Charting the influences on the judicial mind: An empirical study of judicial reasoning

New York University Law Review

Volumn 73, Issue 5, 1998, Pages 1377-1500

  Sisk, Gregory C ^a   Heise, Michael ^b,c   Morriss, Andrew P ^d  

^a DRAKE UNIVERSITY   (United States)

^b Department of Law ^*  (United States)

^c INDIANA UNIVERSITY   (United States)

^d CASE WESTERN RESERVE UNIVERSITY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 22444452813     PISSN: 00287881     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (570)

1. See infra Part II.A.2.
2. See infra Part II.A.2.
3. Cf. Glendon Schubert, The Judicial Mind (1965);
4. Glendon Schubert, The Judicial Mind Revisited (1974). Schubert developed a psychometric model of Supreme Court decisionmaking under which the beliefs of the justices motivate their voting behavior. By contrast, we use the term "judicial mind" here more generically to describe the reasoning process by which a judge reaches a result.
5. See Thomas G. Walker & Deborah J. Barrow, The Diversification of the Federal Bench: Policy and Process Ramifications, 47 J. Pol. 596, 603 (1985) (stating that obtaining responses by judges "to identical case stimuli . . . will never be the case in studies of trial court decision making");
6. see also Jon Gottschall, Reagan's Appointments to the U.S. Courts of Appeals: The Continuation of a Judicial Revolution, 70 Judicature 48, 51 (1986) (stating that "the problem of comparability of cases cannot be completely resolved at the lower federal court level"). As this study demonstrates, the prediction that this scenario "never" would arise proved to be too pessimistic, but "rarely" would be true, as our archetypal scenario appears to be unique and is unlikely to be often, if ever, replicated.
7. See infra Part II.B.2-3.
8. Sentencing Reform Act of 1984 (Title II of the Comprehensive Crime Control Act of 1984), Pub. L. No. 98-473, 98 Stat. 1837, 1987-2034 (codified as amended in scattered sections of 18, 28 U.S.C.).
9. 28 U.S.C. § 994 (1994).
10. See United States Sentencing Commission, Sentencing Guidelines and Policy Statements (1987) [hereinafter Sentencing Guidelines and Policy Statements].
11. See infra Part II.B.1.
12. See United States v. Ortega Lopez, 684 F. Supp. 1506, 1520 & n.12 (C.D. Cal. 1988) (en banc) (Hupp, J., dissenting) (describing "chaos" caused by uncertainty in sentencing practice and procedure during this period).
13. See Mistretta v. United States, 488 U.S. 361 (1989).
14. See infra Part III.
15. See infra Part IV.B.
16. Joel B. Grossman, Social Backgrounds and Judicial Decision-Making, 79 Harv. L. Rev. 1551, 1552 (1966). See generally infra Part II.A.1 (discussing mixed results of prior studies seeking to find connection between judges' social backgrounds and voting behaviors).
17. Paul D. Carrington, Of Law and the River, 34 J. Legal Educ. 222, 227 (1984).
18. See infra Part V.A.
19. See infra Part V.C.
20. See infra Part V.C.1.
21. See infra Part V.C.3.
22. See infra Part V.E.
23. See Mark A. Cohen, Explaining Judicial Behavior or What's "Unconstitutional" about the Sentencing Commission?, 7 J.L. Econ. & Org. 183, 184 (1991) (conducting study of Sentencing Guidelines constitutionality decisions by 196 judges).
24. See infra Part V.A.2.
25. See infra Part V.B.
26. Specific results of prior studies are further developed in Part V of this article, which discusses and interprets the results of this study by sets of independent variables.
27. Norman Dorsen, A Change in Judicial Philosophy?, Nat'l L.J., Feb. 18, 1985, at 13;
28. see also S. Sidney Ulmer, Are Social Background Models Time-Bound?, 80 Am. Pol. Sci. Rev. 957, 957 (1986) (calling Dorsen's comment "about as concise a statement of the social background theory as could be penned").
29. See C. Neal Tate & Roger Handberg, Time Binding and Theory Building in Personal Attribute Models of Supreme Court Voting Behavior, 1916-88, 35 Am. J. Pol. Sci. 460, 460-61 & n.2 (1991);
30. see also Henry R. Glick, Courts, Politics, and Justice 313 (3d ed. 1993) ("The study of judges' personal backgrounds assumes basically that people behave according to who they are."). An alternative approach examines courts as the unit of analysis and seeks to determine the influence of aggregate measures of a jurisdiction's characteristics (e.g., method of judicial selection) on evolution of doctrine.
31. See, e.g., Bradley C. Canon & Lawrence Baum, Patterns of Adoption of Tort Law Innovations: An Application of Diffusion Theory to Judicial Doctrines, 75 Am. Pol. Sci. Rev. 975, 975 (1981) (studying whether judicial doctrines diffuse on regional basis, and whether organizations adopt innovations consistently across issues, based on underlying characteristics of organization);
32. Andrew P. Morriss, Developing a Framework for Empirical Research on the Common Law: General Principles and Case Studies of the Decline of Employment-at-Will, 45 Case W. Res. L. Rev. 999, 1020-56 (1995) (studying content of judicial opinions in context of evolution of common law doctrine).
33. See Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model 231-34 (1993) (concluding that empirical evidence provides little support for relationship between social background factors and judicial attitudes);
34. Orley Ashenfelter, Theodore Eisenberg & Stewart J. Schwab, Politics and the Judiciary: The Influence of Judicial Background on Case Outcomes, 24 J. Legal Stud. 257, 281 (1995) (concluding, in context of federal district court outcomes in civil rights cases, "that individual judge characteristics cannot be assumed to influence substantially the mass of cases");
35. Gerard S. Gryski & Eleanor C. Main, Social Backgrounds as Predictors of Votes on State Courts of Last Resort: The Case of Sex Discrimination, 39 W. Pol. Q. 528, 528-29, 536 (1986) (describing criticism of social background theory but concluding that theory remains viable with significant limitations);
36. see also Peter J. Van Koppen & Jan Ten Kate, Individual Differences in Judicial Behavior: Personal Characteristics and Private Law Decision-Making, 18 L. & Soc'y Rev. 225, 225-41 (1984) (finding that Dutch judicial decisions in civil cases are only moderately influenced by personality characteristics of judges, measured by psychological tests through questionnaire to judges participating in simulation).
37. See Sue Davis, Susan Haire & Donald R. Songer, Voting Behavior and Gender on the U.S. Courts of Appeals, 77 Judicature 129, 130 (1993) (describing prior empirical research on behavior of women decisionmakers, including judges, as producing "mixed results");
38. Sheldon Goldman, Voting Behavior on the U.S. Courts of Appeals Revisited, 69 Am. Pol. Sci. Rev. 491, 496 (1975) (describing multitude of prior studies on relationship of background variables to judicial voting behavior as having "mixed" results);
39. John Gruhl, Cassia Spohn & Susan Welch, Women as Policymakers: The Case of Trial Judges, 25 Am. J. Pol. Sci. 308, 311 (1981) (describing studies using judges' sex as independent variable as yielding "mixed results"); Tate & Handberg, supra note 26, at 470 (describing results of prior studies on influence of prior judicial experience as "mixed"); Ulmer, supra note 25, at 957 (describing prior research on social background influences upon judicial decisionmaking as producing "mixed results").
40. See, e.g., Davis, Haire & Songer, supra note 28, at 131-32 (finding no significant differences between male and female judges in search and seizure and obscenity cases, when controlling for party of appointing president, although finding female judges more liberal in employment discrimination cases); Jon Gottschall, Carter's Judicial Appointments: The Influence of Affirmative Action and Merit Selection on Voting on the U.S. Courts of Appeals, 67 Judicature 164, 171-73 (1983) (finding relative similarity between President Carter's male and female appointees to courts of appeals and, with exception of criminal cases, minimal variances between black and white judges, even in racial discrimination cases); Gruhl, Spohn & Welch, supra note 28, at 318-20 (finding few significant differences in conviction rates of male and female judges, although finding female judges more likely to sentence female defendants to prison);
41. Herbert M. Kritzer & Thomas M. Uhlman, Sisterhood in the Courtroom: Sex of Judge and Defendant in Criminal Case Disposition, Soc. Sci. J., April 1977, at 77, 86 (concluding that female judges "behave no differently than their male colleagues" in study of sentencing);
42. Cassia Spohn, The Sentencing Decisions of Black and White Judges: Expected and Unexpected Similarities, 24 L. & Soc'y Rev. 1197, 1211-14 (1990) (finding "remarkable" similarities in sentencing decisions of black and white judges and concluding that race of judge has little predictive power);
43. Thomas M. Uhlman, Black Elite Decision Making: The Case of Trial Judges, 22 Am. J. Pol. Sci. 884, 891-94 (1978) (finding no important differences between black and white judges in criminal conviction rates and sentencing); Walker & Barrow, supra note 4, at 613-15 (finding marked similarity in decisionmaking records between black and white federal district judges in several fields and few differences between male and female judges, with exception of tendency of female judges to rule in favor of government entities);
44. Susan Welch, Michael Combs & John Gruhl, Do Black Judges Make a Difference?, 32 Am. J. Pol. Sci. 126, 131-35 (1988) (finding little impact of black judges in overall severity of criminal sentencing, but finding evidence of more equal treatment by black judges of white and black defendants in decisions to incarcerate);
45. Richard L. Fox & Robert W. Van Sickel, Gender Dynamics and Judicial Behavior in Criminal Trial Courts 12-13, 20-22 (1998) (unpublished manuscript on file with the New York University Law Review) (finding no clear pattern of male and female traits or "voice" in exercise of judicial discretion by local criminal judges, although finding some gender differences, such as that female judges were more likely to rely on prosecution while male judges were more likely to side with defense);
46. Kenneth Luis Manning, Como Decide? (How Do You Decide?): Decision-Making by Hispanic Judges in the Federal District Courts 6 (1998) (unpublished manuscript on file with the New York University Law Review) (finding that "the difference in ideology of Hispanic and Non-Hispanic judge decisions [on federal district courts] is virtually nonexistent").
47. See J. Woodford Howard, Jr., Courts of Appeals in the Federal Judicial System 182-83 (1981) (finding in study of circuit judges' votes across multiple fields that prior judicial experience was significant only on discrete issue of civil rights); Ashenfelter, Eisenberg & Schwab, supra note 27, at 277-81 (finding that individual judge characteristics, including prior judgeship, did not appear to influence substantially mass of cases decided by district court judges); Gryski & Main, supra note 27, at 532 (finding that prior career characteristics of judges were "not useful predictors of state high court judicial behavior in sex discrimination cases").
48. See Tate & Handberg, supra note 26, at 470 (describing finding in C. Neal Tate, Personal Attribute Models of the Voting Behavior of United States Supreme Court Justices, 75 Am. Pol. Sci. Rev. 355, 361-63 (1981)).
49. For example, prosecutorial experience has been associated with a more conservative behavior in civil liberties cases, see id. at 474-76, but with a more liberal or favorable response to racial equal protection claims, see Theodore Eisenberg & Sheri Lynn Johnson, The Effects of Intent: Do We Know How Legal Standards Work?, 76 Cornell L. Rev. 1151, 1190 (1991);
50. see also Stuart S. Nagel, Judicial Backgrounds and Criminal Cases, 53 J. Crim. L. & Criminology 333, 336 (1962) (finding in early study that former prosecutors were significantly more likely to vote against defense in criminal cases).
51. Don Bowen, The Explanation of Judicial Voting Behavior from Sociological Characteristics of Judges (1965) (unpublished Ph.D. dissertation, Yale University) (on file with author),
52. quoted in S. Sidney Ulmer, Social Background as an Indicator to the Votes of Supreme Court Justices in Criminal Cases: 1947-1956 Terms, 17 Am. J. Pol. Sci. 622, 622 (1973); see also Howard, supra note 30, at 182 (finding, in study of decisions by circuit judges in multiple fields, that "[n]o single background characteristic was a strong determinant of voting outcomes across the board"); Gryski & Main, supra note 27, at 528 (acknowledging that "social background factors have not proven to be particularly effective means by which to explain judicial behavior").
53. See Walker & Barrow, supra note 4, at 604-11.
54. See Davis, Haire & Songer, supra note 28, at 131-32. But see Gryski & Main, supra note 27, at 531-32, 536 (finding sex not to be significant influence in study of state high court judges in sex discrimination cases, although number of female judges was too small for reliable conclusions).
55. See Gottschall, supra note 29, at 172-73.
56. See Walker & Barrow, supra note 4, at 613-15 (finding that "black and white [federal district] judges displayed markedly similar decision-making records" in multiple legal fields, including criminal law and procedure). Studies of trial judges have consistently found little or no difference between black and white judges in adjudication of criminal cases, including conviction rates and sentencing severity. See, e.g., Spohn, supra note 29, at 1211-14; Uhlman, supra note 29, at 891-94; Welch, Combs & Gruhl, supra note 29, at 131-35.
57. See Howard, supra note 30, at 182-83, 186 (finding, in study of circuit judge decisions across multiple fields, that party identification was weakest indicator on votes, and concluding that "[t]he predictive power of political indicators was negligible and indirect"); Ashenfelter, Eisenberg & Schwab, supra note 27, at 281 (concluding in study of district court decisions "that we cannot find that Republican judges differ from Democratic judges in their treatment of civil rights cases").
58. See, e.g., C. K. Rowland & Robert A. Carp, Politics and Judgment in Federal District Courts 24-57 (1996);
59. Jilda M. Aliotta, Combining Judges' Attributes and Case Characteristics: An Alternative Approach to Explaining Supreme Court Decisionmaking, 71 Judicature 277, 278, 280 (1988);
60. Frank B. Cross & Emerson H. Tiller, Judicial Partisanship and Obedience to Legal Doctrine: Whistleblowing on the Federal Courts of Appeals, 107 Yale L.J. 2155, 2175-76 (1998); Goldman, supra note 28, at 496-506; Gottschall, supra note 29, at 169-71; Gottschall, supra note 4, at 51-54; Gryski & Main, supra note 27, at 531, 533-34;
61. Stuart S. Nagel, Political Party Affiliation and Judges' Decisions, 55 Am. Pol. Sci. Rev. 843, 845 (1961);
62. Richard L. Revesz, Environmental Regulation, Ideology, and the D.C. Circuit, 83 Va. L. Rev. 1717, 1717-21, 1738-66 (1997).
63. See generally American Court Systems 382 (Sheldon Goldman & Austin Sarat eds., 2d ed. 1989).
64. See Rowland & Carp, supra note 39, at 24 (noting that "Democratic judges are 1.42 times more likely to render a liberal decision than are judges of Republican backgrounds"); Gryski & Main, supra note 27, at 534.
65. See Rowland & Carp, supra note 39, at 46 (finding that judicial "appointees of Democratic presidents are clearly more liberal" in decisionmaking than judges chosen by Republicans); Gottschall, supra note 29, at 169-71; Gottschall, supra note 4, at 51. For a discussion of the relationship between a judge's party identification and the party of the appointing president, see infra Part III.B.
66. See Rowland & Carp, supra note 39, at 34 (finding difference of 10-13% between Democratic and Republican cases for all types of cases). However, there were significantly higher partisan voting differences in certain areas, such as race discrimination and religion. See id. at 40 (finding difference between Democratic and Republican judges of 28% on race discrimination cases and 24% on religion cases); id. at 48-50 (finding some dramatic voting differences between Carter and Reagan appointed judges on such issues as race (60%) and right to privacy (33%)); see also Gottschall, supra note 4, at 53 (finding, in study of court of appeals judges, that when looking at results in universe of both unanimous and nonunanimous cases, margin of difference between appointees of Democratic and Republican presidents was 20% in civil rights and liberties cases and 10% in economic cases).
67. See Rowland & Carp, supra note 39, at 56. At least two studies of judicial behavior post-1986 have confirmed the decline of partisan variance among federal judges. See Nancy Scherer, Reexamining the Politics of Crime in the Federal Courts: Are Bill Clinton's Judicial Appointees "New" Democrats or "Old" Democrats? 29 (1998) (unpublished manuscript, on file with the New York University Law Review) (concluding that "partisan affiliation is no longer sufficient to distinguish between the voting behavior of federal appeals court judges, at least with respect to issues of criminal law and civil liberties," and finding in study of search and seizure cases that voting behavior of Clinton's judicial nominees to federal appellate courts is "virtually indistinguishable" from that of appointees of his Republican predecessor);
68. see also Ronald Stidham, Robert A. Carp & Donald R. Songer, The Voting Behavior of President Clinton's Judicial Appointees, 80 Judicature 16, 19-20 (1996) (concluding that Clinton's appointees have demonstrated moderate decisional tendencies, and finding small differences in "liberal" voting rates, generally under 10% across categories of cases, for both district and court of appeals judges). However, another recent study found significant partisan voting on the multimember appellate panels of the District of Columbia Circuit. See Revesz, supra note 39, at 1717-19. Revesz concluded that "judges generally vote consistently with their ideological preferences only when they sit with at least one other judge of the same political party" and therefore prior studies that "fail[ed] to control for such panel composition effects" have "substantial[ly] underestimat[ed] the frequency of ideological voting." Id. at 1719-20. However, Revesz acknowledged that "judges on the D.C. Circuit have a far higher political profile than do federal judges generally" and "[t]hus ideology might have a greater-than-average impact on the votes of these judges than on the universe of federal circuit court judges." Id. at 1720-21.
69. Rowland & Carp, supra note 39, at 14.
70. This is especially true when most such studies rely on published opinions which, particularly at the federal district court level, are an unrepresentative sample of peculiarly controversial cases. See id. at 16 (noting that less than five percent of district court decisions are published and that "published decisions tend to be policy judgments with greater political consequences than their unpublished counterparts"). See generally Susan M. Olson, Studying Federal District Courts Through Published Cases: A Research Note, 15 Just. Sys. J. 782 (1992) (arguing that published decisions are not representative of all district court cases). When unpublished decisions are examined, partisan differences fade considerably, although this may be at least partially attributable to the routine, precedent-bound nature of such cases that leaves less room for expression by judges of political proclivities. See Rowland & Carp, supra note 39, at 134 (finding that while link between political appointment and judicial rulings may extend to certain types of unpublished decisions, "the magnitude and consistency of appointment effects on unpublished opinions do not compare with appointment effects on published decisions"); Ashenfelter, Eisenberg & Schwab, supra note 27, at 258-60, 281 (finding no partisan difference in district court rulings in day-to-day docket). But see Rowland & Carp, supra note 39, at 21 (reporting other studies finding that differences in partisan voting by district judges between published and unpublished opinions were negligible); see also Morriss, supra note 26, at 1038-47 (summarizing literature on unpublished decisions and their role in empirical work).
71. See Ulmer, supra note 25, at 961 (stating that, unlike other experiential variables, party choice is "a surrogate for the aggregate of attitudes" that other experiences produce); see also American Court Systems, supra note 39, at 382 (stating that party affiliation and appointment by particular president are "likely reasonably accurate surrogates for attitudes and values").
72. Segal & Spaeth, supra note 27, at 232; see also Ulmer, supra note 25, at 961 (stating that "[i]t may not be strictly logical . . . to argue that party 'causes' or influences judicial votes" because party choice is "a surrogate for the aggregate of attitudes" possessed by judge).
73. See Howard, supra note 30, at xxiii (stating that "judicial roles become intervening variables between institutional and personality factors in the judicial process"); Frank B. Cross, Political Science and the New Legal Realism: A Case of Unfortunate Interdisciplinary Ignorance, 92 Nw. U. L. Rev. 251, 299 (1997) ("The judicial process is meant to create a role orientation centering decisions on the legal model. While research into judicial role orientation is somewhat limited, there is reason to think that role responsibilities play an important part in judging.").
