. Canon, Bradley C. and Charles A. Johnson. 1999. Judicial Policies: Implementation and Impact. 2'“i ed. Washington DC: CQ Press. Carp, Robert A. and C. K. Rowland. 1983. Palicymaking and Politics in Federal District Courts. Knoxville: University of Tennessee Press. Carp, Robert A. and Ronald Stidham. 1996. The Federal Courts. Washington DC: CQ Press. Criminal Justice Institute. 1976. Law of Admissions and Confessions. Criminal Justice Sedes Davis, Sue and Donald R. Songer. 1988. “The Changing Role of the United States Courts of Appeals: The Flow of Litigation Revisited.” The Justice System Journal 13:323-340. 184 Ditslear, Corey A. and Kevin M. Scott. 1999. “The Behavior of Elevated Judges: A Cross-Jurisdictional Analysis.” Prepared for Presentation at the 1999 Annual Meeting of the Midwestern Political Science Association, Chicago, IL, April 15- 17. Eggertsson, Thrainn. 1990. Economic behavior and institutions. Cambridge: Cambridge University Press. Eliason, Scott R. 1993. Maximum Likelihood Estimation: Logic and Practice. Newbury Park: Sage. Epstein, Lee, Jeffrey A. Segal, Harold J. Spaeth, and Thomas G. Walker. 1996. The Supreme C aurt C ampendium: Data, Decisions and Developments. 2'“l Ed. Washington, D.C.: CQ Press. Everson, David H. 1972. Ed. The Supreme Court as Policy-Maker: Three Studies on the Impact of Judicial Decisions. 2"d Edition. Carbondale, IL: Public Affairs Research Bureau. Fisher, Franklin M. 1958. "The Mathematical Analysis of Supreme Court Decisions: The Use and Abuse of Quantitative Methods." American Political Science Review. 52:321-338. George, Tracey E. and Lee Epstein. 1992. “On the Nature of Supreme Court Decision Making.” American Political Science Review 86:323-337. Gibson, James L. 1998. “Selecting Units for Analysis.” Paper presented at the Third Annual Conference on the Scientific Study of Judicial Politics. East Lansing: Michigan State University, October 3-5. Giles, Michael W., Virginia A. Hettinger, and Todd C. Peppers. 1998. “Alternative Measures of Preferences for Judges of the Courts of Appeals.” Presented a the Annual Meeting of the Midwest Political Science Association, Chicago, IL. Glick, Henry R. 1994. “The Impact of Permissive Judicial Policies: The US. Supreme Court and the Right to Die.” Political Research Quarterly 47:207-222. Goldman, Sheldon. 1966. “Voting Behavior on the United States Courts of Appeals, 1961-1964.” American Political Science Review 60:374-383. Goldman, Sheldon. 1975. “Voting Behavior 11 the United States Courts of Appeals Revisited.” American Political Science Review 69:491-506. Granato, Jim. 1991. “An Agenda for Econometric Model Building.” Political Analysis 3: 123-154. 185 Grano, Joseph D. 1996. Confessions, Truth, and the Law. Ann Arbor: The University of Michigan Press. Greene, William C. 1997. Econometric Analysis. 3" Ed. New York: Prentice Hall. Gruhl, John. 1980. “The Supreme Court’s Impact on the Law of Libel: Compliance By Lower Federal Courts.” Western Political Quarterly 33:502-519. Hagle, Timothy M. 1991. “But Do They Have to See It to Know It? The Supreme Court's Obscenity and Pornography Decisions.” Western Political Quarterly. 44:1039-1054. Hagle, Timothy M. and Glenn E. Mitchell 11. 1992. “Goodness-of-Fit Measures for Probit and Logit.” American Jaunral ofPaliticaI Science 36:762-784. Hagle, Timothy M. and Harold J. Spaeth. 1992. “The Emergence of a New Ideology: The Business Decisions of the Burger Court.” The Journal afPalitics. 54: 120- 134. Hall, Melinda Gann and Paul Brace. 