74. Rowland & Carp, supra note 39, at 136. They observe that, at least in terms of self-perception, "many trial judges appear to be motivated primarily by their role orientation - that is, their 'secondary' perceptions of what a judge should do - and not by their personal preferences." Id. at 190. "A role orientation is a psychological construct which is the combination of the occupant's perception of the role expectations of significant others and his or her own norms and expectations of proper behavior for a judge." James L. Gibson, Judges' Role Orientations, Attitudes, and Decisions: An Interactive Model, 72 Am. Pol. Sci. Rev. 911, 917 (1978). Gibson found that a judge's role orientation acts as an intervening variable, such that a judge's belief about the legitimacy of allowing nonlegal criteria to influence decisions does indeed affect "the relationship between attitudes and behavior." Id. at 922; see also American Court Systems, supra note 39, at 437 (explaining that "the judge's concept of role can inhibit the full flowering of political attitudes and values" in exercise of judicial discretion).
75. See Rowland & Carp, supra note 39, at 150-73 (proposing new model of trial court judgment as special case of "social judgment" within context of legal system that, while allowing discretion, also places meaningful constraints on judges' decisionmaking authority).
76. Richard A. Epstein, The Independence of Judges: The Uses and Limitations of Public Choice Theory, 1990 BYU L. Rev. 827, 831-32.
77. See, e.g., Richard A. Posner, What Do Judges and Justices Maximize? (The Same Thing Everybody Else Does), 3 Sup. Ct. Econ. Rev. 1, 2 (1993)
78. [hereinafter Posner, What Do Judges Maximize] ("The economic analyst has a model of how criminals and contract parties, injurers and accident victims, parents and spouses - even legislators, and executive officials such as prosecutors - act, but falters when asked to produce a model of how judges act.");
79. Edward L. Rubin, The New Legal Process, The Synthesis of Discourse, and the Microanalysis of Institutions, 109 Harv. L. Rev. 1393, 1399 (1996) ("[N]o stable [public choice] theory has emerged to explain the behavior of judges because most American judges, regardless of their substantive decisions, are insulated from monetary rewards or punishments, guaranteed of their position[s], and unable to affect their own jurisdiction in any direct fashion.").
80. Epstein, supra note 51, at 838.
81. Id. at 851.
82. See, e.g., Howard, supra note 30, at xix, 173-84 (reviewing 4,941 decisions by 35 judges of Second, Fifth, and District of Columbia Circuits, from fiscal years 1965-1967, in such categories as civil rights, criminal, and labor, with certain outcomes labeled liberal or conservative); Rowland & Carp, supra note 39, at 18 (reviewing 45,826 published opinions issued by 1500 district court judges from 1933-1987, categorized into 26 case types with liberal/conservative dimension); Gottschall, supra note 29, at 167-68 (reviewing all court of appeals decisions in criminal procedure, racial discrimination, and sex discrimination cases over two-year period, labeling votes in favor of claims by criminally accused and discrimination claimants as liberal); Gottschall, supra note 4, at 51 (reviewing all court of appeals decisions in multiple general fields, including racial or sexual discrimination and criminal procedure, during year-and-a-half period, with certain outcomes assigned liberal label).
83. See, e.g., Davis, Haire & Songer, supra note 28, at 130-31 (reviewing votes of all federal court of appeals judges from 1981-1990 in employment discrimination, criminal procedure, and obscenity cases); Goldman, supra note 28, at 492 (reviewing decisions by court of appeals judges in multiple issue areas, including criminal procedure, civil liberties, labor, and government fiscal); Gryski & Main, supra note 27, at 529, 536 (reviewing state high court rulings in sex discrimination cases from 1971-1981); Nagel, supra note 32, at 333 (reviewing criminal cases decisions from 1955 involving 313 state and federal supreme court judges listed in directory).
84. See Van Koppen & Ten Kate, supra note 27, at 226 ("The problem in working with actual decisions is to ensure that the different cases heard by different judges are comparable on all relevant dimensions."). See generally John M. Conley & William M. O'Barr, Fundamentals of Jurisprudence: An Ethnography of Judicial Decision Making in Informal Courts, 66 N.C. L. Rev. 467, 472-73 (1988) (discussing alternative problems of incomparability among cases or unrealistic simulation in empirical research of judicial decisionmaking). Conley and O'Barr also criticize prior quantitative studies as inadequate to "answer the question of what caused the result in a particular case," id. at 473, and adduce the limitations of a priori judgments by researchers in selecting the variables to be evaluated in a study, see id. at 474-75. They propose an alternative method of observing judges as they make decisions and then evaluating judicial statements of reasoning in a group workshop discussion applying conversational analysis. See id. at 475-79. However, the Conley-O'Barr method cultivates its own weaknesses, such as the absence of quantification, the difficulty in replication by other researchers, and heavy dependence on the subjective interpretations of the workshop participants. In any event, we have adhered to a traditional quantitative statistical model, which requires developing (through review of the raw data, prior research, and theory) independent variables for coding and analysis.
85. See Aliotta, supra note 39, at 277 (observing that "[b]oth studies of judicial attributes and judicial attitudes are concerned with judges' voting propensities over a large number of cases," such that "the facts or legal principles involved in particular cases are not considered relevant"); Gottschall, supra note 29, at 169 n.13 (observing that "lower federal court judges decide different cases in different settings and their votes are not directly comparable" and thus "generalizations from such quantitative data" about relative attitudes of lower court judges "must be assessed cautiously"). Even when the study focuses upon United States Supreme Court decisions, thereby including generally the same set of judges hearing the same cases, see, e.g., Segal & Spaeth, supra note 27; Tate & Handberg, supra note 26; Ulmer, supra note 33, the incomparability problem persists through the assumption that a justice's resolution of one discrete case in a field, such as criminal law, provides evidence of an attitude on that subject rather than evaluation of anomalous facts, circumstances, or legal doctrinal implications of that individual case. Moreover, because the number of justices included in a study of the Supreme Court is limited, such studies may produce poignantly detailed descriptions of the attributes, experiences, attitudes, philosophies, and predilections of a small, atypical collection of individual human beings, but add little knowledge about the general subject of judicial decisionmaking. In other words, detailed studies of Supreme Court justices may come dangerously close to biography rather than sociology. Some researchers have attempted to compensate for the incomparability of aggregate votes in different cases by incorporating certain fact-pattern or other case-specific characteristics to the analysis. See, e.g., Segal & Spaeth, supra note 27, at 216-21, 229-31 (in study of Supreme Court rulings in search and seizure cases, combining derived attitude values for justices and twelve fact-based factors, such as justification for search, place of intrusion, etc., as independent variables, resulting in greater prediction rate, although concluding that justices' attitudes are more important than facts of case in predicting votes); Aliotta, supra note 39, at 279-80 (devising case characteristic variables for equal protection cases, such as whether case involved race, fundamental rights, or education, or was brought as class action). While this approach is an important refinement of prior research techniques, the case characteristics chosen remain those the researcher deems important (based in part on prior research or content analysis of opinions) and cannot fully account for the multidimensional aspects of each individual case. Moreover, even while incorporating case characteristics into the analysis, these studies continue to focus upon general outcome votes, for example, in favor of or against a civil rights or an equal protection claim as the dependent variable. See, e.g., Segal & Spaeth, supra note 27, at 242-55; Aliotta, supra note 39, at 278. The divergent paths that would lead to the same result, that is, the judges' reasoning, are not explored.
86. Conley & O'Barr, supra note 57, at 474-75.
87. See Van Koppen & Ten Kate, supra note 27, at 225. Van Koppen and Ten Kate explain that Dutch civil cases generally do proceed with exchange of written documents, such that presentation of written protocols was a legitimate simulation of the tasks that face trial judges in that nation. See id. at 227.
88. Id. at 227.
89. Id. at 240.
90. See Goldman, supra note 28, at 491 (explaining that outcomes in general issue areas were examined in "basic political terms of who wins and who loses and by implication what political values are seemingly being fostered").
91. See Van Koppen & Ten Kate, supra note 27, at 227 (explaining that study asked participating judges "for their decisions but did not ask them to follow the common procedure in actual cases of providing justifications for the decisions reached").
92. Rowland & Carp, supra note 39, at 149.
93. The Sentencing Reform Act of 1984 was enacted as Chapter II of the Comprehensive Crime Control Act of 1984, Pub. L. No. 98-473, 98 Stat. 1837, 1987-2034 (codified as amended in scattered sections of 18, 28 U.S.C.).
94. 28 U.S.C. § 991(a) (1994).
95. See id. The judicial members of the Commission are not required to resign as federal judges while serving on the Commission. See id. § 992(c).
96. See id. §§ 991(a), 992(a).
97. See id. § 991(a).
98. Id. § 994(a)(1). Congress directed the Commission to formulate guidelines that "provide certainty and fairness in meeting the purposes of sentencing," while "avoiding unwarranted sentencing disparities among defendants with similar records who have been found guilty of similar criminal conduct" and "maintaining sufficient flexibility to permit individualized sentences" where appropriate. Id. § 991(b)(1)(B). The Guidelines are to establish, "for each category of offense involving each category of defendant, . . . a sentencing range." Id. § 994(b)(1). The range must be consistent with provisions of the federal criminal code and may vary by no more than 25% or six months from the minimum to the maximum, except that where the minimum sentence is 30 years or more, the maximum may be life imprisonment. See id. §994(b)(1)-(2). The Commission also has continuing responsibility to review the Sentencing Guidelines regularly, including considering petitions by defendants to modify the Guidelines. See id. § 994(o)-(u). Amendments to the Guidelines take effect automatically unless, within 180 days after they are reported specific legislation provides otherwise. See id. § 994(p).
99. See Sentencing Guidelines and Policy Statements, supra note 8, at 1.1-1.12. The Federal Sentencing Guidelines are, in a sense, nothing more than a set of instructions for one chart - the Sentencing Table. The goal of guidelines calculations is to arrive at numbers for the vertical (offense level) and horizontal (criminal history category) axes on the Sentencing Table grid, which in turn generate an intersection in the body of the grid. Each such intersection designates a sentencing range expressed in months. Frank O. Bowman, III, Coping With "Loss": A Re-Examination of Sentencing Federal Economic Crimes Under the Guidelines, 51 Vand. L. Rev. 461, 472-73 (1998) (footnote omitted).
100. For a readable and concise explanation of the Sentencing Guidelines "grid" and the calculation of a sentence under the Guidelines, see generally Frank O. Bowman, III, The Quality of Mercy Must be Restrained, and Other Lessons in Learning to Love the Federal Sentencing Guidelines, 1996 Wis. L. Rev. 679, 693-704
101. [hereinafter Bowman, Quality of Mercy].
102. See Comprehensive Crime Control Act of 1984, Pub. L. No. 98-473, tit. 2, ch. 2, § 235(a)(1)(B)(ii)(III), 98 Stat. 2032 (reprinted in note to 18 U.S.C. § 3551).
103. The House of Representatives defeated a proposal to delay implementation of the Guidelines. See 133 Cong. Rec. 26, 683-84 (1987).
104. See Sentencing Guidelines and Policy Statements, supra note 8 at 1.1, 1.12 (noting that "[b]ecause of time constraints and the nonexistence of statistical information, some offenses that occur infrequently are not considered in this initial set of guidelines," and that Commission had also deferred promulgation of Guidelines pertaining to certain sanctions for organizational defendants). It is somewhat inaccurate to term the Commission's work as mere "guidelines." The criminal code states that a sentencing court: [S]hall impose a sentence of the kind, and within the range [set forth in the Guidelines] unless the court finds that there exists an aggravating or mitigating circumstance of a kind, or to a degree, that was not adequately taken into consideration by the Sentencing Commission in formulating the guidelines [and] that should result in a sentence different from that described. 18 U.S.C. § 3553(b) (1994). The sentencing judge must state the reasons for imposing the sentence selected and must give "the specific reason for the imposition of a sentence different from that described" in the applicable guideline. Id. § 3553(c). Congress expected that less than 20% of sentences would be set outside the Guidelines. See S. Rep. No. 98-225, at 52 n.71 (1983), reprinted in 1984 U.S.C.C.A.N. 3182, 3235. On the power of sentencing judges to depart from the Guidelines, see generally Koon v. United States, 518 U.S. 81, 95-100 (1996) (expanding authority of district judges to depart downward from Guidelines by adopting abuse of discretion as standard for appellate review of district court evaluation of whether factors had been adequately considered by Commission in formulating guideline sentence).
105. 133 Cong. Rec. 26, 367 (1987) (remarks of Rep. Conyers).
106. Id. at 26, 372 (remarks of Rep. Synar).
107. See Bowman, Quality of Mercy, supra note 72, at 682 ("Prior to [enactment of the Guidelines], the judge had virtually unlimited discretion to sentence a convicted defendant anywhere within the range created by the statutory maximum and minimum penalties for the offense or offenses of conviction.").
108. On the history of sentencing, the problems in disparities in sentencing among judges, and Congress's conclusions in adopting a mandatory guideline system, see generally S. Rep. No. 98-225, at 37-65 (1983), reprinted in 1984 U.S.C.C.A.N. 3182, 3220-48. For concise statements of the history and background of sentencing reform and the Sentencing Guidelines from different positions (one favorable toward and one critical of the Guidelines), see Bowman, Quality of Mercy, supra note 72, at 680-704 (favorable); Charles J. Ogletree, Jr., The Death of Discretion? Reflections on the Federal Sentencing Guidelines, 101 Harv. L. Rev. 1938, 1940-51, 1951-60 (1988) (critical).
109. José A. Cabranes, Sentencing Guidelines: A Dismal Failure, 207 N.Y. L.J., Feb. 11, 1992, at 2
110. [hereinafter Cabranes, Dismal Failure].
111. See Bowman, Quality of Mercy, supra note 72, at 740 (stating that "the Guidelines, taken together with mandatory minimum sentences, compel the imposition of very long sentences on drug sellers," sentences that "are long relative to the previously settled expectations of the participants in the federal system");
112. Marvin E. Frankel, Sentencing Guidelines: A Need for Creative Collaboration, 101 Yale L.J. 2043, 2047 (1992) ("[T]he Commission produced guidelines that actually increase the overall severity [of federal sentences.]");
113. Kate Stith & Steve Y. Koh, The Politics of Sentencing Reform: The Legislative History of the Federal Sentencing Guidelines, 28 Wake Forest L. Rev. 223, 284-85 (1993) (stating that, in keeping with congressional intent, "the percentage of defendants being imprisoned and the length of prison terms have increased" under Sentencing Guidelines).
114. Charles Fried, Order and Law 165 (1991) ("The defense lawyers . . . hated the guidelines, which meant stiffer sentences and less to bargain over.").
115. See Ruth Marcus, Federal Sentences in Limbo as Guidelines Challenged, Wash. Post, Mar. 2, 1988, at A19 (reporting that defendants charged with crimes committed after November 1, 1987, effective date of new federal Sentencing Guidelines, were objecting to their constitutionality, resulting in spreading "epidemic of challenges");
116. David G. Savage, Court Fights Ahead Over New Sentencing Rules: Novel Attempt at Fairness is Attacked as a Violation of Separation of Powers, L.A. Times, Apr. 8, 1988, at 18 (reporting growing number of challenges to Sentencing Guidelines filed by defense lawyers across country and quoting leading participants from both sides as agreeing that result would be uncertainty, even chaos, in sentencing procedures until Supreme Court resolved issue).
117. Each of these categories of constitutional attacks serves as a dependent variable unit for statistical analysis in our study. See infra Part IV.B.2.
118. See, e.g., United States v. Velez-Naranjo, 691 F. Supp. 584, 587 (D. Conn. 1988) (accepting defendant's argument that Sentencing Commission exercises executive power and thus its placement in judicial branch offends Constitution); United States v. Tolbert, 682 F. Supp. 1517, 1523-25 (D. Kan. 1988) (holding that placement of Sentencing Commission within judiciary violates separation of powers as Commission's power to issue substantive Sentencing Guidelines goes beyond scope of judiciary's authority to adjudicate cases or controversies).
119. In this and the following paragraphs, we use the term "government" to refer to positions held both by the Department of Justice and by the Sentencing Commission. Whenever the two entities differed on the appropriate understanding of the Sentencing Commission's place in the constitutional scheme of government (the separation of powers issue), we have identified them individually as the "Department" or "Department of Justice" and "Sentencing Commission" or "Commission."
120. "A formalist approach seeks to maintain three distinct branches of government, and to prevent commingling of the three central powers of making the laws, executing the laws, and interpreting the laws." The Supreme Court, 1988 Term - Leading Cases, 103 Harv. L. Rev. 137, 280 n.6 (1989)
121. (citing Harold J. Krent, Separating the Strands in Separation of Powers Controversies, 74 Va. L. Rev. 1253, 1254 (1988)).
122. Draft Brief on the Constitutionality of the Work of the Sentencing Commission, Department of Justice, Civil Division (Redraft March 28, 1988) at 4, 8-17, 26-34 [hereinafter DOJ Model Brief] (on file with the New York University Law Review). This model brief, which was obtained from the Department of Justice under a Freedom of Information Act request, was filed pursuant to the Attorney General's direction in every district court proceeding in which a constitutional challenge to the Sentencing Guidelines was made. We rely on model briefs here for description of the parties' arguments for several reasons. First, although the arguments underwent some refinement during the year, the essentials of each party's position were clearly articulated in model briefs. Second, examining the actual briefs from each of the hundreds of cases would not only be costly and inordinately time-consuming (even if the briefs could be retrieved from federal records centers), but would add almost no relevant information because most briefs followed the format of a model brief with the exception of individually-tailored statements of the nature of the charge and procedural history of particular defendant's cases. (In addition, the DOJ Model Brief was supplemented, as appropriate, with argument addressing the "core" function nondelegation issue. See infra note 100.) Third, the written district court opinions, which were reviewed for this study, confirm the continued adherence of the parties to the positions articulated in the model briefs.
123. See id. at 4-5, 40-50; see also, e.g., United States v. Smith, 686 F. Supp. 1246, 1251-53 (W.D. Tenn. 1988) (holding that Sentencing Commission was actually in executive branch despite recitation in statute that it was part of judicial branch); United States v. Ortega Lopez, 684 F. Supp. 1506, 1516 & n.1 (C.D. Cal. 1988) (en banc) (Hupp, J., dissenting) (accepting Department's position that judicial branch "label" is not controlling).
124. See Motion of the United States Sentencing Commission for Leave to File a Brief as Amicus Curiae at 2, United States v. Allen, Cr. No. S-88-024-EJG (E.D. Cal. 1988) ("The Department of Justice has consented to the Commission's request that it be allowed to file a brief as amicus curiae in this and other cases involving a challenge to the Sentencing Guidelines."). The Sentencing Commission also developed a model brief that it filed as amicus curiae in district court proceedings. See infra note 91.
125. See Brief for the United States Sentencing Commission as Amicus Curiae in Support of the Constitutionality of the Sentencing Guidelines, United States Sentencing Commission, at 1-3, 25-27, 35-48 (Feb. 8, 1988) [hereinafter Sentencing Commission Model Brief] (on file with the New York University Law Review); Fried, supra note 82, at 165 ("The Chairman and the most influential members of the Commission were all judges, and they resented any suggestion that they were now part of the Executive Branch as well as any aspersions cast on their independence."); see also, e.g., United States v. Macias-Pedroza, 694 F. Supp. 1406, 1413-16 (D. Ariz. 1988) (en banc) (holding that Commission is properly regarded as being located in judicial branch because it operates in aid of judicial function); United States v. Ruiz-Villanueva, 680 F. Supp. 1411, 1421-22 (S.D. Cal. 1988) (same).
126. Fried, supra note 82, at 166; see also Marcus, supra note 83, at 32 (describing different approaches of Department of Justice and Sentencing Commission in defending Guidelines in district courts).
127. See, e.g., United States v. Wilson, 686 F. Supp. 284, 285-86 (W.D. Okla. 1988) (accepting defendant's argument that mandatory service of three judges on Commission impairs their ability to perform their constitutional judicial role and undermines their impartiality and independence); United States v. Elliott, 684 F. Supp. 1535, 1539-40 (D. Colo. 1988) (accepting defendant's argument that mandatory participation of three Article III judges on Sentencing Commission contravenes separation of powers doctrine).
128. See, e.g., United States v. Serpa, 688 F. Supp. 1398, 1401 (D. Neb. 1988) (accepting defendant's argument that president's removal power over commissioners, including judge members, violates separation of powers doctrine); United States v. Cortes, 697 F. Supp. 1305, 1308-11 (S.D.N.Y. 1988) (same).
129. See DOJ Model Brief, supra note 88, at 18-19, 24-25; Sentencing Commission Model Brief, supra note 91, at 6, 48-55; see also, e.g., United States v. Franz, 693 F. Supp. 687, 692 (N.D. Ill. 1988) (holding that presence of Article III judges on Sentencing Commission is not unconstitutional because judges serve voluntarily); United States v. Alves, 688 F. Supp. 70, 76-78 (D. Mass. 1988) (holding that service of Article III judges as voluntary participants on Sentencing Commission does not impair judicial function).
130. See DOJ Model Brief, supra note 88, at 19-24.
131. See id. at 17-19; Sentencing Commission Model Brief, supra note 91, at 5, 63-64; see also, e.g., United States v. Weidner, 692 F. Supp. 968, 985-87 (N.D. Ind. 1988) (holding that president's power to remove commissioners does not impair independence of judiciary); United States v. Landers, 690 F. Supp. 615, 623 (W.D. Tenn. 1988) (same).
132. See, e.g., United States v. Dahlin, 701 F. Supp. 148, 150-51 (N.D. Ill. 1988) (holding that decisions involving policy choices on core fundamental liberties questions, such as defining and punishing criminal conduct, are "nondelegable" and may be exercised only by Congress); United States v. Williams, 691 F. Supp. 36, 41-53 (M.D. Tenn. 1988) (en banc) (holding that promulgation of Sentencing Guidelines that are binding on court is nondelegable legislative function because only Congress may fix punishment for engaging in criminal conduct).
133. See, e.g., United States v. Bogle, 689 F. Supp. 1121, 1162 (S.D. Fla. 1988) (en banc) (Aronovitz, J., specially concurring) (arguing that "the delegation of legislative power to the Sentencing Commission is so excessive as to violate the basic premises of our constitutional scheme of government" because "[f]ew meaningful restrictions on [the Commission's] discretion are set forth" in statute); United States v. Brodie, 686 F. Supp. 941, 950 (D.D.C. 1988) (stating that Sentencing Reform Act "would probably fail to pass muster [under delegation doctrine], for Congress has given to the Sentencing Commission a mandate of such vagueness that it constitutes no real direction at all").
134. Although the DOJ Model Brief did not address the "core" function nondelegation argument, government briefs filed in the district courts added that argument, and the government's response is well-stated in the briefs submitted to appellate courts during this same time period. See Brief for the United States at 60-62, United States v. Frank, 864 F.2d 992, 1010-12 (3d Cir. 1988) (No. 88-3220); Sentencing Commission Model Brief, supra note 91, at 27-30; see also, e.g., United States v. Myers, 687 F. Supp. 1403, 1409 (N.D. Cal. 1988) (rejecting argument that any legislative function is per se nondelegable as "core function"); United States v. Sparks, 687 F. Supp. 1145, 1150-51 (E.D. Mich. 1988) (same).
135. See DOJ Model Brief, supra note 88, at 50-56; Sentencing Commission Model Brief, supra note 91, at 31-35; see also, e.g., United States v. Macias-Pedroza, 694 F. Supp. 1406, 1411-13 (D. Ariz. 1988) (en banc) (holding that Sentencing Reform Act set out intelligible standards and statements of purpose and thus was constitutional delegation of legislative power to Sentencing Commission); United States v. Richardson, 685 F. Supp. 111, 113-14 (E.D.N.C. 1988) (same).
136. See, e.g., United States v. Brittman, 687 F. Supp. 1329, 1354-57 (E.D. Ark. 1988) (holding that Guidelines violate defendant's due process right to individualized sentencing); United States v. Frank, 682 F. Supp. 815, 817-19 (W.D. Pa. 1988) (same).
137. See, e.g., United States v. Alafriz, 690 F. Supp. 1303, 1310 (S.D.N.Y. 1988) (invalidating Guidelines on due process grounds, saying that "sentencing requires a specific assessment of an individual and an individual's circumstances, as it involves an individual's constitutionally protected right to liberty"); United States v. Martinez-Ortega, 684 F. Supp. 634, 636 (D. Idaho 1988) (holding that "the negation of the sentencing judge's discretion violates the due process clause by preventing the defendant from having an opportunity to convince the sentencing judge that there are circumstances which override the point allocations of the Guidelines").
138. See Brief for the United States at 67-75, United States v. Frank, 864 F.2d 992 (3d Cir. 1988) (No. 88-3220); see also, e.g., United States v. Weidner, 692 F. Supp. 968, 971-72 (N.D. Ind. 1988) (rejecting due process right to individualized sentencing); United States v. Seluk, 691 F. Supp. 525, 539 (D. Mass. 1988) (holding that congressional choice to reduce exercise of discretion by judges was not offensive to due process standards).
139. The district court level Sentencing Guidelines litigation continued from January 25, 1988 (the first district court ruling) to December 19, 1988 (the last district court ruling).
140. Compare United States v. Ruiz-Villanueva, 680 F. Supp. 1411 (S.D: Cal. 1988) (upholding Guidelines), with United States v. Lopez-Barron, 685 F. Supp. 725 (S.D. Cal. 1988), and United States v. Arnold, 678 F. Supp. 1463 (S.D. Cal. 1988) (both invalidating Guidelines); compare United States v. Knox, 694 F. Supp. 777 (W.D. Wash. 1988) and United States v. Amesquita-Padilla, 691 F. Supp. 277 (W.D. Wash. 1988) (both upholding Guidelines), with United States v. Wylie, No. CR88-04T (W.D. Wash. Mar. 29, 1988), United States v. Hines, No. CR88-004WD (W.D. Wash. May 11, 1988), United States v. Dalton, No. CR88-2C (W.D. Wash. June 16, 1988), and United States v. Nordall, No. CR87-067TB (W.D. Wash. April 21, 1988) (all invalidating Guidelines). Based upon our research, judges divided internally over the validity of the Guidelines in 28 district courts.
141. See infra Table 3.
142. See United States v. Ortega Lopez, 684 F. Supp. 1506, 1520 & n.12 (C.D. Cal. 1988) (en banc) (Hupp, J., dissenting).
143. 488 U.S. 361 (1989).
144. Id. at 371 (explaining that Supreme Court had granted certiorari before judgment "[b]ecause of the 'imperative public importance' of the issue" and "because of the disarray among the Federal District Courts"). The Mistretta case was selected as the vehicle for a certiorari petition because it was one of the first district rulings that had been perfected for appeal, because it raised the central separation of powers and nondelegation doctrine issues, and because both the Department of Justice and Mistretta's counsel agreed to seek Supreme Court review. As the coauthor of this study who was involved in the litigation of these cases reports, the Solicitor General's office had initially identified another case as the candidate to bring this matter to the Supreme Court's attention, but the defendant's counsel in that case proved uncooperative. Solicitor General Charles Fried, on behalf of the United States, concluded that it would both expedite resolution of this pressing matter and enhance the likelihood that the Court would accept review of a case off the ordinary track through the court of appeals if both sides joined in petitioning for review.
145. Under 28 U.S.C. § 1254(1) (1994), the Supreme Court may grant a petition for a writ of certiorari to review any case that is "in" the court of appeals, even if a final judgment has not yet been entered by that court. See United States v. Nixon, 418 U.S. 683 690-92 (1974). If a notice of appeal has been filed from the district court, and it is properly docketed in the court of appeals, then a party may seek direct review by the Supreme Court before judgment in the court of appeals. See id. The Supreme Court grants "certiorari before judgment" about once a decade and only in cases of pressing importance See, e.g., Dames & Moore v. Regan, 453 U.S. 654, 667-68 (1981) (Iran hostage agreement); Nixon, 418 U.S. at 691-92 (subpoena to president); Youngstown Sheet & Tube Co. v. Sawyer, 343 U.S. 579 (1932) (steel seizure case); Ex parte Quirin, 317 U.S. 1 (1942) (president's assignment of jurisdiction over trial of belligerent saboteurs to military tribunal).
146. See Mistretta, 488 U.S. at 371 n.6 (1989) (observing that two courts of appeals had addressed issue since it had granted certiorari, one invalidating Guidelines and another upholding them).
147. See Gubiensio-Ortiz v. Kanahele, 857 F.2d 1245 (9th Cir. 1988), vacated sub nom. United States v. Chavez-Sanchez, 488 U.S. 1036 (1989).
148. See id. at 1251-66.
149. See United States v. Frank, 864 F.2d 992 (3d Cir. 1988). 116 See id. at 1008-10.
150. See id. at 1010-12.
151. See id. at 1012-16.
152. 488 U.S. 361, 311-19 (1989). For relatively contemporaneous scholarly comment on Mistretta, see Stephen L. Carter, Constitutional Improprieties: Reflections on Mistretta, Morrison, and Administrative Government, 57 U. Chi. L. Rev. 357 (1990);
153. Martin H. Redish, Separation of Powers, Judicial Authority, and the Scope of Article III: The Troubling Cases of Morrison and Mistretta, 39 DePaul L. Rev. 299, 311-19 (1989);
154. Julia L. Black, Note, The Constitutionality of Federal Sentences Imposed Under the Sentencing Reform Act of 1984 After Mistretta v. United States, 75 Iowa L. Rev. 767 (1990);
155. Mary Buffington, Comment, Separation of Powers and the Independent Governmental Entity After Mistretta v. United States, 50 La. L. Rev. 117 (1989);
156. Mark Nielsen, Comment, Mistretta v. United States and the Eroding Separation of Powers, 12 Harv. J.L. & Pub. Pol'y 1049 (1989);
157. Kristin L. Timm, Note, "The Judge Would Then Be the Legislator": Dismantling Separation of Powers in the Name of Sentencing Reform - Mistretta v. United States, 65 Wash. L. Rev. 249 (1990);
158. The Supreme Court, 1988 Term-Leading Cases, supra note 87, at 279-90.
159. See Mistretta, 488 U.S. at 372-97.
160. See id. at 372-79.
161. Id. at 381.
162. See id. at 380-411.
163. Id. at 395.
164. Id. at 388.
165. See id. at 390-91.
166. Id. at 395.
167. See id. at 397-404 (citing simultaneous service of first Chief Justice, John Jay, as Ambassador to England; of Chief Justice Ellsworth as minister to France; and of Chief Justice Marshall as Secretary of State; service of five justices on election commission that resolved disputed election of 1876; service of Justice Jackson as prosecutor at Nuremberg trials; presiding of Chief Justice Warren over commission investigating assassination of President Kennedy).
168. Id. at 404.
169. See id. at 407-08.
170. Id. at 412.
171. See id. at 409-11.
172. See id.
173. See id. at 413-27 (Scalia, J., dissenting).
174. Id. at 413.
175. See id. at 413, 422-27.
176. Id. at 425.
177. Id. at 420.
178. Id.
179. Id. at 427.
180. John Steer, the General Counsel of the Sentencing Commission, generously granted us access to this list, subject to an agreement not to publicly release the names of those ruling judges who had not issued written decisions. A caveat should be noted concerning the Sentencing Commission information. When the Commission staff conducted the supplemental telephone survey, they focused primarily upon district judges who were not following the Guidelines. Thus, when an individual district judge was not applying the Guidelines, the staff recorded information about any case decision by that judge invalidating the Guidelines on constitutional grounds. But when an individual district judge was applying the Guidelines, no additional information was solicited. In substantial part, this reflects the fact that many judges applying the Guidelines did so as a matter of course and without a formal decision upholding the constitutionality of the Guidelines. Thus, the fact that a judge followed the Guidelines in sentencing did not necessarily mean that the judge had ruled favorably or at all upon a constitutional challenge to the Guidelines. By contrast, nearly all judges who declined to apply the Guidelines in sentencing did so by reason of a constitutional objection. (The Sentencing Commission staff did note a few judges who had made no formal ruling on the constitutionality question but were not applying the Guidelines to sentences. These judges have not been included within this study, which is limited to study of constitutionality rulings.) In sum, the supplemental information obtained from the telephone survey provides rather complete information about a few additional district judges who had invalidated the Guidelines, which information is included in the database for this study. However, a judge who was recorded in the supplemental survey as applying the Guidelines is included in our database only if there is an independent confirmation (either from the primary Sentencing Commission decision list or written decisins available on Westlaw or Lexis) that an actual decision addressing the constitutionality of the Guidelines had been rendered. For this reason, the database prepared for this study should be complete (or as complete as humanly possible) with respect to decisions by district judges invalidating the Guidelines. However, the database may fall short of perfect completeness with respect to decisions by district judges upholding the Guidelines. Nevertheless, the Sentencing Commission staffs telephone survey added only 18 judges to the list of judges invalidating the Guidelines, and we therefore assume that a similarly small number of additional judges would have been added to the list of judges upholding the Guidelines had sufficient information been available to designate them. In any event, at this point in time, it would be impossible to replicate the original sources of information and create a more complete database.
181. The Westlaw and Lexis electronic databases were searched for all rulings on the constitutionality of the Sentencing Guidelines. These databases primarily contain decisions that were officially reported in the Federal Supplement, although quite a few unpublished decisions are also included.
182. For purposes of our study, three judges were eliminated from the database: (1) a district judge from the Virgin Islands for whom crucial data were unavailable or nonexistent given the anomalous locale; (2) a district judge who was included in the Sentencing Commission's tally as upholding the constitutionality of the Guidelines but who, as we discovered upon review of his written opinion, had applied the Guidelines without addressing the constitutionality question; and (3) a court of appeals judge who ruled on a constitutional challenge to the Guidelines while sitting as a district court judge by designation. See United States v. Estrada, 680 F. Supp. 1312 (D. Minn. 1988) (Heany, J.), rev'd, 873 F.2d 1449 (8th Cir. 1989) (unpublished table decision). Only six court of appeals judges ruled upon the constitutionality of the Guidelines, including Judge Heaney, the members of the panels in the Third Circuit (one of whom dissented on other grounds without addressing the constitutionality issue), and the Ninth Circuit. See United States v. Frank, 864 F.2d 992 (3d Cir. 1988); Gubiensio-Ortiz v. Kanahele, 857 F.2d 1245 (9th Cir. 1988), vacated sub nom. United States v. Chavez-Sanchez, 488 U.S. 1036 (1989). Accordingly, court of appeals judges were excluded from our study. Moreover, many of the variables included in our study are relevant only with respect to district court judges.
183. We were unable to retrieve opinions for only two judges who we had at least some indication may have prepared written opinions. These two judges are necessarily treated in our study as though they issued unwritten decisions.
184. Upon reviewing the written decisions, we determined that a few decisions included in the initial set from the Sentencing Commission were so cursory and lacking in analysis as to amount to nothing more than written confirmations of bench announcements. Because these documents provided no basis for analyzing the judge's reasoning, we removed them from the opinion content analysis stage of the study. However, if an opinion provided any statement of reasons, even if limited to the adoption by reference of another judge's opinion, the opinion was included in our content analysis at least for some purposes. A few judges issued more than one written decision, but with only one exception these multiple decisions were identically-worded opinions rendered in different cases within a few days of one another. On the view that the more important decision is the one upon which the judge initially "took a stand" for or against the Guidelines, the first opinion is used for dating purposes. For the one judge who issued a second opinion supplementing the substance of the first, both opinions were reviewed together as divulging the judge's overall reasoning and treatment of the panoply of constitutional issues. Judge Edward Devitt of the District of Minnesota initially upheld the Sentencing Commission and the Guidelines on July 27, 1988, ruling that the Commission was properly regarded as an independent commission in the judicial branch and holding that the presidential removal power did not intrude unconstitutionally upon the judiciary. See United States v. Whitfield, 689 F. Supp. 954, 956-57 (D. Minn. 1988). Two months later, on September 13, 1988, Judge Devitt reaffirmed his earlier ruling and supplemented his analysis, primarily by addressing (and rejecting) the government's argument that the Commission should be conceived as an executive branch agency. See United States v. Roy, 694 F. Supp. 635, 637-41 (D. Minn. 1988).
185. For a discussion of the role of en banc district courts in the Sentencing Guidelines Crisis, see infra note 172.
186. Questions of sampling and potential bias are avoided in this study because (with the qualifications mentioned in notes 141-45) we have evaluated the entire universe of judicial decisions on the constitutionality of the Sentencing Guidelines in 1988. While the 293 judges involved do not constitute the entire population of district judges, the inclusion of a judge within this universe of decisions was based on the presumably random factors of a particular criminal case being assigned to that judge and the defendant raising a challenge to the Sentencing Guidelines.
187. See United States v. Thomas, 699 F. Supp. 147, 148 (W.D. Tenn. 1988) ("The same arguments are presented to each court, resulting in decisions for and against the guidelines on varying grounds."). As discussed earlier, the Department of Justice and the Sentencing Commission drafted model briefs to be presented to all district judges addressing challenges to the Sentencing Guidelines. See DOJ Model Brief, supra note 88; Sentencing Commission Model Brief, supra note 91. In addition, members of a team of attorneys from Main Justice in Washington, D.C., including one of the authors of this article specially appeared to argue on behalf of the government in many of the cases in the district courts. Although the criminal defense bar did not present a single model brief, the cooperative efforts of two organizations ensured a fair degree of comparability in the legal work presented on behalf of defendants challenging the Guidelines. Public Citizen Litigation Group frequently appeared as counsel for defendants challenging the Guidelines See e.g., United States v. Brodie, 686 F. Supp. 941, 942 n.2 (D.D.C. 1988); United States v. Ruiz-Villanueva, 680 F. Supp. 1411, 1412 (S.D. Cal. 1988); see also Marcus, supra note 83 at 31 (reporting involvement of Public Citizen in challenges to Guidelines by criminal defendants). Alan Morrison of Public Citizen ultimately argued the case against the Guidelines before the Supreme Court in Mistretta v. United States, 488 U.S. 361, 362 (1989). The National Association of Criminal Defense Lawyers routinely submitted amicus curiae briefs attacking the Guidelines in these cases. See, e.g., United States v. Swapp 695 F. Supp. 1140, 1149 (D. Utah 1988); United States v. Macias-Pedroza, 694 F. Supp. 1406, 1414 (D. Ariz. 1988) (en banc); United States v. Weidner, 692 F. Supp. 968, 970 (N.D. Ind. 1988); United States v. Amesquita-Padilla, 691 F. Supp. 277, 279 (W.D. Wash. 1988); United States v. Bogle, 689 F. Supp. 1121, 1124 (S.D. Fla. 1988) (en banc); United States v. Perez, 685 F. Supp. 990, 992 n.1 (W.D. Tex. 1988); see also Cohen, supra note 21, at 184 n.3 (referring to "significant amount of 'sharing' of legal briefs and arguments on both sides of the [Sentencing Guidelines] issue").
188. See Posner, What Do Judges Maximize, supra note 52, at 26 ("[J]udicial outcomes reflect both the judges' preferences going in and the quality of the briefing and argument in particular cases.").
189. For another, likewise unique, example of content analysis of opinions in an empirical study of judicial behavior and judge characteristics (in that case, gender), although involving comparison of a sample of opinions in different cases by appellate judges on panels with shifting membership, see Sue Davis, Do Women Judges Speak "In a Different Voice?": Carol Gilligan, Feminist Legal Theory, and the Ninth Circuit, 8 Wis. Women's L.J. 143 (1992-1993). For an example of empirical study of the content of judicial opinions in another context, that of the evolution of a common law doctrine, see Morriss, supra note 151 See Rowland & Carp, supra note 39, at 172 (encouraging revival of interest by empirical researchers in opinions as "judges' codified explanation of their judgment protocols"); Davis, supra note 150, at 151-52 (explaining that study of hypothesized gender differences in judging "may not be readily quantifiable in terms of voting behavior" and must instead "focus at least as much on judges' methods of reasoning as . . . on the way they vote").