1992. “Toward an Integrated Model of Judicial Voting Behavior.” American Politics Quarterly 20: 147-68. Hamilton, Lawrence C. 1992. Regression With Graphics: A Second Course in Applied Statistics. Belmont, CA: Duxbury Press. Howard, J. Woodford. 1981. Courts of Appeals in the Federal Judicial System. Princeton: Princeton University Press. Humphries, Martha Anne, Tammy A. Sarver, and Donald R. Sanger. 1998. “Going All the Way: How to Seduce the Supreme Court (into granting cert).” Presented at the 1998 Meeting of the American Political Science Association. Ignagni, Joseph 1990. Explaining and Predicting Supreme Court Decision-making: The Establishment Clause Cases, 19 70-1986. Unpublished Ph.D. Dissertation. East Lansing: Michigan State University. Inbau, Fred E. and John E. Reid. 1962. Criminal Interrogation and Confessions. Baltimore: The Williams and Wilkins Company. Israel, Jerold H. and Wayne R. LaF ave. 1971. Criminal Procedure In a Nutshell. St. Paul: West Publishing. Johnson, Charles A. 1987. “Law, Politics, and Judicial Decision Making: Lower Federal Court Uses of Supreme Court Decisions.” Law and Society Review 21:325-340. 186 Johnson, Charles A. and Bradley C. Canon 1984. Judicial Policies: Implementation and Impact. Washington DC: CQ Press. Kiewiet, D. Roderick and Matthew D. McCubbins. 1991. The Logic of Delegation: C angressianal parties and the appropriation process. Chicago: University of Chicago Press. King, Gary. 1986. “How Not to Lie With Statistics: Avoiding Common Mistakes in Quantitative Political Science.” American Journal ofPalitical Science 30:666- 687. King, Gary. 1989. Unifying Political Methodology: The Likelihood Theory of Statistical Inference. Cambridge University Press. King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry : Scientific Inference in Qualitative Research. Princeton, NJ. : Princeton University Press. Kort, Fred. 1957. "Predicting Supreme Court Decisions Mathematically: A Quantitative Analysis of the 'Right to Counsel' Cases. " American Political Science Review. 51: 1-12. Kort, Fred. 1958. "Reply to Fisher's Mathematical Analysis of Supreme Court Decisions. " American Political Science Review. 52:339-348. Kort, Fred. 1963. "Content Analysis of Judicial Opinions and Rules of Law. " In G. Schubert (Ed) Judicial Decision Making. Glencoe, IL: Free Press. Kort, Fred. 1964. "Simultaneous Equations and Boolean Algebra in the Analysis of Judicial Decisions. " Law and Contemporary Problems. Kort, Fred. 1966. "Quantitative Analysis of Fact-Pattems in Cases and Their Impact on Judicial Decisions." Harvard Law Review. Vol. 79:1551, 1966. Manwaring, David R. 1972. “The Impact of Mapp v. Ohio.” In David H. Everson, ed. The Supreme Court as Policy-Maker: Three Studies on the Impact Of Judicial Decisions. 2"" Edition. Carbondale, H3: Public Affairs Research Bureau. Miller, Gary J. 1992. Managerial Dilemmas: the political economy of hierarchy. New York: Cambridge University Press. Miller, Gary J. and Terry M. Moe. 1986. “The Positive Theory of Hierarchies.” In Herbert F. Weisberg, ed. Political Science: The Science afPalitics. 1986. New York: Agathon Press. 187 Moe, Terry M. 1990. “Political Institutions: The Neglected Side of the Story.” Journal of Law, Economics, and Organization. Vol. 6, Special Issue 1990:213-253. Morris, Jeffrey B. “Burger, Warren Earl.” In The Oxford Companion to the Supreme Court of the United States. Kermit L. Hall, Editor. New York: Oxford University Press. Murphy, Walter F. 1959. “Lower Court Checks on Supreme Court Power.” Americrm Political Science Review 53 : 1017-103 1. Pacelle, Richard L. and Lawrence Baum. 1992. “Supreme Court Authority in the Judiciary: A Study of Remands.” American Politics Quarterly 20: 169-191. Perry, Jr., H.W. 1991. Deciding to Decide: Agenda Setting in the United States Supreme Court. ” Cambridge: Harvard University Press. Pinello, Daniel R. 1998. “Linking Political Party to Judicial Ideology in American Courts: A Meta-Analysis.” Presented at the Annual Meeting of the American Political Science Association, Boston, MA. Pratt, John W. and Richard J. Zeckhauser. 