190. Cross, supra note 48, at 267.
191. Edward L. Rubin, The Concept of Law and the New Public Law Scholarship, 89 Mich. L. Rev. 792, 801 (1991).
192. Cross, supra note 48, at 270 (citing Frederick Schauer, Giving Reasons, 47 Stan. L. Rev. 633, 652 (1995));
193. see also Alvin B. Rubin, Does Law Matter? A Judge's Response to the Critical Legal Studies Movement, 37 J. Legal Educ. 307, 311 (1987);
194. Kathleen Waits, Values, Intuitions, and Opinion Writing: The Judicial Process and State Court Jurisdiction, 1983 U. Ill. L. Rev. 917, 931.
195. Some might question our assumption that these written district court decisions reflect the reasoning of the judges themselves. Among federal appellate judges, informed commentators believe that few judges still write their own opinions. "[T]he tendency has been for more and more of the initial opinion-drafting responsibility to be delegated to law clerks, transforming the judge from a draftsman to an editor." Richard A. Posner, The Federal Courts 141 (1996) [hereinafter Posner, Federal Courts]. However, in contrast to court of appeals judges who generally are obliged to render written decisions in appeals, district court judges produces a written ruling on a particular matter only when they so choose to record their opinions. Thus, we believe it is more likely that a district judge takes a personal interest and a hand in drafting on the special occasions in which an opinion is produced. Indeed, we believe this is particularly likely when the issue is one of great significance - such as the Sentencing Guidelines constitutionality issue - and the judge wishes to publicly participate in the legal debate. In any event, even when a judge delegates drafting responsibilities to a law clerk, the judge maintains control over the basic structure of the opinion and the grounds upon which the issue is resolved. As discussed below, see infra Part IV.B.2, we coded opinions primarily on the basis of the constitutional theories raised and resolved by each judge. Whether the judge acts as author or architect of the opinion, our coding should be adequate to uncover the judge's underlying views.
196. Cross, supra note 48, at 268.
197. Anthony D'Amato, Aspects of Deconstruction: Refuting Indeterminancy with One Bold Thought, 85 Nw. U. L. Rev. 113, 118 (1990). 153 See supra Part II.B.1-2.
198. See, e.g., Raines v. Byrd, 117 S. Ct. 2312 (1997) (dismissing for lack of standing law-suit brought by individual members of Congress challenging constitutionality of Line Item Veto Act on separation of powers grounds); Clinton v. Jones, 520 U.S. 681 (1997) (holding doctrine of separation of powers does not require federal courts to stay all private actions against president while in office); Loving v. United States, 517 U.S. 748 (1996) (holding that separation of powers principles do not preclude Congress from delegating its constitutional authority to president to define aggravating factors that permit imposition of statutory penalty of death in military capital cases).
199. 488 U.S. 361 (1989).
200. For a recent description of the continued controversy over the Sentencing Guidelines, including the views of both critics and defenders of the system, see generally Mary Pat Flaherty & Joan Biskupic, Justice by the Numbers (pts. 1-5), Wash. Post, October 6, 1996, at A1, October 7, 1996, at A1, October 8, 1996, at A1, October 9, 1996, at A1, October 10, 1996, at A1.
201. For scholarly treatment of the subject, see generally Symposium on Federal Sentencing, 66 S. Cal. L. Rev. 99 (1992).
202. Cabranes, Dismal Failure, supra note 80, at 2;
203. see also Michael Tonry, Sentencing Matters 11 (1996) ("Few outside the federal [sentencing] commission would disagree that the federal guidelines have been a disaster.").
204. José A. Cabranes, Letter to the Editor: Incoherent Sentencing Guidelines, Wall St. J., Aug. 28, 1992, at All [hereinafter Cabranes, Letter to Editor];
205. see also Kate Stith & José A. Cabranes, Judging Under the Federal Sentencing Guidelines, 91 Nw. U. L. Rev. 1247, 1254 (1997) (decrying reduction of trial judges' sentencing discretion and substitution of "bureaucratic penalization" under Guidelines regime);
206. Naftali Bendavid, Breyer's Role as Sentencing Pioneer Still Rankles, Legal Times, May 16, 1994, at 7 (quoting district Judge Jack Weinstein as saying that Guidelines "require, in the main, cruel imposition of excessive sentences and unnecessary micromanaging of sentencings").
207. See Don J. DeBenedictis, The Verdict is In, A.B.A. J., Oct. 1993, at 78-79 (finding majority of judges and equal number of federal judges supported "scrapping" the Guidelines);
208. Henry J. Reske, Judges Irked by Tough-on-Crime Laws, A.B.A. J., Oct. 1994, at 18 (finding that more than 50% of district judges supported elimination of Sentencing Guidelines); see also Bendavid, supra note 163, at 7 (describing opposition of federal judges to Guidelines, even to point that "[m]any judges regard as a traitor any colleague who serves on the U.S. Sentencing Commission").
209. One federal judge resigned from the bench in protest of the Sentencing Guidelines, see Criticizing Sentencing Rules, U.S. Judge Resigns, N.Y. Times, Sept. 30, 1990, at A22, and two federal senior judges declared they would refuse to impose sentences in drug cases to avoid the Guidelines and mandatory minimum sentences, see DeBenedictis, supra, at 78.
210. See Bowman, Quality of Mercy, supra note 72, at 680, 707-20, 748-49 (rebutting criticisms that Guidelines are too rigid);
211. see also Theresa Walker Karle & Thomas Sager, Are the Federal Sentencing Guidelines Meeting Congressional Goals? An Empirical and Case Law Analysis, 40 Emory L.J. 393, 444 (1991) (concluding, after statistical study of Guidelines sentences, "that the Guidelines generally are meeting their goal of reducing disparity," although "disparity persists on a more limited scale as a result of differing approaches to ambiguous Guidelines provisions");
212. Andrew J. Kleinfeld, The Sentencing Guidelines Promote Truth and Justice, 55 Fed. Probation 16 (1991) (supporting, as a federal court of appeals judge, the federal Sentencing Guidelines system);
213. Julie R. O'Sullivan, In Defense of the U.S. Sentencing Guidelines' Modified Real-Offense System, 91 Nw. U. L. Rev. 1342, 1343-44 (1997) (stating that, although Guidelines have produced structure that, "judging from the scholarly commentary, virtually everyone loves to hate," the system is "a significant advance in sentencing practice [that] is workable in the day-to-day administration of justice");
214. Thomas N. Whiteside, The Reality of Federal Sentencing: Beyond the Criticism, 91 Nw. U. L. Rev. 1574, 1581, 1598 (1997) ("The [Guidelines] system is rigid enough to demand compliance with congressional statutory mandates, complex enough to capture culpability in a multitude of criminal behaviors in consistent fashion, and flexible enough to consider significant individual characteristics when appropriate.");
215. cf. Paul H. Robinson, The Federal Sentencing Guidelines: Ten Years Later, An Introduction and Comments, 91 Nw. U. L. Rev. 1231, 1235, 1241 (1997) (concluding that Sentencing Reform Act establishes "appropriate balance" between goal of consistency through system of Guidelines that all judges must work from and power of judges to depart from Guidelines in cases differing from the paradigm envisioned by applicable Guidelines, although criticizing Guidelines as promulgated because "the Commission did not deduce sentencing guidelines from principles of justice, or logical principles of any sort," but rather "based its sentences on mathematical averages of past practice of federal sentencing judges").
216. As recently as 1994, a panel of the United States Court of Appeals for the Eleventh Circuit in an unpublished opinion questioned the dual role of one district judge as both a member of the Sentencing Commission and a sentencing judge. See Ira Bloom, The Aftermath of Mistretta: The Demonstrated Incompatibility of the United States Sentencing Commission and Separation of Powers Principles, 24 Am. J. Crim. L. 1, 2-3, 9-11 (1996) (discussing United States v. McLellan, 28 F.3d 117 (11th Cir. 1994) (unpublished table decision));
217. Ann Woolner, Sentencing Panel Hit Over Dual Role, Legal Times, March 20, 1995, at 2 (same). The panel hinted that the constitutional question was not closed, saying that the Supreme Court's Mistretta decision '"was decided . . . before the Commission's practices and powers had been developed much.'" Woolner, supra, at 2 (quoting McLellan) (ellipsis in original). In a subsequent published opinion, the Eleventh Circuit "emphatically disavow[ed]" adoption of the earlier panel's suggestion of a general conflict for a judicial member of the Sentencing Commission, leaving the issue of recusal from sentencing duties for future resolution. In re United States, 60 F.3d 729, 731 n.2 (11th Cir. 1995); see also Ann Woolner, 11th Circuit Rebuffs Sentencing Challenge, Legal Times, July 31, 1995, at 8 (reporting on Eleventh Circuit's ruling). With respect to the constitutionality of the Commission, the court stated that "[t]he Mistretta decision sweeps broadly, and if it is to be overruled, the Supreme Court itself will have to do it." In re United States, 60 F.3d at 733 n.4. In 1993, a district judge renewed the substantive due process line of attack upon the Sentencing Guidelines - which the Supreme Court had not addressed in its Mistretta opinion - by holding that the Guidelines unconstitutionally imposed a 30-year prison sentence on a drug dealer without allowing the judge to take into account the minor nature of prior convictions or extenuating circumstances. See United States v. Spencer, 817 F. Supp. 176, 181-84 (D.D.C. 1993), aff'd on other grounds, 25 F.3d 1105 (D.C. Cir. 1994); see also United States v. Silverman, 976 F.2d 1502, 1534-35 (6th Cir. 1992) (Martin, J., dissenting) ("I now . . . see the error of my early endorsement of the goals of the guidelines . . . . I now believe I was wrong in endorsing the guidelines [because they] have become more than just guidelines; they are rigid mandates [that] disregard fundamental notions of due process.").
218. For recent scholarly critiques of the Sentencing Guidelines on constitutional grounds - separation of powers and due process, see generally Bloom, supra note 166; Ronald J. Krotoszynski, Jr., On the Danger of Wearing Two Hats: Mistretta and Morrison Revisited, 38 Wm. & Mary L. Rev. 417 (1997);
219. Jack H. McCall, Jr., The Emperor's New Clothes: Due Process Considerations Under the Federal Sentencing Guidelines, 60 Tenn. L. Rev. 467 (1993).
220. See Rowland & Carp, supra note 39, at 145.
221. See Ashenfelter, Eisenberg & Schwab, supra note 27, at 263-64 ("On many issues . . . , district judge discretion governs . . . [and] in their isolation, district judges may feel less constrained by legal doctrine and thus be more likely to follow political inclinations . . . ."); see also Rowland & Carp, supra note 39, at vii (stating that political scientists have recognized that "federal trial courts [are] highly politicized policy-making institutions and . . . trial judges, like their appellate counterparts, [are] influenced by their personal values and policy preferences").
222. Howard, supra note 30, at 135.
223. See id. at 178-221; Ashenfelter, Eisenberg & Schwab, supra note 27, at 263-64 ("Appellate judges decide cases in panels of three or more. Multiple decision makers may check judicial discretion. If one's fellow judges evaluate the legal arguments a certain way, concern for collegiality and respect pushes one to conform to these views . . . .").
224. See Daniel J. Freed, Federal Sentencing in the Wake of the Guidelines: Unacceptable Limitations on the Discretion of Sentencers, 101 Yale L.J. 1681, 1719 (1992) (describing how, during 1988, "[s]ome district courts held unprecedented en banc hearings to receive arguments from the Department of Justice and the Commission . . . and from defense attorneys"). In our study, 90 of the 293 judges participated in en banc decisions. The en banc proceedings by district court judges were themselves unusual, interesting, and arguably unauthorized. While the efficiency of a single proceeding is clear, the validity of a district judge's compliance with an en banc order with which he disagrees is not. Some of these purported en banc proceedings were nothing more than joint hearings of argument in which the participating judges reserved the right to determine whether to join the en banc opinion and follow that ruling in future cases. However, in some districts, the judges agreed in advance to adopt the ruling of the majority on the constitutionality of the Guidelines, even if one or more judges dissented. See, e.g., United States v. Ortega Lopez, 684 F. Supp. 1506, 1515 (C.D. Cal. 1988) (en banc) ("This decision is binding upon the members of this Court in all relevant cases unless and until we receive a contrary ruling from the Ninth Circuit or the Supreme Court."). Since there is no general statutory authorization for federal trial judges to reach decisions as panels and apply those decisions as mandatory precedent for all judges in the district, the validity and merit of these en banc proceedings warrant further examination.
225. See, e.g., United States v. Macias-Pedroza, 694 F. Supp. 1406, 1419 (D. Ariz. 1988) (en banc) (Carroll, J., concurring in part); id. at 1419 (Rosenblatt, J., dissenting); United States v. Bogle, 689 F. Supp. 1121, 1161 (S.D. Fla. 1988) (en banc) (Aronovitz, J., specially concurring); id. at 1163 (Gonzalez, J., dissenting); id. at 1164 (Nesbitt, J., dissenting); Ortega Lopez, 684 F. Supp. at 1515 (Hupp, J., dissenting).
226. This part of the Article describes how we translated the information on each independent variable into a format suitable for statistical analysis. Discussion of the theoretical basis for study of the particular variable, relevant prior research findings on that variable or type of variable, and an interpretation of the findings in this study are provided in Part V of this Article.
227. Sources checked include various editions of: Almanac of the Federal Judiciary; The American Bench; Who's Who in American Law.
228. In particular, Sheldon Goldman generously shared information on each judge's racial background and American Bar Association rating.
229. 28 minority judges were included in our dataset, 10 appointed by Republican presidents and 18 appointed by Democratic presidents.
230. See Federal Bureau of Investigation, U.S. Dep't of Justice, Crime in the United States: Uniform Crime Reports for the United States 1988, at 1-6 (1989). Although Congress directed federal law enforcement agencies to participate in this program, implementation at the federal level had not yet occurred in 1988, the subject date for this study. See id. at 6.
231. We had initially hoped to use a crime index for each of the nation's metropolitan statistical areas, believing this would be the best measure of the level of criminal activity in the area surrounding each judge's courthouse. (The crime index for metropolitan statistical areas is reported in Appendix IV of the 1989 report. See id. at 324-50.) Unfortunately, data were missing for a large number of the judges because no crime rates were included for Illinois cities as its reporting methods did not comport with FBI standards, see id. at 350 n.4, no report was made for Florida, see id. at 350 n.5, and several judges did not maintain their chambers in metropolitan statistical areas for which crime rates are calculated. Crime rate data by metropolitan statistical area were missing for 61 of the 291 judges in the largest regression analysis, which we regarded as an unacceptable number of missing cases. However, we did prepare an alternative regression analysis using the crime rate by metropolitan statistical area for the 241 judges for whom data was available. This analysis produced no significant findings with respect to crime rate influence. The documentation for this alternative analysis is available from the authors.
232. The crime index by state is reported in Table 5 of the 1989 FBI report. See id. at 60-70. Although the Illinois report did not comply with FBI guidelines and no report was made by Florida, the FBI provided estimated crime counts at the state level. See id. at 70 nn.5-6.
233. See id. at 320.
234. Judges were coded according to the law school from which they received their first legal degree, thus we did not consider advanced legal degrees. Our theory is that the primary socializing effect of legal education is felt during the basic three-year experience leading to the initial legal degree.
235. See, e.g., James Lindgren & Daniel Seltzer, The Most Prolific Law Professors and Faculties, 71 Chi.-Kent L. Rev. 781, 793-94 (1996).
236. See, e.g., The Top 25 Law Schools, U.S. News & World Report, Mar. 1996, at 82.
237. See, e.g., Jack Gourman, The Gourman Report: A Rating of Graduate and Professional Programs in American and International Universities 90-91 (8th ed. 1997).
238. We adopted the following formula: Using the three most recent rankings from each of the three sources, we set up two sets for scoring, one based on position in the top five and one based on position in the top ten. Then, we assigned one point to each law school for each time it is listed in the top five or the top ten of these rankings. Accordingly, we created a total of nine lists (three for each year, thus making 9 points the maximum score), and each source was given equal weight. Based upon listings among the top five law schools, we found that four schools stood out: Chicago (9), Yale (9), Harvard (8), and Stanford (5). See also Ashenfelter, Eisenberg & Schwab, supra note 27, at 274 n.44 (classifying these four as elite law schools for empirical study of judicial decisionmaking). Looking at listings among the top ten law schools, we added Columbia (9) and Michigan (7) to the list; both schools also had ranked just after Stanford in the top five listings. Two very recent entries into the surveys of law schools, both based upon scholarly productivity and impact, confirm our choice of elite law schools. First, one professor has assembled a ranking of law schools that emphasizes the scholarly distinction of a school's faculty, measured by the volume of publication and citations to publications. See David E. Rovella, A Survey of Surveys Ranks the Top U.S. Law Schools, Nat'l L.J., June 2, 1997, at Al (discussing the "Ranking of U.S. Law Schools by Educational Quality" produced by Professor Brian Leiter of University of Texas School of Law). Of the seven law schools designated as "elite" for our study, six fall within this new survey's top six law schools, with our seventh "elite" school, Michigan, being ranked tenth in the new survey. See id. Second, Professors Theodore Eisenberg and Martin T. Wells measured the academic reputations of 32 law schools by citations to their faculties' works. See generally Theodore Eisenberg & Martin T. Wells, Ranking and Explaining the Scholarly Impact of Law Schools, 27 J. Legal Stud. 373 (1998). Their study identified Yale, Chicago, Harvard, and Stanford as standing alone at the top of the ranking for scholarly impact. See id. at 374. In developing a list of elite law schools for federal judges, one would ideally use a historical classification, that is, the ranking of the school at the time that a particular judge attended law school and began his or her legal socialization. Unfortunately, such an effort to create historical rankings would likely prove impossible and would certainly be unduly burdensome, both because of the difficulty of remeasuring the ranking, judge-by-judge, year-by-year, and because the sources we relied upon for law school rankings did not exist during those periods. Accordingly, we began with current rankings (1995-97) to create our list. Because status as an elite law school depends primarily upon reputation, a characteristic that develops over substantial periods of time and changes slowly, these rankings should be roughly accurate. Moreover, as noted infra note 187 and the accompanying text, we supplemented and confirmed this process with a separate ranking based on the number of federal judges who graduated from a particular law school, thus distinguishing "national" from "regional" law schools. This latter ranking should correspond with historical reputation at the relevant point in time.
239. Virginia (11) was second only to Harvard (25) in the number of out-of-state graduates represented among the federal judges in our sample, and ranked well above Michigan (7), Yale (6), Columbia (3), Stanford (3), and Chicago (2).
240. We included in this category those who had prosecution experience at the state or federal level, those who acted as prosecutors in the Judge Advocate General (JAG) Corps of a military service branch, and those who had served as agents of the Federal Bureau of Investigation.
241. For example, while the listing of a judge's previous position as an Assistant United States Attorney (AUSA) likely reflects criminal prosecution experience, some AUSAs handle civil litigation against the federal government. Based on the responses we obtained to our survey, we determined that this was the one position that we could safely treat as a proxy for criminal prosecution; the few judges who had been AUSAs in a noncriminal capacity were so designated in biographies.
242. Of the 34 surveyed judges, we received responses from 30, a return rate of nearly 90%. The surveys were sent to all judges whose biographies listed service as an AUSA without additional information confirming criminal prosecution (as to whom the returned surveys uniformly confirmed prosecutorial duties), service in the military JAG Corps, or municipal or county official positions whose description was unclear. With the exception of AUSAs, from whom the survey response was uniform, thus justifying treatment of that service as a proxy for prosecution, judges who did not respond to the survey or, in the case of one judge who was deceased, were coded as nonprosecutors.
243. Of the 293 judges in our study, 39 are coded as having criminal defense experience. See infra Table 3.
244. We also created alternative variables ELECTED (consisting of state and federal elected positions), APPOINTED (consisting of high-level appointed administrative positions in state and federal government), EXECUTIVE (consisting of elected or appointed high-level administrative positions in both state and federal government), and LEGISLATIVE (consisting of members of Congress or a state legislature). For these alternative variables, judges who had experience at both the executive and legislative level were coded for both EXECUTIVE and LEGISLATIVE, and judges who had held both ELECTED and APPOINTED office were coded for both.
245. See Aliotta, supra note 39, at 279 (describing similar variable for study of judicial decisionmaking). For determining which state offices were elective (other than service as legislator, which was presumed elective), we referred to Table 2.9 in Council of State Governments, 31 The Book of the States 33-34 (1996-97).
246. Council of State Governments, supra note 193, at 35-39 (Table 2.10).
247. For a list of leading administrative and independent agency positions and the type of appointment, see generally Senate Comm. on Govt. Affairs, 102d Cong., 2d Sess., Policy and Supporting Positions (Comm. Print 1992). For information on type of appointment for the defunct Civil Aeronautics Board, we referred to House Comm. on Post Office and Civil Service, 94th Cong., 2d Sess., Policy and Supporting Positions (Comm. Print 1976).
248. This category thus includes state, county, and municipal judgeships, but not federal adjunct judicial positions such as bankruptcy judge or magistrate.
249. For a description of the Standing Committee and its rating of potential nominees for federal judgeships, see Public Citizen v. United States Dep't of Justice, 491 U.S. 440, 443-45 (1989).
250. The Standing Committee has since modified its rating of prospective nominees to eliminate the "Exceptionally Well Qualified" rating, thus leaving three possible ratings, "Well Qualified," "Qualified," or "Not Qualified." See American Bar Association, The ABA Standing Committee on Federal Judiciary: What It Is and How It Works 7 (1991) (explaining three possible ratings of prospective judicial nominees).
251. See, e.g., American Bar Association, Standing Committee on Federal Judiciary: What It Is and How It Works 4 (1983) (briefly describing each rating);
252. American Bar Association, Standing Committee on Federal Judiciary: What It Is and How It Works 4-5 (1980) (same);
253. American Bar Association, The ABA's Standing Committee on Federal Judiciary: What It Is and How It Works 9-10 (1977) (same).
254. Sheldon Goldman, Reagan's Judicial Legacy: Completing the Puzzle and Summing Up, 72 Judicature 318, 320 (1989) (italics omitted).
255. A "split rating" reflected a division on the committee between any two ratings, and the dissenting minority may have believed the prospective nominee deserved a higher, as opposed to a lower, rating than did the majority. For purposes of this study, we have only recorded split ratings when the Standing Committee divided between "Qualified" and "Not Qualified" (recorded for this study as "Q/NQ"). As Sheldon Goldman explains: "The ABA committee insists that anyone receiving a Qualified rating, even if there is dissent among some members, is fully qualified for the federal bench. Yet there is the suspicion that those receiving this split rating are only marginally qualified." Id. at 320.
256. See Annual Report of the Director of the Administrative Office of the United States Courts (1989) [hereinafter Annual Report of Admin. Office].
257. See id. at 456-57 tbl.X-1; see also Cohen, supra note 21, at 192 n.21 (using same source for measuring caseload in study of Sentencing Guidelines constitutionality decisions).
258. See Annual Report of Admin. Office, supra note 202, at 457 n.1.
259. The source for the appointment date is Volume 700 of the Federal Supplement. This volume, which is dated early in 1989, lists those judges on the bench as of the end of 1988, including the date of appointment for each judge.
260. For nine judges, the Sentencing Commission had information concerning how each judge had ruled upon the Guidelines but had no date for the decisions. Because these did not involve written decisions, there was no way to recover this information. Rather than lose nine judges as missing cases in our regression analyses, we adopted an estimated date of decision, choosing the mid-year date of June 30, 1988 as the default. Because all of the Sentencing Guidelines constitutionality decisions occurred during an eleven-month period during 1988, our estimated date is off, at most, by five-and-a-half months. Moreover, since most of the decisions cluster around the middle of the year, it is likely that the deviation from the estimated June 30 date is smaller. Compared to a mean seniority figure of 113.12 months, see infra Table 2, we believed this estimation was acceptable and preferable to the alternative of excluding nine judges for whom we had otherwise complete data. In any event, as a check, we ran an alternative regression analysis in which these nine judges were excluded, with no effect on the correlation for SENIORITY. In addition, not having participated in written opinions, these nine judges are necessarily excluded from all analyses involving dependent variables, other than the basic outcome variable reported in Table 4.
261. See Cohen, supra note 21, at 188-89.
262. See id. at 192.
263. See id. at 193. Subsequently, in a study of federal district judge behavior in the context of criminal antitrust sanctions, Cohen developed a more sophisticated set of eight different variables to measure promotion potential. See Mark A. Cohen, The Motives of Judges: Empirical Evidence from Antitrust Sentencing, 12 Int'l Rev. L. & Econ. 13, 19, 27-29 (1992). Some of these variables reflect the longitudinal nature of that study, such as the availability of newly created court positions over time, that are not relevant to our fixed-time study. See id. Another variable - the age and length of a subject judge's judicial service, which theoretically may either enhance or decrease a judge's promotion potential or responsiveness to that influence in decisionmaking - did not prove useful in that study. See id. at 23. Cohen also introduced variables taking into account the party affiliation of the judge and the party of the President at the time of a vacancy. See id. at 27-29. We have controlled for political party somewhat differently in our study by conducting alternative regression analyses that select only for judges appointed by Republican or Democratic Presidents. In the final analysis, we adhered to the simpler promotion potential scheme in Cohen's earlier study as better suited for our study and also as allowing more direct comparison of his results with ours.
264. The number of district court positions authorized for a state was taken from Table X-1 of the Annual Report of the Director of the Administrative Office of the United States Courts covering the 1988 period. See Annual Report of Admin. Office, supra note 202, at 456-57.
265. This determination was a two-step process: First, we identified all active circuit judges listed in volume 700 of West's Federal Supplement reporter of decisions. Based on the state in which each such active judge had his or her chambers, that circuit court judgeship was treated as reserved to that state. (The assumption is that any vacancy in that judgeship will be filled by a nominee from that state, although this is not invariably the case.) Second, because there were unfilled vacancies in the circuit courts at the end of 1988, the listing of active judges in volume 700 of the Federal Supplement is not a complete listing of all authorized circuit court judgeships and the states from which they are normally appointed. To finalize the list, we also examined each court of appeals vacancy listed in the December 1988 release of "Vacancies in the Federal Judiciary" published by the Administrative Office of the Courts. For each such vacancy listed, we looked at earlier volumes of the
266. For example, determination of the number of circuit judge positions treated as reserved for California changed during 1988. Judge Joseph Sneed, who had his chambers in California, retired in 1987. His successor was Judge Stephen Trott, who was confirmed on March 25, 1988, but who set up his chambers in Idaho. Thus, the Sneed/Trott position was counted as a California position prior to March 25, 1988, but counted as an Idaho position after that date. Thus, for the few California decisions dated prior to March 25, 1988, the ratio was 16/47 (0.34), but after that date was reduced to 15/47 (0.32), reflecting competition over fewer circuit seats reserved for that state.
267. Existence of a vacancy was determined by use of the "Vacancies in the Federal Judiciary" mimeographs from the Administrative Office of the United States Courts from 1988. For each district judge, we looked at the edition of "Vacancies in the Federal Judiciary" dated at the beginning of the month following the month in which the judge's decision on the Guidelines was rendered (i.e., if the decision upholding or invalidating the Guidelines is dated in June, 1988, then we looked to the July 1 edition of "Vacancies in the Federal Judiciary"). If a vacancy appeared on the court of appeals for the circuit in which that district judge sat, then we determined from which state the court of appeals judge who created the vacancy came (that is, we determined whether the vacancy was in a circuit court judgeship that was normally reserved for the state in which the district court judge sits). In addition, to prevent double-counting of the same circuit court judge for both a vacancy and the age of the oldest circuit judge, the judgeship was recorded as vacant only if the circuit judge was listed as retired or declaring an intention to retire by the date of the particular district judge's Sentencing Guidelines decision. If the judgeship was listed as vacant on that date, then that retiring circuit judge was not considered in terms of age of the oldest circuit judge. Thus, there is no overlap between the vacancy factor and the age of the oldest circuit judge factor.
268. Thus, in addition to identifying the existence of a nominee for the position, we noted also the actual date of the nomination and determined whether that nomination predated the date of the district court decision (if the nomination was made after that date, then we recorded this factor as a "Vacancy").
269. To determine this factor, we first determined who were the active court of appeals judges holding circuit court judgeships normally reserved for that district judge's state at the time of the district judge's decision. This required several steps: First, by looking at volume 700 of the Federal Supplement, we identified the active circuit judges with chambers in that state (that is, active judges in circuit court judgeships normally reserved for that state) as of the end of 1988. Second, we also reviewed earlier and subsequent volumes of the Federal Supplement to ensure that no recently retired or retiring judges were omitted and that all active or retiring circuit judges during this period were accounted for under either the age of oldest circuit judge factor or the vacancy factor. Third, we eliminated those judges whose appointments were dated after the date of the district court's decision on the Guidelines. (These judges thus would not have been sitting at the time the district judge made the decision). Fourth, we looked at the "Vacancies in the Federal Judiciary" for the month following the district court decision. If there was a vacancy listed for a court of appeals judgeship normally reserved for that state, we looked at the date the vacancy was created to determine whether the judge would have been an active court of appeals judge at the time of the district court decision. If so, that judge was added to the list Finally, for each of those court of appeals judges who were active at the time of the district court decision and who had chambers in that state, we determined the date of birth We recorded the age of the oldest active circuit judge in a judgeship in that state by a whole number for number of years. For convenience, we simply subtracted the year of birth from the year 1988 (i.e., for a birthdate of 1946, the number to be recorded would be 42).
270. Cohen, supra note 21, at 192.
271. See Jae-On Kim & Charles W. Mueller, Factor Analysis: Statistical Methods and Practical Issues 8 (1978) (describing factor analysis assumption that "observed variables are linear combinations of some underlying (hypothetical or unobervable) factors").
272. See Cohen, supra note 21, at 193.
273. See id. at 190.
274. As noted earlier, see supra note 206, the Sentencing Commission lacked information on the date of decision for nine judges. Rather than lose nine judges as missing cases in our regression analyses, we adopted an estimated date of decision - choosing the mid-year date of June 30, 1988 as the default. Theoretically, this selected date was relevant not only to seniority but also to whether there were court of appeals vacancies at the crucial point of the district judge's decision. Fortunately, this theoretical concern was not borne out in practice. For four of the judges, there were no vacancies in their circuits all year. For four other judges from the same state, the vacancy situation was stable through most of the year and changed only at a late point after which no other district judge in that circuit issued a Sentencing Guidelines opinion, thus leading us to believe it is most unlikely that our default date (which precedes the date of change) affects the calculation. For the single remaining district judge, the only change during the year was from a vacancy to a vacancy with a nominee. This judge was coded more conservatively as seeing a single vacancy with a nominee. In any event, we also conducted an alternative regression analysis on the outcome dependent variable excluding these nine judges with no effect on the PROMO-POT variable's significance.
275. For purposes of this study, we used volume 700 of the Federal Supplement to determine if the judge was listed as active or senior as of the end of 1988. To some extent, the list of senior judges determined in this manner will be overinclusive, as some judges may have been active as of the time of a decision on the Guidelines earlier in the year and then changed to senior status later in the year. However, this overinclusiveness is actually a more accurate measure for our purposes. We essentially are using senior status as a control for promotion potential, recognizing that senior status judges certainly will not be candidates for nor motivated by the possibility of promotion to the court of appeals. This same analysis applies directly to judges, who while active at the time of a decision, were plainly contemplating an imminent change in status as demonstrated by their transfer to senior status by the end of the year.
276. The lowest promotion potential factor score derived was -2.17 and the highest 2.94, with a mean of -.39. We assigned a score of -2.20 to senior and District of Columbia judges.
277. Believing that judges would be most influenced by and aware of other decisions within their own district, we initially created a "Within District Precedent" variable based upon a chronological chart of decisions by district. Unfortunately, this variable was flawed because of lack of variation. A solid majority of the judges had no prior rulings by other judges in their district, because judges in many districts issued or joined rulings simultaneously with other judges in the district, several judges were the only ruling judges in their particular district, and, of course, every district had to have an initial ruling by a judge. Accordingly, after preliminary tests, we abandoned this variable. An attempt to analyze this variable by excluding all judges for whom there was no prior Guidelines ruling in the district was unsatisfactory - we lost 187 of our 294 cases and, in any event, had no significant results.
278. A precedent was treated as "prior" for a particular judge's decision only if the former preceded the latter by at least seven days. The reasons for our "rule of seven days" were necessity and reality. The dates for decisions included in the information received from the Sentencing Commission were not necessarily precise; in some instances, the date is the date upon which the judge signed the order, but in other cases it is the date upon which the order was entered on the court docket. The difference between the date of signature and date of docketing was ordinarily less than a week. For a precedential factor, however, it means that a decision recorded as issued by Judge A on March 15, and a decision recorded as issued by Judge B on March 18, may as a practical matter have been issued slightly in opposite chronological order. In any event, decisions that were issued only seven or fewer days apart were essentially the equivalent of issuing simultaneous orders in which neither is precedent for the other. We concluded that a district judge: (1) may be unlikely to know about another district judge's decision entered on the docket only a week previously; (2) is less likely to be influenced by another decision that is so new as to be barely entered upon the docket sheet as compared to a decision entered earlier; and (3) is likely to have already been well along in the process of reaching a conclusion on such a momentous issue as the constitutional validity of the Guidelines when that decision is dated within a week of another judge's decision. Moreover, if a judge were profoundly influenced by a ruling that was issued just before he had intended to issue his own, his likely reaction would be to delay issuance of his opinion until it could be rewritten to take into account the new decision, thereby probably separating the date of issuance from the new precedent by at least seven days. For purposes of this precedential variable, we excluded, both as precedential for other judges and in terms of measuring the influence of precedent upon them, the nine judges for whom we did not have a specific date of decision. To avoid treating them as missing cases in the regression, we coded these nine judges as facing zero precedential influence. For the one judge who reversed himself by issuing decisions on both sides of the issue - separated by about one month - both decisions were counted for precedential purposes in opposite directions.
279. We counted precedent by judge, not by decision. This was important because of decisions like United States v. Ortega Lopez, 684 F. Supp. 1506 (C.D. Cal. 1988) (en banc), where 14 judges of the Central District of California joined in an en banc opinion invalidating the Guidelines and 10 joined in a dissent arguing the Guidelines' constitutionality. Counting only decisions would have underestimated the impact of such opinions.
280. For example, where there were 3 constitutional rulings and 7 unconstitutional rulings, the value would be (3-7) * \3-7\ / 10 = -1.60.
281. For example, where there was a prior single-judge decision rinding the Guidelines constitutional, the value would be (1-0) * |1-0| / 1 = 1.0. By contrast, where there were three prior single-judge decisions, two finding the Guidelines constitutional and one finding them unconstitutional, the value would be (2-1) * |2-1| / 3 = 0.33.
282. We are in the beginning stages of further work on the influence of precedent which would extend to measuring precedent by theoretical category.
283. Table 3 reports only the distribution of categorical variables. Other important, noncategorical variables are included in our standard set of variables as summarized in Tables 1 and 2 and reflected in later tables reporting results from logistic regression analyses.
284. Our selection of logistic regression may warrant some discussion. Our dependent variable - OUTCOME - is dichotomous, coded "1" if the judge ruled the Guidelines constitutional and "0" if not, and, as such, the usual linear regression models, such as Ordinary Least Squares (OLS), are not appropriate. OLS models, for example, allow the predicted values to fall outside the 0 to 1 range of our dependent variable. Moreover, OLS is relatively less efficient as the error cannot be normally distributed nor can it have constant variance. For a fuller discussion of these points, see Michael O. Finkelstein & Bruce Levin, Statistics for Lawyers 447-52 (1990); John Fox, Applied Regression Analysis, Linear Models, and Related Methods 442 (1997). In contrast to OLS models, logit and probit models deal quite well with a dichotomous dependent variable and either model is generally appropriate. We settled on a logit model for two reasons. First, logit models possess the practical advantages of relative ease and interpretability. See Fox, supra, at 444-46. Second, the weight of the empirical literature favors logit models. See, e.g., Eisenberg & Johnson, supra note 32, at 1185 n.155; Vicki Schultz & Stephen Petterson, Race, Gender, Work, and Choice: An Empirical Study of the Lack of Interest Defense in Title VII Cases Challenging Job Segregation, 59 U. Chi. L. Rev. 1073, 1121 (1992). In logistic regression, the dependent variable is the natural log of the odds ratio of the probability that an event occurs to the probability that it does not occur [L=log[p/(1-p)]]. For more on logistic regression models, see generally John H. Aldrich & Forrest D. Nelson, Linear Probability, Logit, and Probit Models (1984) (deriving equations for logit and probit models); Alfred DeMaris, Logit Modeling: Practical Applications (1992) (describing use and application of logit model); Eric A. Hanushek & John E. Jackson, Statistical Methods for Social Scientists 179-216 (1977) (explaining discrete variable problem and deriving logit and probit models). Although logit and probit models both transform the actual proportion responding on the independent variables, probit models do so by replacing the observed proportions with the value of the standard normal curve below which the observed proportion of the area is located.
285. Through data gathering (and subsequent statistical manipulation) we generated more than 100 variables, most different formulations of the same type of variable.
286. For example, as discussed in Part V.A.3 infra, we generated a variable AGE for a judge's age. Not surprisingly, a judge's age was closely related to the length of a judge's tenure on the bench, which was coded in the SENIORITY variable. Thus, for multicollinearity reasons, we could not, of course, include both variables in a single regression. Moreover, multicollinearity concerns precluded the inclusion of both ELECTED and POLITICAL or of both REAGAN and PARTY in the same logistic regression equation.
287. We adopted a two-fold approach to guard against multicollinearity: First, for every model presented we generated bivariate correlation matrices for all independent variables. No firm "rule" exists within the literature. See, e.g., George W. Bohrnstedt & David Knoke, Statistics for Social Data Analysis 407 (2d ed. 1988) (suggesting exclusion of variables where coefficients exceed 50%); Finkelstein & Levin, supra note 230, at 352 ("A simple (but not foolproof) test for multicollinearity involves looking for high correlations (e.g., in excess of .9) in pairs of explanatory variables . . . ."); Michael S. Lewis-Beck, Applied Regression: An Introduction 60 (1980) ("For diagnosis, we must look directly at the intercorrelation of the independent variables. A frequent practice is to examine the bivariate correlations among the independent variables, looking for coefficients of about .8, or larger."). In every instance, we adopted the more conservative approach. Thus, where two potential independent variables' coefficients exceeded 50%, we excluded one of the related variables. Second, even where coefficients did not exceed the 50% threshold, variables were excluded if two or more were intended to serve as proxies for the same idea or where, in theory, two variables were too closely linked. For example, although the bivariate coefficient for variables CASELOAD and CRIM-CASELOAD (-.15) did not exceed our .5 threshold, we did not include both variables in the same equation as both were designed to serve as proxies for a judge's workload.
288. To create the categories and codes, we initially analyzed a more or less random sample of opinions and discussed the possible codes. As we coded and discovered the need for additional codes, such as unusual variations or combinations of other codes, we added them.
289. A complete list of opinions and their codes is available from the authors on request.
290. A judge upholding the Sentencing Guidelines would have to address and reject all constitutional theories raised by the defendant, while a judge invalidating the Guidelines could seize upon a single theory to accomplish that end. However, any bias in the data is likely minimized by the following four factors. First, a judge upholding the Guidelines was obliged to address only those constitutional theories actually raised by a defendant in challenging the application of the Sentencing Guidelines in his or her case, and many defendants did not raise the due process theory. Second, a judge invalidating the Guidelines would often apply a "belt-and-suspenders" approach, striking the Guidelines on more than one ground. Third, several judges affirmatively rejected certain challenges to the Guidelines, even when ultimately accepting another as a basis for invalidation. Fourth, in the end, there were only four theories in the universe of possible constitutional challenges, and the two separation of powers arguments were frequently combined in presentation and resolution.