1985. Principals and Agents: The Structure of Business. Boston: Harvard Business School Press. Pratt, John W. and Richard J. Zeckhauser. 1985. “Principals and Agents: An Overview.” In John W. Pratt and Richard J. Zeckhauser, eds. 1985. Principals and Agents: The Structure of Business. Boston: Harvard Business School Press. Reddick, Malia. 1997. The Applicability of the Legal and Attitudinal Models to the Treatment of Precedent in the Courts Of Appeals. Unpublished Ph.D. dissertation. East Lansing: Michigan State University. Redmond, Karen. 1996. “Record-Setting Workloads Confi'ont Federal Judges.” [Online] Available http:\\www.uscourts.gov\Press_Release\workld.htm, June 24, 1996. Reid, Traciel V. 1987. “Judicial Policy-Making and Implementation: An Empirical Examination.” Western Political Quarterly 41:509-527. Richardson, Richard J. and Kenneth N. Vines 1970. The Politics of Federal Courts: Lower Courts in the United States. Boston: Little, Brown and Company. Ringel, William E. 1966. Arrests, Searches, Confessions. Jamaica, NY: Gould Publications. 188 Romans, Neil T. 1974. “The Role of State Supreme Courts in Judicial Policy Making: Escobedo, Miranda and the Use of Judicial Impact Analysis.” Western Political Quarterly 27:38-59. Rutledge, Devallis. 1994. Criminal Interrogation: Law and Tactics. Placerville, CA: Copperhouse Publishing Company. Sanders, Francine. 1995. “Brown v. Board of Education: An Empirical Reexamination of Its Effects on Federal District Courts.” Law and Society Review 29:731-756. Schafer, William J., 111. 1968. Confessions and Statements. Springfield, IL: Charles C. Thomas, Publisher. Segal, Jeffrey A. 1984. "Predicting Supreme Court Cases Probabilistically: The Search and Seizure Cases, 1962-1981." American Political Science Review. 78:891-900. Segal, Jeffrey A. 1986. "Supreme Court Justices as Human Decision Makers: An Individual-Level Analysis of the Search and Seizure Cases." Journal afPalitics. 48:938-955. Segal, Jeffrey A. and Albert D. Cover. 1989. “Ideological Values and the Votes of US. Supreme Court Justices.” American Political Science Review 83:557-564. Segal, Jeffrey A., Lee Epstein, Charles M. Cameron and Harold J. Spaeth. 1995. “Ideological values and the votes of US. Supreme Court Justices Revisited.” The Journal ofPalitics 57:812-823. Segal, Jeffrey A., Donald R. Songer, Charles M. Cameron. 1995. “Decision Making on the US. Courts of Appeals.” In Lee Epstein, ed. Contemplating Courts. 1995. Washington, D.C.: CQ Press, Inc. Segal, Jeffrey A. and Harold H. Spaeth. 1992. The Supreme Court and the Attitudinal Model. New York: Cambridge University Press. Sheehan, Reginald 8., William Mishler and Donald R. Songer. 1992. “Ideology, Status, and the Differential Success of Direct Parties Before the Supreme Court.” American Political Science Review 86:464-471. Shepsle, Kenneth A. and Mark S. Bonchek. 1996. Analyzing Politics: Rationality, Behavior, and Institutions. Draft. Cambridge, MA: Harvard University Press. Sanger, Donald R. 1987a. “Case Selection in Judicial Impact Research.” The Western Political Quarterly. 