291. See infra Part V.A.2 (discussing disproportionate numbers of minority judges who adopted due process constitutional theory in invalidating Sentencing Guidelines).
292. The set of independent variables we employed to generate results for the content analysis dependent variables is structured quite similarly to that used for the general outcome dependent variable. See supra Part IV.A. However, because the precedent variable (PREC-CIR) was measured solely by outcome, see supra Part III.F, this variable is dropped from our standard set.
293. See supra Part II.B.1.
294. For example, in United States v. Christman, No. CR88-4-2 (D. Vt. Nov. 19, 1988) (en banc), three district judges invalidated the Guidelines in part on the grounds that judicial authority in the sentencing process includes the discretion "to apply reasonable and appropriate factors in the light of the circumstances of the case presented to the court" and that the Guidelines improperly "replace the judicial discretion vested in the courts by substituting the sentencing factors and punishment structured by the Sentencing Commission." Id. at 7-8. Although this reasoning is very similar to the due process claim of a right to individualized sentencing by a judge with full discretion, these judges had expressly refused to consider the due process argument, holding that the claim would not be ripe until a Guidelines sentence was actually imposed. See id. at 4-5. Thus, despite our impression that the reasoning fit more comfortably under a due process analysis, these judges considered their holding to be based upon a separation of powers violation, and we coded it as such.
295. See supra notes 85-92 and accompanying text.
296. If a judge found that the Sentencing Commission was appropriate in terms of function and operation in either the judicial branch or the executive branch, but objected to the mandatory participation of judges upon the Commission, the opinion was classified for branch location purposes without considering the deciding judge's view of the validity of judges serving on the Commission. The question of mandatory judicial participation upon the Commission was analyzed in the Separation of Powers-Judge Member category discussed below. Thus, for example, if a judge concluded that relocation of the Commission to the executive branch would not be enough to make the Commission constitutional only because of the participation of judges in an executive branch entity, the judge was treated for branch location purposes as concluding that the Commission would be constitutional in terms of function and operation in the executive branch. The invalidation of the Commission by reason of judge membership was reserved for separate analysis in that distinct phase of the study. See infra Part IV.B.2.a.ii.
297. We substituted these two variables for POLITICAL rather than adding them to our set of variables due to multicollinearity concerns. See supra notes 192-95 and accompanying text. Out of an abundance of caution, we also excluded the PROSECUTOR variable in this regression as well. Although EXECUTIVE was not defined to include federal prosecutors, we believe the variables are theoretically correlated and thus properly treated as though collinear.
298. See supra notes 93-97 and accompanying text.
299. The question of whether the Sentencing Reform Act validly authorizes the President to remove the judge members from the Commission was classified as a branch location concern for our study, rather than as a judge members concern. The removal question asked whether it was appropriate for the head of one branch to exercise removal power over officials in another branch, which thus focused upon the purported location of the Commission in the judicial branch. By contrast, the requirement that judges serve upon the Commission, whatever its branch location, raised other issues about judges participating in a lawmaking or regulationissuing body and whether that participation impaired the impartiality of the individual judges or the judiciary as a whole.
300. See supra notes 98-101 and accompanying text.
301. See supra notes 102-04 and accompanying text.
302. For coding purposes, we treated claims by defendants that the availability of probation as a sentence is compelled by due process as a species of the individualized sentencing right theory. The purported requirement that a judge be able to consider a certain sentencing option - whether it is probation or a shorter period of years - is inescapably an individualized sentencing claim.
303. We had and still have doubts about whether this categorization effort was fully successful and made objective and adequate distinctions. See infra Part V.A.6 (discussing correlation between law school education and practical reasoning). One reviewer criticized our practical/theoretical dichotomy, suggesting we may have confused formalism with theoretical reasoning. In his view, for example, we may have mistakenly characterized a formalistic discussion of separation of powers as theoretical although it did not evidence high-minded thinking. However, it was not our intent in this study to distinguish legal formalism from other theoretical forms of reasoning, but rather to distinguish both from the nitty gritty, "street smarts" practical legal thinking displayed from the bench by some trial judges. Nonetheless, the critique is well-taken, both in expressing some dissatisfaction with our implementation of this variable and in prompting us to undertake further exploration. Recognizing that the Sentencing Guidelines decisions are a treasure trove of data, we intend in a future study to further examine legal reasoning by coding these opinions by such analytical categories as legal formalism, legal realism/political, and aspirational/ prophetic.
304. See, e.g., United States v. Schwartz, 692 F. Supp. 331, 337-40 (D. Del. 1988) (engaging in pragmatic analysis of Sentencing Commission activities).
305. See, e.g., United States v. Chambless, 680 F. Supp. 793, 797-802 (E.D. La. 1988) (engaging in theoretical analysis of separation of powers issues).
306. When an opinion authored by a judge contained more than a de minimis amount of analysis, we coded it by reasoning approach, both in terms of practical versus theoretical reasoning and originalist versus nonoriginalist reasoning (as discussed infra Part IV.B.2.b.ii). When a judge recited that he was adopting the holding of another opinion, but offered no further analysis, we were unable to classify that opinion by reasoning approach. Thus, those opinions were coded only by the constitutional claims addressed and theories adopted in the cited opinion and are not included in this study of practical versus theoretical reasoning or the following study of originalist versus nonoriginalist reasoning. Likewise, when a judge participating in an en banc proceeding merely joined an opinion written by another judge, the joining judge was coded according to the constitutional claims addressed and theories adopted by the authored opinion, but not by reasoning approach.
307. See Stephen M. Griffin, American Constitutionalism 155 (1996) (arguing that textual, doctrinal, and prudential approaches to constitutional interpretation cannot be captured in originalist/nonoriginalist dichotomy).
308. See generally id. at 143-91.
309. Compare Robert H. Bork, The Tempting of America: The Political Seduction of the Law (1990),
310. Earl M. Maltz, Rethinking Constitutional Law: Originalism, Interventionism, and the Politics of Judicial Review (1994),
311. Michael W. McConnell, Originalism and the Desegregation Decisions, 81 Va. L. Rev. 947 (1995),
312. Henry P. Monaghan, Our Perfect Constitution, 56 N.Y.U. L. Rev. 353 (1981),
313. and Antonin Scalia, Originalism: The Lesser Evil, 57 U. Cin. L. Rev. 849 (1989),
314. with Laurence H. Tribe & Michael C. Dorf, On Reading the Constitution 8-13, 97-117 (1991),
315. Speech by Justice William J. Brennan to the Text and Teaching Symposium, Georgetown University (1985),
316. reprinted in The Great Debate: Interpreting Our Written Constitution 11 (1986);
317. see also Paul Brest, The Misconceived Quest for the Original Understanding, 60 B.U. L. Rev. 204 (1980);
318. Ronald Dworkin, The Arduous Virtue of Fidelity: Originalism, Scalia, Tribe, and Nerve, 65 Fordham L. Rev. 1249 (1997);
319. and Mark Tushnet, The U.S. Constitution and the Intent of the Framers, 36 Buff. L. Rev. 217 (1987).
320. See generally Barry Friedman & Scott B. Smith, The Sedimentary Constitution, 147 U. Pa. L. Rev. 1 (forthcoming 1998) (manuscript at 10-42, on file with the New York University Law Review) (presenting history of originalism, living constitutionalism, and historical fidelity as theories of interpretation, prevalent in many eras, including period between 1973 and 1997).
321. For commentary on the Bork nomination to the Supreme Court, see generally Bork, supra note 255, at 267-349; David G. Savage, Turning Right: The Making of the Rehnquist Supreme Court 133-46 (1992);
322. Stephen M. Griffin, Politics and the Supreme Court: The Case of the Bork Nomination, 5 J.L. & Pol. 551, 551-604 (1989).
323. See Griffin, supra note 256, at 567-604 (concluding that Senate rejected as unduly strict Bork's judicial philosophy of neutral principles and original intent, but more generally that Bork was defeated because of his criticism of Supreme Court decisions on fundamental rights and civil rights that had come to be generally accepted as legitimate achievements); Tribe & Dorf, supra note 255, at 3 (claiming that "the Senate's decision to withhold its consent was based in large part on its rejection of Judge Bork's belief that a quest for the 'original intent' of the Framers of the Constitution is the only proper method of interpreting the Constitution"). But see Bork, supra note 255, at 9 (arguing that "the difference about the proper role of the courts is what the battle over [Bork's] confirmation was underneath" but that "the public campaign, designed to influence senators through public opinion polls, consisted of the systematic distortion of [his] academic writings and [his] judicial record and, it must be said, employed racial and gender politics of a most pernicious variety").
324. See Mistretta v. United States, 488 U.S. 361, 413-27 (1989) (Scalia, J., dissenting).
325. Cf. Robert K. Rasmussen, A Study of the Costs and Benefits of Textualism: The Supreme Court's Bankruptcy Cases, 71 Wash. U. L.Q. 535, 540 (1993) (declining, for purposes of study of Supreme Court opinions, to define "textualism" as theory of statutory interpretation so narrowly as to exclude any opinion that includes discussion of legislative history even if only to confirm result "gleaned from a close examination of the text," saying that "[t]he Justices, like all good lawyers, attempt to marshal all possible arguments for their side").
326. See Antonin Scalia, A Matter of Interpretation 39 (1997) (observing that constitutional discussion is rarely "addressed to the text of the constitutional provision that is at issue, or to the question of what was the originally understood or even the originally intended meaning of that text");
327. Richard S. Kay, "Originalist" Values and Constitutional Interpretation, 19 Harv. J.L. & Pub. Pol'y 335, 336-40 (1996) (describing four methods advanced as originalism: (1) original text, (2) original intention, (3) original understanding, and (4) original values, and arguing that method of original intentions is most consistent with originalist values).
328. See Scalia, supra note 260, at 39 (stating that typical "starting point" for constitutional analysis today is "Supreme Court cases, and the new issue will presumptively be decided according to the logic that those cases expressed, with no regard for how far that logic, thus extended, has distanced us from the original text and understanding").
329. The results must be qualified both because we are looking at trial judges who may be less likely to be concerned with interpretation theory than appellate judges and because lower federal courts must follow Supreme Court precedents that may restrict the ability of a lower court judge to engage in originalist analysis.
330. We coded 14 judges as using an originalist approach to constitutional interpretation. All 14 originalists were independently coded as also engaging in theoretical (as opposed to practical) reasoning, as described in the previous subsection. This finding is not remarkable, as originalism by definition is a theoretical mode of analyzing a constitutional problem. Nonetheless, the two reasoning categories are not congruent and do not measure the same aspect of judicial reasoning. Of 59 judges who were coded as engaging in theoretical rather than practical reasoning, only a minority of 14 adopted an originalist as opposed to a nonoriginalist theoretical mode. Thus, the question remains, what influences a theoretically-oriented judge to select an originalist versus a nonoriginalist approach to resolving a constitutional problem?
331. Carol Gilligan, In a Different Voice: Psychological Theory and Women's Development (1982) (suggesting that men and women develop different forms of moral reasoning, with women tending to apply an interdependent "ethic of care" while men focus on rights-based abstract moral principles).
332. See Davis, Haire & Songer, supra note 28, at 129-30 (describing theories of gender difference and judging); Fox & Van Sickel, supra note 29, at 2-7 (same).
333. Davis, Haire & Songer, supra note 28, at 129.
334. Suzanna Sherry, Civic Virtue and the Feminine Voice in Constitutional Adjudication, 72 Va. L. Rev. 543, 582 (1986). Subsequently, Sherry has described her view as "moderate" - that life experiences and gender "may have a subtle effect on beliefs, attitudes, or approaches" - and has sharply criticized the more radical feminist view "that women have an entire world view that differs substantially from that of men and that is in some sense generally inaccessible to men."
335. Daniel A. Farber & Suzanna Sherry, Beyond All Reason: The Radical Assault on Truth in American Law 16, 30 (1997). Our study, especially in the context of other studies mentioned below, casts substantial doubt upon the viability of even this "moderate" view of a distinctive "feminine" approach to judicial decisionmaking.
336. Davis, Haire & Songer, supra note 28, at 130; see also Fox & Van Sickel, supra note 29, at 9-11 (describing theorized female voice traits of "community" and "context" and male voice traits of "individualism" and "rules").
337. See Gruhl, Spohn & Welch, supra note 28, at 311, 318-20 (hypothesizing that "women judges would be slightly more lenient than men judges," but actually finding little difference in sentencing behavior, although female judges were somewhat more likely to sentence female convicts to prison). Interestingly, Sherry fits Justice O'Connor's conservative approach to criminal issues into her feminine paradigm by asserting that this is evidence of a communitarian attitude: "If the community is more important than individual rights, it is quite predictable that Justice O'Connor would be a strong law and order proponent: she will protect the community from crime even at the expense of the individual rights of criminal defendants." Sherry, supra note 267, at 604. Whether this desire for community order or, alternatively, a preference for contextual evaluation, both of which Sherry describes as feminine virtues, would take priority in the context of the Sentencing Guidelines is debatable. In any event, the strong emphasis in feminist theory on contextual values versus abstract rules would appear to be poignantly implicated in the context of the Sentencing Guidelines. However, the presence of some doubt about how "different voice" female attitudes cut on the Sentencing Guidelines question may also indicate that feminist or feminine values do not provide a "coherent and consistent core of principles" to guide actual decisionmaking in adjudication. See Michael E. Solimine & Susan E. Wheatley, Rethinking Feminist Judging, 70 Ind. L.J. 891, 893 (1995) (contending that feminist jurisprudence is insufficiently coherent to direct judges in deciding cases).
338. Lucinda M. Finley, Breaking Women's Silence in Law: The Dilemma of the Gendered Nature of Legal Reasoning, 64 Notre Dame L. Rev. 886, 893 (1989).
339. See, e.g., Gruhl, Spohn & Welch, supra note 28, at 313-14 (including seven female judges in study and acknowledging that "findings need to be interpreted with caution"); Walker & Barrow, supra note 4, at 602 (including twelve female judges in study and even fewer for certain tests).
340. See supra notes 29, 34-35 and accompanying text; see also Ashenfelter, Eisenberg & Schwab, supra note 27, at 277-81 (finding that individual judge characteristics, including sex, did not appear to substantially influence mass of cases decided by district court judges). But see Davis, Haire & Songer, supra note 28, at 131-32 (finding female judges more supportive of claimants in employment discrimination cases, although when controlled for party of appointing President, variance declined and, for Republican-appointed judges, disappeared); Walker & Barrow, supra note 4, at 604-11 (finding that female judges were more deferential to positions taken by government in personal right and economic regulations cases).
341. See infra Part V.A.2.
342. See Davis, Haire & Songer, supra note 28, at 133 (suggesting possible reasons for absence of differences in judging between male and female judges).
343. See Walker & Barrow, supra note 4, at 615 (suggesting that common socialization experiences of legal education and screening of selection process may mute gender differences). But see Sheldon Goldman, Carter's Judicial Appointments: A Lasting Legacy, 64 Judicature 344, 351-52 (1981) (finding President Carter's women appointees to federal bench were less likely to have been political activists and less likely to have had prior judicial experience);
344. Elaine Martin, Women on the Federal Bench: A Comparative Profile, 65 Judicature 306, 310 (1982) (finding "several marked differences" in backgrounds of women recently appointed to federal bench, including that they were "far more likely to have been judges at the time of their appointment, far less likely to have been working for a large corporate law firm, and much less likely to have been party activists");
345. Elliott E. Slotnick, The Paths to the Federal Bench: Gender, Race and Judicial Recruitment Variation, 67 Judicature 371, 378-88 (1984) (finding distinctively different paths to federal bench for women, including alternative career backgrounds, such as being less likely to have held elective political office, less likely to have been in prominent private practice, and fewer years at bar).
346. Davis, Haire & Songer, supra note 28, at 133; see also Fox & Van Sickel, supra note 29, at 22 (speculating that "a natural feminine voice may actually be 'socialized out' of a woman judge's behavior over time (of course, this would still not account for the fact that in our study, male judges sometimes displayed feminine voice traits more than did women judges)"); Gruhl, Spohn & Welch, supra note 28, at 309 (suggesting that gender contrasts may be diluted by "powerful influences of socialization to the legal profession and to the judicial role" and influences of "courtroom 'workgroups'" such as prosecutors and defense attorneys).
347. Solimine & Wheatley, supra note 269, at 919; see also Fox & Van Sickel, supra note 29, at 12, 16 (finding, in study of local criminal trial judges, that "there was no clear pattern of women employing 'feminine voice traits' and men employing 'masculine voice traits'" and concluding that results "offer almost no support for the contention that women judges, by definition, bring to the judiciary a philosophy which embodies the feminine voice"). For a fascinating empirical exploration of the "different voice" judging theory involving the study of reasoning in opinions by five federal court of appeals judges, see Davis, supra note 150, at 148, 171 (finding that men adopted contextual approaches as often as women and that both chose rule-based approaches more often, thus concluding that results "do not provide empirical support for the theory that the presence of women judges will transform the very nature of the law"). For a future study, the set of written opinions in these Sentencing Guidelines constitutionality cases would provide an excellent source for evaluating the language and approaches used by female and male judges through an empirical model with a good number of women judges and involving the identical case problem.
348. Miriam Goldman Cedarbaum, Women on the Federal Bench, 73 B.U. L. Rev. 39, 44 (1993).
349. See supra notes 29, 36-37 and accompanying text.
350. Welch, Combs & Gruhl, supra note 29, at 127.
351. See Stephen Breyer, The Federal Sentencing Guidelines and the Key Compromises upon Which They Rest, 17 Hofstra L. Rev. 1, 4-5 (1988) (noting that Congress "relied upon statistical studies" indicating disparities in sentencing across gender and race). Indeed, the Sentencing Reform Act explicitly directs that the Guidelines are to be neutral as to the offender's race, sex, national origin, creed, and socioeconomic status. See 28 U.S.C. § 994(d) (1994).
352. See, e.g., Gerald W. Heaney, The Reality of Guidelines Sentencing: No End to Disparity, 28 Am. Crim. L. Rev. 161, 204-05 (1991) (concluding that study of Guidelines in four districts demonstrated disparate impact on Hispanic and African American defendants);
353. José Cabranes, Cabranes Rips Sentencing Rules, Legal Times, Apr. 11, 1994, at 17 (saying that "members of racial and ethnic minorities continue to fare worse under the guidelines than wealthier non-minorities do") [hereinafter Cabranes, Rips Sentencing Rules].
354. But see Joe B. Brown, The Sentencing Guidelines Are Reducing Disparity, 29 Am. Crim. L. Rev. 875, 877-78 (1992) (criticizing Heaney's interpretation of data as failing to account for different categories of offenses). Somewhat ironically, given the charge of continued racial disparity under the Guidelines, a very recent study found that two-thirds of the racial, gender, and economic disparities in federal sentencing are accounted for by the relatively small number of cases involving judicial departures from the Guidelines, rather than differential sentencing within the Guidelines.
355. See David B. Mustard, Racial, Ethnic and Gender Disparities in Sentencing: Evidence from the U.S. Federal Courts 17-20 tbl.6 (1998) (unpublished manuscript on file with the New York University Law Review). In addition, the Sentencing Guidelines, or more accurately, the underlying mandatory minimum sentence statutes, have been attacked as racially biased due to the heavier penalties imposed for possession with intent to distribute crack cocaine (more prevalent in minority communities) than powdered cocaine (more prevalent in white communities).
356. See Randall Kennedy, Race, Crime, and the Law 364-86 (1997) (discussing race and differences in sentencing for crack cocaine versus powder cocaine);
357. Paul Butler, (Color) Blind Faith: The Tragedy of Race, Crime, and the Law, 111 Harv. L. Rev. 1270, 1276-79 (1998) (criticizing Randall Kennedy's analysis and characterizing sentencing disparity for crack versus powder cocaine as racist);
358. see also David H. Angeli, A "Second Look" at Crack Cocaine Sentencing Policies: One More Try for Federal Equal Protection, 34 Am. Crim. L. Rev. 1211, 1213 (1997) (reporting that, in 1993, blacks accounted for 88.3% of federal crack cocaine distribution convictions (while 4.1% were white), but only 27.4% of those convicted of powder cocaine distribution (while 32% were white))
359. (citing United States Sentencing Commission, Special Report to the Congress: Cocaine and Federal Sentencing Policy (1995)). However, while the first federal sentencing statute imposing harsher penalties for trafficking in crack cocaine was enacted in 1986, see Anti-Drug Abuse Act of 1986, Pub. L. No. 99-570, 100 Stat. 3207 (1986) (codified as amended in scattered sections of 18, 21, 31 U.S.C.), the controversy concerning the racially disparate effect of this differential did not flare up until after the Sentencing Guidelines crisis and thus likely would not have affected attitudes toward the Guidelines during the period relevant for this study. See generally Kennedy, supra, at 364-80, 459-60 nn.39-44 (discussing and citing scholarly, media, and judicial treatment of crack versus powder cocaine differential in sentencing and disparate racial impact, which illustrate greater attention to issue arising around 1993 with a few earlier treatments dating to 1991). Indeed, as Randall Kennedy notes, about half of the members of the Congressional Black Caucus voted in favor of the law in 1986. See id. at 301. In any event, the crack versus powder cocaine differential is really more a consequence of statutory minimum sentences than of the Guidelines themselves. Indeed, in 1995, the Sentencing Commission recommended that Congress narrow the differences in sentencing for crack versus powder cocaine possession and distribution and proposed amendments to the Sentencing Guidelines equalizing "sentences for offenses involving similar amounts of crack cocaine and powder cocaine." United States Sentencing Commission, Amendments to the Sentencing Guidelines for United States Courts, 60 Fed. Reg. 25,074, 25,076 (1995). However, Congress passed and President Clinton signed a bill disapproving these proposals. See Federal Sentencing Guidelines, Amendment, Disapproval, Pub. L. No. 104-38, 109 Stat. 334 (1995).
360. See Paula C. Johnson, At the Intersection of Injustice: Experiences of African American Women in Crime and Sentencing, 4 Am. U. J. Gender & L. 1, 41-42 (1995). But cf. Kennedy, supra note 282, at 8-12, 21-24, 351-86 (disputing charge that disproportionate criminal prosecution and punishment of blacks for drug and other offenses reflect impermissible racial discrimination, and observing that, relative to their percentage of population, blacks both commit more and are more likely to be victims of street crime, thus leaving lawabiding black citizens in "dire need of protection against criminality").
361. None of the written opinions on the constitutionality of the Sentencing Guidelines included in our study mentioned disparate racial impact as an issue.
362. Cabranes, Rips Sentencing Rules, supra note 282, at 17.
363. Walker & Barrow, supra note 4, at 614.
364. See, e.g., Manning, supra note 29, at 6; Spohn, supra note 29, at 1211-14; Uhlman, supra note 29, at 891-94; Walker & Barrow, supra note 4, at 613-15; Welch, Combs & Gruhl, supra note 29, at 131-35. The one exception appears to be a study of federal appellate judges, see Gottschall, supra note 29, at 171-73 (finding pronounced disparity in voting by black and white judges in criminal cases).
365. Spohn, supra note 29, at 1212. See id. (suggesting that judicial socialization "produces a subculture of justice and encourages judges to adhere to prevailing norms, practices, and precedents" (citation omitted)); Uhlman, supra note 29, at 885 (suggesting that "atypically successful pre-judicial careers, a rigorous process of legal socialization, and special scrutiny for highly visible black jurists may attenuate the uniqueness of [a minority judge's] role" (citations omitted)); Walker & Barrow, supra note 4, at 615 (suggesting that judicial selection screening may mute differences among minority judges).
366. Ashenfelter, Eisenberg & Schwab, supra note 27, at 277-81 (finding that individual judge characteristics, including race, did not appear to influence mass of cases decided by district court judges).
367. Of the 28 minority judges in our study - a relatively healthy sample in comparison with many prior studies - 20 (71%) invalidated the Guidelines and 8 (29%) upheld the Guidelines.
368. The odds multiplier for RACE on the Due Process individualized sentencing issue is .02. The odds multiplier, a standard way of measuring the size of a variable's influence, is obtained by taking the antilog of the regression coefficient. See David W. Hosmer, Jr. & Stanley Lemeshow, Applied Logistic Regression 40-41 (1989).
369. The regression coefficients for each independent variable are reported in the tables for this study. An odds multiplier greater than 1.0 indicates that the variable's presence, holding other variables constant, increases the chance of a positive outcome on the dependent variable. An odds multiplier of less than 1.0 indicates that the presence of the variable reduces the chance of a positive outcome on the dependent variable. See Kevin M. Clermont & Theodore Eisenberg, Trial by Jury or Judge: Transcending Empiricism, 77 Cornell L. Rev. 1124, 1146 (1992);
370. Deborah Jones Merritt & Barbara F. Reskin, Sex, Race, and Credentials: The Truth About Affirmative Action in Law Faculty Hiring, 97 Colum. L. Rev. 199, 243 n.134 (1997). The greater the distance from 1.0 in either direction, the more sizable the positive or negative influence of the variable. However, a "change in the Odds' does not translate directly into a change in the probability that the outcome would occur." Merritt & Reskin, supra, at 243 n.134; see also Clermont & Eisenberg, supra, at 1146 n.54 (discussing difference between "odds" and "probability"). For statisticians, "[t]he odds of an event occurring are defined as the ratio of the probability that it will occur to the probability that it will not."
371. Marija J. Norusis, SPSS Advanced Statistics User's Guide 49 (1990). Thus, when applying the odds multiplier as a measure of the size of influence of a variable, "[t]he magnitude of the impact on probability depends on the initial odds." Merritt & Reskin, supra, at 243 n.134. In other words, we must first determine the initial odds ratio based upon the probabilities of the event occurring and not occurring, apply the odds multiplier to those initial odds, and then translate the increased or decreased odds into a probability percentage. See, e.g., id. at 257 n.178 (applying odds multiplier and translating result into probability); Norusis, supra, at 49-50 (applying odds multiplier). With respect to the Due Process Claim dependent variable, 42% of judges ruling upon this issue sustained the constitutionality of the Guidelines; thus, the "odds" of a ruling upholding the Guidelines were approximately .72 (.42 divided by .58). As the odds multiplier for the RACE variable is .02, the odds of a ruling sustaining the constitutionality of the Guidelines were decreased in the presence of this independent variable to approximately .01 (.02 multiplied by .72). These odds are equivalent to a probability of approximately 1%. Thus, holding all other variables constant, the RACE variable reduces the probability of a positive constitutionality ruling on the Due Process Claim from 42% to 1%.
372. The small number of minority judges in this sample (10) suggests some caution is due in evaluating this finding or generalizing from it. However, we do note that prior empirical studies of the influence of race on the behavior of judges also have involved relatively small numbers of minority judges. See Spohn, supra note 29, at 1200 (describing study as involving 13 black and 25 white judges); Uhlman, supra note 29, at 886 (contrasting judicial performance of 16 black and 75 white judges); Welch, Combs & Gruhl, supra note 29, at 129 (describing data as including decisions made by 10 black judges and 130 white judges).
373. A judge striking the Guidelines as unconstitutional on one theory did not need to address other challenges raised by the defendant, although some judges relied on multiple constitutional grounds to invalidate the Guidelines. See supra Tables 5, 6. Thus, not all of the judges addressed all four of the constitutional claims outlined in Part II.B.1. See supra Tables 5, 6 and accompanying text.
374. See, e.g., Fried, supra note 82, at 165 & 239 n.61 (describing due process ruling in United States v. Bolding, 683 F. Supp. 1003 (D. Md. 1988), rev'd, 876 F.2d 21 (4th Cir. 1989), as example of "absurd" ground for constitutional challenge to Guidelines, as contrasted with serious separation of powers grounds).
375. See Lockett v. Ohio, 438 U.S. 586, 604-05 (1978) (plurality opinion) (stating that "in noncapital cases, the established practice of individualized sentences rests not on constitutional commands, but on public policy enacted into statutes"); Woodson v. North Carolina, 428 U.S. 280, 304 (1976) (plurality opinion) (stating that "the prevailing practice of individualizing sentencing determinations generally reflects simply enlightened policy rather than a constitutional imperative").
376. 488 U.S. 361 (1989).
377. For court of appeals decisions rejecting the due process individualized sentencing challenge to the Guidelines, by order of circuit, see United States v. Doe, 934 F.2d 353, 356-57 (D.C. Cir. 1991); United States v. Sanchez, 917 F.2d 607, 614-15 (1st Cir. 1990); United States v. Vizcaino, 870 F.2d 52, 53-56 (2d Cir. 1989); United States v. Frank, 864 F.2d 992, 1008-10 (3d Cir. 1988); United States v. Bolding, 876 F.2d 21, 22-23 (4th Cir. 1989); United States v. White, 869 F.2d 822, 825 (5th Cir. 1989); United States v. Jacobs, 877 F.2d 460, 461-62 (6th Cir. 1989); United States v. Pinto, 875 F.2d 143, 144-45 (7th Cir. 1989); United States v. Blackman, 897 F.2d 309, 318 (8th Cir.), aff'd on reh'g, 904 F.2d 1250, 1258-59 (8th Cir. 1990); United States v. Brady, 895 F.2d 538, 543-44 (9th Cir. 1990); United States v. Thomas, 884 F.2d 540, 542-43 (10th Cir. 1989); United States v. Erves, 880 F.2d 376, 379 (11th Cir. 1989). See generally McCall, supra note 167, at 490-92 & n.129.
378. See Ulmer, supra note 33, at 625 (finding three factors - age at appointment, federal administrative experience, and religious affiliation - to have some explanatory value for decision variance in sample of 14 United States Supreme Court justices).
379. See Goldman, supra note 28, at 498-506 (finding age variable statistically significant and, along with party affiliation, best in accounting for variance in decisions on political measures by court of appeals judges).
380. See Gryski & Main, supra note 27, at 532-36 (finding chronological age significant in one model but not holding significance through multivariate analysis).
381. Beverly Blair Cook, Sentencing Behavior of Federal Judges: Draft Cases - 1972, 42 U. Cin. L. Rev. 597, 623 (1973).
382. Oliver Wendell Holmes, Law in Science and Science in Law, 12 Harv. L. Rev. 443, 455 (1899).
383. See infra Part V.D.3.
384. For this AGE variable, each judge's age was recorded by a whole number for number of years. For convenience, we simply subtracted the judge's year of birth from the year 1988 (i.e., for a birthdate of 1946, the number to be recorded would be 42).
385. Documentation for these and other alternative analyses mentioned in this study but not reported in the tables is available from the authors.
386. Not only was age not significant when included in an alternative regression on the outcome dependent variable, but inclusion of this variable in the set did not substantially affect the correlations of the other variables with the dependent variable. In particular, the signs of the coefficients remained unchanged, and no variable that was insignificant in the standard set analysis emerged as significant. The only variable that was significant in the standard set but insignificant in the alternative analysis (PROMO-POT) slipped just below significance at the 95% probability level. Indeed, the consistency of the alternative sets tends to confirm the collinearity of AGE and SENIORITY as variables.
387. Cf. Jack Bass, Unlikely Heroes: The Dramatic Story of the Southern Judges of the Fifth Circuit Who Translated the Supreme Court's Brown Decision into a Revolution for Equality (1981).
388. See fate & Handberg, supra note 26, at 467 (suggesting political orientations of American geographic regions likely have become more similar).
389. See Rowland & Carp, supra note 39, at 58-61, 85 (observing "that in aggregate, variations between judges in the North and South have markedly declined since 1977 although they are still to be found for some specific case types").
390. See id. at 63-65 (noting that, in category of criminal cases, variation between Northern and Southern judges dropped dramatically after 1977).
391. See id. at 66, 85; see also Sheldon Goldman, Picking Federal Judges: Lower Court Selection from Roosevelt Through Reagan 361 (1997) (stating that his comprehensive study "suggests that presidential agendas and judicial selection are intimately tied and that the policy agenda tends to predominate in times of political realignment").
392. Each judge was assigned to one of four geographical regions, based on the judge's district, NORTHEAST, MIDWEST, SOUTH, and WEST. See Bureau of Justice Statistics, Sourcebook of Criminal Justice Statistics-1994, at 625-26 app. 3 (1995).
393. Although the regional variables were found to be statistically significant at the 95% probability level in an exploratory chi-square analysis, this result warrants relatively little weight given the inherent limitations of chi-square analysis and the superiority of regression analysis in measuring the influence of independent variables.
394. Not only was region not significant when included in an alternative regression analysis, but addition of this variable to the alternative set did not substantially affect the correlations of the other variables with the
395. In a separate regression analysis, which is not reported in Table 4, we substituted the Crime Index for the Metropolitan Statistical Areas for each judge for whom we had information. Because of missing data, particularly for the states of Illinois and Florida, the N was reduced from 291 to 240, which is why we concluded this crime rate measure was unacceptable for our study. In any event, in this alternative regression analysis, the Crime Index for Metropolitan Statistical Areas did not approach significance on the outcome dependent variable.
396. See supra Part III.A (discussing CRIME-RATE variable).
397. See supra Part III.D.2 (discussing CASELOAD and CRIM-CASELOAD variables).
398. See James L. Gibson, Environmental Constraints on the Behavior of Judges: A Representational Model of Judicial Decision Making, 14 L. & Soc'y Rev. 343, 358-60 (1980) (finding that "incidence of crime does indeed influence judges' decisions, accounting for approximately one-sixth of the variance in sentencing behavior"). By contrast, another study of federal district judges found a slight correlation between higher crime rates for a city and milder sentences. See Cook, supra note 301, at 611. However, this study is limited because it examined only sentencing for military draft offenders. One might expect judges in high crime urban areas, when confronted with the basically passive misconduct of failure to comply with military conscription laws, to view them less seriously for sentencing purposes in comparison with more familiar and disturbing violent and property crimes.
399. See supra Table 7, Column 2.
400. Letter from Felix Frankfurter, Professor, Harvard Law School, to Mr. Rosenwald 3 (May 13, 1927) (Felix Frankfurter Papers, Harvard Law School library), quoted in Rand Jack & Dana Crowley Jack, Moral Vision and Professional Decisions: The Changing Values of Women and Men Lawyers 156 (1989).
401. But see James J. White, Letter to Judge Harry Edwards, 91 Mich. L. Rev. 2177, 2188 (1993) (stating that "[w]ith characteristic arrogance, Justice Frankfurter exaggerates the role of law schools in 'making law and lawyers'" and arguing that law professors "have a modest influence on the students and an even more limited impact on the law").
402. See Neal Devins, The Interactive Constitution: An Essay on Clothing Emperors and Searching for Constitutional Truth, 85 Geo. L.J. 691, 704 n.92 (1997) (reporting that 80.4% of law professors are Democrats, compared with 46.2% of full-time working population (citing James Lindgren, Measuring Diversity, Speech to the National Association of Scholars (Jan. 5, 1997)));
403. Michael Stokes Paulsen, Reverse Discrimination and Law School Faculty Hiring: The Undiscovered Opinion, 71 Tex. L. Rev. 993, 1001 (1993) (observing "the lack of conservative legal scholars on [law school] faculties and the hugely disproportionate percentage of faculty members who are political Democrats");
404. Eugene Volokh, Diversity, Race as Proxy, and Religion as Proxy, 43 U.C.L.A. L. Rev. 2059, 2073 n.23 (1996) (reporting that 12.9% of law professors are Republicans, compared with 41.0% of working population (citing James Lindgren, Measuring Diversity tbl.2 (unpublished manuscript))); see also Romer v. Evans, 116 S. Ct. 1620, 1637 (1996) (Scalia, J., dissenting) (contrasting "views and values of the lawyer class" on homosexuality, as evidenced by "law-school view" that disapproval of homosexuality is prejudice that must be "stamped out," with "the more plebeian attitudes" of members of Congress who have declined to extend federal civil rights laws to homosexuals).
405. See Richard A. Posner, Legal Scholarship Today, 45 Stan. L. Rev. 1647, 1652 (1993) (stating that "faculties of the leading American law schools are now substantially to the left of the judiciary . . . and of the public at large");
406. Christopher Wolfe, The Ideal of a (Catholic) Law School, 78 Marq. L. Rev. 487, 503 (1995) (stating that "political conservatives and traditional religious believers are . . . little represented at 'big-name' law schools"). Nor is this a new phenomenon. See Joan Chalmers Williams, At the Fusion of Horizons: Incommensurability and the Public Interest, 20 Vt. L. Rev. 625, 628 (1996) (describing Yale Law School as "a center of liberal thought" in 1930s and 1940s, with many professors playing central roles in New Deal).
407. See Howard, supra note 30, at 167-68.
408. See Mistretta v. United States, 488 U.S. 361, 413-27 (1989) (Scalia, J., dissenting); see also supra notes 134-40 and accompanying text (discussing Scalia's dissent in Mistretta). 325 See infra Part V.B.
409. The odds multiplier for the variable LAW-SCHOOL on the Practical dependent variable is 4.21, one of the higher values for significant variables derived in our study. For a discussion of the odds multiplier, see supra note 291. With respect to the Practical dependent variable, 34% of judges were coded as using practical reasoning, see supra Table 5; thus the odds of practical reasoning for the universe of judges were approximately .52 (.34 divided by .66). As the odds multiplier for LAW-SCHOOL is 4.21, the odds of use of practical reasoning were increased by the presence of this variable to approximately 2.19 (4.21 multiplied by .52). Thus, holding all other variables constant, the LAW-SCHOOL variable increased the probability of use of practical reasoning from 34% to 69%.
410. See supra note 249.
411. See supra notes 38-47 and accompanying text.
412. See Scherer, supra note 43, at 5 (stating that "there is a large body of literature which demonstrates that Presidents appoint judges who reflect their own political ideology"); see also Rowland & Carp, supra note 39, at 183-84 (finding that with Carter and Reagan appointees, differences between judges of same and opposite political party appointed by same President "have virtually disappeared"); Revesz, supra note 39, at 1718-19 (studying "the impact that a judge's ideology, using as a proxy the views generally held by the party of the appointing President, has on judicial decisionmaking" (footnote omitted)).
413. See Robert A. Carp & Ronald Stidham, Judicial Process in America 237, 241 (3d ed. 1996).
414. See Fried, supra note 82, at 165 (saying that "[i]n their heart of hearts I suspect [the Justice Department's Office of Legal Counsel] felt the [Sentencing Reform] Act was unconstitutional, and they were willing to see a defense go forward only on the most limited terms"); Douglas W. Kmiec, The Attorney General's Lawyer 79 (1992) (agreeing that he, as Assistant Attorney General in Department of Justice, was more concerned with constitutional separation of powers questions than with uniformity of sentences). One of the authors of this study was a member of the Department of Justice litigation team defending the Guidelines and participated in several of the meetings in which the Department's position was developed. He confirms the accuracy of Fried's and Kmiec's accounts and agrees with their perception of the internal debate over whether and how to defend the Guidelines while maintaining the Administration's strict view of separation of powers.
415. See supra notes 86-89, 92 and accompanying text; see also Fried, supra note 82, at 165 (saying that as far as Department of Justice "was concerned, the guidelines would be defended as an exercise of executive power - in spite of the words of the Act which lodged the Commission in the Judicial Branch - or it would not be defended at all").
416. Fried, supra note 82, at 133. See generally id. at 132-71; Kmiec, supra note 331, at 47-68.
417. See Mistretta v. United States, 488 U.S. 361, 422-27 (1989) (Scalia, J., dissenting) (arguing that Guidelines violate separation of powers); Gubiensio-Ortiz v. Kanahele, 857 F.2d 1245, 1251-66 (9th Cir. 1988) (Kozinski, J.) (striking down Guidelines on separation of powers grounds), vacated sub nom. United States v. Chavez-Sanchez, 488 U.S. 1036 (1989); see also supra notes 113-14; supra notes 134-40 and accompanying text.