41 2569-584. 189 Songer, Donald R. 1987b. “The Impact of the Supreme Court on Trends in Economic Policy Making in the United States Courts of Appeals.” The Journal of Politics 49:830-841. Songer, Donald R. and Susan Davis. 1989. “The Impact of Party and Region on Voting Decisions in the United States Courts of Appeals, 1955-1986.” The Western Political Quarterly 43 :3 17-333. Songer, Donald R. and Susan Haire. 1992. "Integrating Alternative Approaches to the Study of Judicial Voting: Obscenity Cases in the United States Court of Appeals.” American Journal of Political Science. 362963 -982. Songer, Donald R., Jeffrey A. Segal, and Charles M. Cameron. 1994. "The Hierarchy of Justice: Testing a Principal-Agent Model of Supreme Court-Circuit Court Interactions." American Journal of Political Science. 38:673-696. Songer, Donald R. and Reginald S. Sheehan. 1990. "Supreme Court Impact on Compliance and Outcomes: Miranda and New York Times in the United States Courts of Appeals." Western Political Quarterly. 43 :297-3 19. Spaeth, Harold J. 1994. The Supreme Court Judicial Database and the Expanded Supreme Court Database. Ann Arbor: Inter-University Consortium for Political and Social Research. Spriggs, James and Thomas Hansford. 1999. “Measuring Legal Change: The Reliability and Validity of Shepard’s Citation.” Unpublished manuscript. Stephens, Otis H., Jr. 1973. The Supreme Court and Confessions of Guilt. Knoxville: The University of Tennessee Press. Stevens, Joe B. 1993. The Economics of Collective Choice. Boulder: Westview Press. Stidham, Ronald and Robert A. Carp 1982. “Trial Courts’ Responses to Supreme Court Policy Changes.” Law and Policy Quarterly 41215-234. Tarr, G. Alan. 1977. Judicial Impact and State Supreme Courts. Lexington, MA: Lexington Books. Ulmer, S. Sidney. 1963. "Quantitative Analysis of Judicial Processes: Some Practical and Theoretical Applications. " Law and Contemporary Problems. 28: 164-176. Van Meter, CH. 1973. Principles of Police Interrogation. Springfield, IL: Thomas. Wasby, Stephen L. 1970. The Impact of the United States Supreme Court: Some Perspectives. Homewood, IL: Dorsey Press. 190 Wasby, Stephen L. 1972. “The Pure and the Prurient: The Supreme Court, Obscenity, and Oregon Policy.” In David H. Everson, ed. The Supreme Court as Policy- Maker: Three Studies on the Impact of Judicial Decisions. 2'“l Edition. Carbondale, IL: Public Affairs Research Bureau. Wasby, Stephen L. 1973. “The Communication of the Supreme Court’s Criminal Procedure Decisions: A Preliminary Mapping.” Villanova Law Review 18: 1086. Wasby, Stephen L. 1992. “Blackmun, Harry Andrew.” In The Oxford Companion to the Supreme Court of the United States. Kermit L. Hall, Editor. New York: Oxford University Press. Wasby, Stephen L. 1999. “Internal court procedure.” Email to the Law and Courts Discussion List (lawcourts—l@usc.edu). Howard Gillman, Moderator. Sponsored by the Law and Courts Section of the American Political Science Association. White, Harrison C. 1985. “Agency as Control.” In John W. Pratt and Richard J. Zeckhauser, eds. 1985. Principals and Agents: The Structure of Business. Boston: Harvard Business School Press. Wigmore, John Henry. 1942. Wigmore ’s Code of the Rules of Evidence in Trials at Law. Boston: Little, Brown and Company. Wrightsman, Lawrence S. and Saul M. Kassin. 1993. Confessions in the Courtroom. Newbury Park: Sage Publications. Zuk, Gary, Deborah J. Barrow, Gerard S. Gryski. Auburn University. A Mulit-User Data Base on the Attributes of US. Appeals Court Judges, 1801-1994. NSF Grant #SBR-93-1 1999. 19]