418. See Fried, supra note 82, at 161-62; Cohen, supra note 21, at 185.
419. See Cohen, supra note 21, at 195.
420. See supra Table 4.
421. See Rowland & Carp, supra note 39, at 47-48 (explaining that Nixon was associated with law-and-order or "get tough on crime" political theme); Gottschall, supra note 29, at 168 (stating that Nixon focused on criminal issues when making judicial appointments) (citing Jon Gottschall, Nixon Appointees to the U.S. Courts of Appeals: The Impact of the Law and Order Issue on the Rights of the Accused, ch. 1 (unpublished Ph.D. dissertation, University of Massachusetts (Amherst))); Scherer, supra note 43, at 2 (explaining that Nixon promised to appoint "law and order" judges to federal bench).
422. The base case, which was not included in the regression run to prevent collinearity, was the combined dummy variable for judges appointed by Truman, Eisenhower, and Kennedy.
423. Ashenfelter, Eisenberg & Schwab, supra note 27, at 264 (comparing contract with civil rights cases).
424. Prior empirical studies of judicial behavior on an ideological axis have often fastened upon claims of individual rights as illustrating a classic liberal/conservative split. See, e.g., Stidham, Carp & Songer, supra note 43, at 19 (describing judicial outcomes as liberal or conservative based on rulings in civil rights and liberties cases); Walker & Barrow, supra note 4, at 604 (same).
425. See supra notes 293-97 and accompanying text (discussing nonmainstream nature of due process rulings).
426. The REAGAN dummy variable was significant at the 95% probability level. The odds multiplier was .11. For a discussion of the odds multiplier, see supra note 291. With respect to this dependent variable, 9% of judges addressing this issue located the Sentencing Commission in the executive branch; thus the odds of such a ruling were approximately .10 (.09 divided by .91). As the odds multiplier for REAGAN is .11, the odds of a ruling locating the Commission in the executive branch were decreased by the presence of this variable to approximately .01 (.11 multiplied by .10). Thus, holding all other variables constant, the REAGAN variable decreased the probability of a proexecutive branch ruling from 9% to 1%.
427. See supra Table 12.
428. See Kmiec, supra note 331, at 17-46; see also Fried, supra note 82, at 61 (stating that originalism "almost became the motto of the Meese Justice Department").
429. See supra Table 12, Column 1.
430. See supra Table 12, Column 2.
431. The odds multiplier on the Originalism dependent variable is .02 for the variable PARTY and .01 for the variable REAGAN. For a discussion of the odds multiplier, see supra note 291. With respect to the Originalism dependent variable, 34% of judges were coded as using practical reasoning, see supra Table 5; thus the odds of practical reasoning for the universe of judges were approximately .52 (.34 divided by .66). As the odds multiplier for PARTY is .02, the odds of originalist interpretation were decreased by the presence of this variable to approximately .01 (.02 multiplied by .52). As the odds multiplier for REAGAN is .01, the odds of originalist interpretation were decreased by the presence of this variable to less than .01. Thus, holding all other variables constant, the PARTY and the REAGAN variables each decreased the probability of originalist reasoning from 34% to 1%.
432. See Krotoszynski, supra note 167, at 418 n.4 ("Formalists believe that the text of the Constitution and the intent of the Framers concerning the proper relationship of the branches (to the extent that their intent can be determined) should control separation of powers analyses.").
433. See supra note 334 and accompanying text.
434. See Lawrence B. Solum, Originalism as Transformative Politics, 63 Tul. L. Rev. 1599, 1622 (1989) ("When one examines the actual practice of constitutional interpretation, it becomes quite clear that self-professed originalists . . . do not always (or even frequently) adhere to originalism as a practical strategy for deciding cases and writing opinions."); see also Mary Ann Glendon, Comment, in Scalia, supra note 260, at 95, 112 (stating that during era in which "the notions of principled judging and objective scholarship" have been abandoned by judges and scholars "if textualism, structuralism, and originalism advance, it can be predicted that selective deployment of textualism, structuralism, and originalism will advance as well").
435. Glick, supra note 26, at 292.
436. For example, Rowland and Carp propose a revised model of judicial behavior grounded in social cognition theory of human perception, memory, and interpretation of ambiguous information. See Rowland & Carp, supra note 39, at 164-69. Under this theory, human beings (including judges) respond to cues by reference to stored episodic or impressionistic knowledge - memory "schemata" or frames - that may influence behavior without conscious motivation. See id. Applying this theory to the Sentencing Guidelines decisions, a judge with prior experience as a criminal defense attorney who observed particular instances of believed injustice in the sentencing of clients might, upon assuming the bench, view sentencing issues through the framework of those personal experiences when interpreting the importance of an element or the meaning of a circumstance.
437. See supra notes 27-37 and accompanying text.
438. Goldman, supra note 28, at 500.
439. See infra Part V.E.
440. See supra Table 4, Column 2.
441. In addition, on the basic outcome variable, DEFENSE was significant at the 90% probability level on the standard set of variables, see supra Table 4, Column 1. On the "Separation of Powers-Branch Location" dependent variable, DEFENSE was also significant at the 90% probability level on the combined outcome subcategory, see supra Table 7, Column 1. Although we would not report significance at less than the 95% level as worthy of independent consideration, it does provide some additional support for the consistency of significance of the DEFENSE variable across dependent variable analyses. Even when it did not reach the 95% level, it frequently hovered close by.
442. See supra Table 8.
443. See supra Table 10.
444. The odds multiplier for DEFENSE on the "Separation of Powers-Judge Members" dependent variable is .08. The odds multiplier for DEFENSE on the "Due Process Claim Rulings" dependent variable is even stronger, at .03. For a discussion of the odds multiplier, see supra note 291. With respect to the "Separation of Powers-Judge Members" dependent variable, 35% of judges upheld the constitutionality of the Guidelines against this challenge, see supra Table 5; thus the odds of a ruling upholding the Guidelines were approximately .54 (.35 divided by .65). As the odds multiplier for DEFENSE is .08, the odds of a ruling sustaining the constitutionality of the Guidelines were decreased by the presence of this variable to approximately .04 (.08 multiplied by .54). Thus, holding all other variables constant, the DEFENSE variable decreased the probability of a positive constitutionality ruling on the Separation of Powers-Judge Members issue from 35% to 4%. With respect to the Due Process Claim dependent variable, 42% of judges deciding this issue upheld the Guidelines, see supra Table 5; thus the odds of a ruling upholding the Guidelines against this challenge were approximately .72 (.42 divided .58). As the odds multiplier for DEFENSE here is .03, the odds of a ruling sustaining the Guidelines were decreased to approximately .02 (.03 multiplied by .72). Thus, the DEFENSE variable decreased the probability of a positive constitutionality ruling on the Due Process Claim issue from 42% to 2%.
445. See supra notes 82-83 and accompanying text.
446. See Daniel C. Richman, Cooperating Clients, 56 Ohio St. L.J. 69, 118-19 (1995) (describing "the ideology that sustains some of the most zealous criminal defense lawyers" as "sense of 'heroism'" against government and suggesting that "the ideologically committed defense lawyer . . . sees the solidarity of the defense bar as a critical counterbalance to overzealous prosecutors").
447. See Davis, Haire & Songer, supra note 28, at 131-32 (noting that significant differences between male and female judges in criminal procedure cases disappeared once control for appointing president's party was added). But see Gryski & Main, supra note 27, at 531-32, 536 (finding sex not to be significant influence in study of state high court judges in sex discrimination cases, although number of female judges was too small for reliable conclusions).
448. Davis, Haire & Songer, supra note 28, at 133.
449. Id.
450. See Cohen, supra note 21, at 196 (stating that tendency of prosecutors to uphold Guidelines "could have been anticipated based on the overwhelming support that government prosecutors have expressed for the Commission"); Tate & Handberg, supra note 26, at 471 (stating that "prosecutor's job is to look after the legal interests of the government").
451. See Cohen, supra note 21, at 196; see also Eisenberg & Johnson, supra note 32, at 1190 (finding prosecutorial experience significant at 90% probability level); Nagel, supra note 32, at 336 (finding in early study that former prosecutors were significantly more likely to rule against defense in criminal cases); Tate & Handberg, supra note 26, at 474-76 (finding prosecutorial experience significant at 90% probability level).
452. See supra Table 4.
453. See supra Table 8, Column 2.
454. The odds multiplier for the PROSECUTOR variable in this alternative analysis (substituting the CRIM-CASELOAD variable for the CASELOAD variable in the standard set) on the "Separation of Powers-Judge Members" dependent variable is the moderately strong 2.46. For a discussion of the odds multiplier, see supra note 291. With respect to the Separation of Powers-Judge Members dependent variable, 35% of judges upheld the constitutionality of the Guidelines against this challenge, see supra Table 5; thus the odds of a ruling sustaining the constitutionality of the Guidelines were approximately .54 (.35 divided by .65). As the odds multiplier for PROSECUTOR is 2.46, the odds of a ruling upholding the Guidelines against this challenge were increased by the presence of this variable to approximately 1.33 (2.46 multiplied by .54). Thus, holding all other variables constant, the PROSECUTOR variable increased the probability of a positive constitutionality ruling on the Separation of Powers-Judge Members issue from 35% to 57%.
455. See supra notes 93-97 and accompanying text.
456. See supra Tables 5, 6.
457. Prosecutors are accustomed to performing multiple roles - exercising prosecutorial discretion in an almost judicial manner and then serving as advocates once the decision to prosecute is made. This could explain why judges with prosecutorial experience were favorably disposed toward judges performing multiple roles by participating on the Sentencing Commission. However, if judges with prosecutorial experience were indeed more supportive of the Commission's structure, it is difficult to explain why this positive disposition would not also be reflected in the basic outcome variable. Why would a judge's strong sympathy toward judicial service on the Commission not translate directly into approval of the Commission's validity? Moreover, as noted, even on the Separation of Powers-Judge Members dependent variable, PROSECUTOR was significant only on one of two alternative regression runs. Thus, while a possible explanation exists, in part, we are reluctant to promote it strongly.
458. Aliotta, supra note 39, at 279; see also Howard, supra note 30, at 169 (concluding, based upon interviews with judges, that "[t]he most active former politicians, contrary to lawyers' myths, did not become judicial activists").
459. See Cohen, supra note 21, at 194 (regression analysis), 196-97.
460. See supra note 192. From Cohen's article, it is not clear how he implemented his "Public Official (appointed)" and "Public Official (elected)" variables. In particular, we do not know whether he attempted to include county and municipal offices, for which information about the nature of selection to office is difficult to find. However, because his cell counts for these two variables approaches ours, even accounting for the larger number of judges overall included in our study (294 versus 196), we believe it likely that we adopted similar measures. Cohen coded 12 judges as holding appointed public office compared to 10 judges included in our APPOINTED variable, and he coded 11 judges as holding elected public office, compared to 23 judges in our ELECTED variable. See id. at 191.
461. See supra Table 7, Columns 2, 3.
462. For these analyses reported in Table 7, Columns 2 and 3, we also excluded the variable PROSECUTOR because of concerns that it was collinear and thus competed with the EXECUTIVE category.
463. We debated what implementation of the EXECUTIVE variable would most appropriately measure an individual's identification with and perhaps loyalty to this branch of our tripartite government. We agreed from the outset that the positions would have to be high-level ones that steeped a person in the policymaking activities of the executive branch, as opposed to lower-level governmental service. Accordingly, we limited it to selected state administrative offices listed by the Council of State Governments and to federal offices requiring Senate confirmation. We then developed three alternatives: "Strict Executive" (including only those who had served in high-level state administrative positions or federal executive branch positions requiring Senate confirmation), "Executive + Independent Agencies" (adding those who had served in positions requiring Senate confirmation in federal independent agencies), and "Expanded Executive" (adding federal prosecutors). We rejected the third alternative as unduly expansive and also because we doubted that service as a prosecutor in a United States Attorney's office engenders the same high-degree of association with the executive branch. Between the first and second alternatives, we chose the second (including both strictly executive branch and independent agency offices requiring Senate confirmation) as less narrowly defined than the stricter first alternative. In any event, the difference between the two alternatives was a single judge (who had served in an independent agency).
464. In this study, we have adhered strictly to the traditional 95% probability level as the minimum threshold for reporting significance, other than noting near significance above the 90% level on alternative analyses for variables that independently met the 95% standard in a reported regression analysis. In this particular case, however, the significance figure is p<.0508, thus meaning it falls short only due to rounding and by less than a tenth of a percentage point. Moreover, in certain preliminary alternative regression analyses, EXECUTIVE surpassed the mark. For these reasons, we believe it worthy of mention and explanation.
465. The odds multiplier for EXECUTIVE on this "Sentencing Commission in Executive Branch" dependent variable is among the highest found for any significant variable on any analysis in this study - 31.84. However, the small cell count could account for this exceptionally high multiplier. For a discussion of the odds multiplier, see supra note 291. With respect to this dependent variable, 9% of judges addressing this issue located the Sentencing Commission in the executive branch. Thus, the odds of a proexecutive branch ruling were approximately .10 (.09 divided by .91). As the odds multiplier for EXECUTIVE is 31.84, the odds of a ruling locating the Commission in the executive branch were increased by the presence of this variable to approximately 3.18 (31.84 multiplied by .10). Thus, holding all other variables constant, the EXECUTIVE variable increased the probability of a proexecutive branch ruling on the branch location issue from 9% to 76%.
466. See supra note 380.
467. See infra Part V.C.3.
468. See supra notes 30-31 and accompanying text.
469. See Aliotta, supra note 39, at 278-80; see also Eisenberg & Johnson, supra note 32, at 1190 (finding that judges with prior judicial experience treated racial equal protection claims more favorably).
470. Aliotta, supra note 39, at 279 (speaking of Supreme Court justices with prior judicial experience).
471. See supra Table 4.
472. The odds multiplier for JUDGE on the outcome dependent variable is 1.85 with the standard set of variables and 1.81 with the alternative set. For a discussion of the odds multiplier, see supra note 291. On the general outcome dependent variable, 39% of judges upheld the constitutionality of the Guidelines, see supra Table 5. Thus, the odds of a ruling sustaining the constitutionality of the Guidelines were approximately .64 (.39 divided by .61). As the odds multiplier for JUDGE is 1.85 with the standard set and 1.81 with the alternative set, the odds of a ruling in favor of the Guidelines were increased by the presence of this variable to approximately 1.18 (1.85 multiplied by .64) and 1.16 (1.81 multiplied by .64), respectively. Thus, holding all other variables constant, the JUDGE variable increased the probability of a positive constitutionality ruling from 39% to approximately 54% for both the standard set and the alternative set of variables.
473. In a study of criminal sentencing, Beverly Blair Cook hypothesized that judges with military experience would impose more severe punishment, but found the opposite. See Cook, supra note 301, at 624. Given the particular context of sentences for military draft offenders, she concluded that "more lenient sentences [by a judge with military experience might be] compensation for his known association with the military" or that he might have "no motivation to prove his devotion to national security by giving severe sentences since he already had earned his credentials." Id. at 624-25. That prior study thus provides little guidance in formulating a hypothesis for the effect of military background in our case.
474. See supra Table 7, Column 2.
475. The variable MILITARY is significant at the 95% probability level and negatively correlated with the Department of Justice's position that the Commission could be moved to the executive branch, see supra Table 7, Column 2. The odds multiplier for MILITARY is .25. For a discussion of the odds multiplier, see supra note 291. With respect to this dependent variable, 9% of judges addressing this issue located the Sentencing Commission in the executive branch. Thus, the odds of a proexecutive branch ruling were approximately .10 (.09 divided by .91). As the odds multiplier for MILITARY is .25, the odds of a ruling locating the Commission in the executive branch were decreased by the presence of this variable to approximately .03 (.25 multiplied by .10). Thus, holding all other variables constant, the MILITARY variable decreased the probability of a proexecutive branch ruling on the branch location issue from 9% to between 2% and 3%.
476. 28 U.S.C. § 991(a) (1994).
477. See Devins, supra note 321 (reporting that 80.4% of law professors are Democrats, compared with 46.2% of full-time working population (citing James Lindgren, Measuring Diversity, Speech to the National Association of Scholars (Jan. 5, 1997))); see also supra notes 321-22 and accompanying text (discussing disproportionate liberalism of law faculties).
478. See supra Part V.B.
479. See Timothy B. Tomasi & Jess A. Velona, Note, All the President's Men? A Study of Ronald Reagan's Appointments to the U.S. Courts of Appeals, 87 Colum. L. Rev. 766, 779 n.66 (1987) (finding in study of courts of appeals judges that "four of the five 'academic' Reagan judges studied . . . were more conservative than other Republican appointees"). During the 1980's, a common joke in conservative circles was that President Reagan was decimating the ranks of conservative law professors at leading law schools by appointing them to the federal bench.
480. See supra Table 11.
481. See supra notes 197-200 and accompanying text.
482. See supra Tables 9, 10.
483. The odds multiplier for ABA-AQ on the "Non-Delegation Doctrine" dependent variable is .34 and on the "Due Process Claims" dependent variable is .24. For a discussion of the odds multiplier, see supra note 291. With respect to the Non-Delegation Doctrine dependent variable, 69% of judges upheld the constitutionality of the Guidelines, see supra Table 5; thus, the odds of a ruling sustaining the constitutionality of the Guidelines were approximately 2.23 (.69 divided by .31). As the odds multiplier for ABA-AQ on this dependent variable is .34, the odds of a ruling favoring the Guidelines were decreased by the presence of this variable to approximately .76 (.34 multiplied by 2.23). Thus, holding all other variables constant, the ABA-AQ variable decreased the probability of a positive constitutionality ruling on the Non-Delegation Doctrine challenge from 69% to 43%. With respect to the Due Process Claims dependent variable, 42% of judges upheld the constitutionality of the Guidelines, see supra Table 5; thus, the odds of a ruling sustaining the constitutionality of the Guidelines were approximately .72 (.42 divided by .58). As the odds multiplier for ABA-AQ on this dependent variable is .24, the odds of a ruling upholding the Guidelines was decreased by the presence of this variable to approximately .17 (.24 multiplied by .72). Thus, holding all other variables constant, the ABA-AQ variable decreased the probability of a positive constitutionality ruling on the Due Process Claim challenge from 42% to 15%.
484. See supra notes 293-97 and accompanying text (explaining how Due Process Clause objection was outside mainstream in Guidelines litigation).
485. Because the dummy variable for judges receiving a Qualified rating (ABA-Q) was used as the reference, it is not reported with ABA-AQ and ABA-BQ in Tables 9 and 10. However, when we conducted alternative regression analyses, in which ABA-AQ was used as the reference, ABA-Q emerged as significant at the 95% probability level with a positive coefficient (i.e., association with a constitutional result) on both the Non-Delegation Doctrine and Due Process Claims dependent variables.
486. Historically, the committee has given lower ratings to female and minority nominees. See Elliot E. Slotnick, The ABA Standing Committee on Federal Judiciary: A Contemporary Assessment - Part 2, 66 Judicature 385, 387 (1983);
487. see also Henry J. Reske, ABA Judicial Ratings Draw Fire, A.B.A. J., Nov. 1994, at 38-39 (reporting charge that "the ABA's system for evaluating judges is erratic, racist and weighted in favor of lawyers who have worked for silk-stocking firms"). Incidents of apparent religious discrimination by the Committee have been reported.
488. See Victor Williams, The ABA Judgemaker Committee Is Exposed, Albeit Shaded from FACA Sunshine, 12 Geo. Mason L. Rev. 249, 260-62 (1990). Others view the Committee as elitist as well as biased in favor of those with prior judicial experience and trial lawyers. See Glick, supra note 26, at 141 ("Critics complain that the ABA procedures are highly elitist, since the committee mainly consults prominent lawyers and presidents of local bar associations in producing its judicial ratings."); Posner, Federal Courts, supra note 155, at 19 & n.27 (criticizing preference of Committee for former judges and trial lawyers).
489. The most controversial example of this purported politicization of the Committee was the minority rating of "Not Qualified" given to Supreme Court nominee Robert H. Bork in 1987 - allegedly based upon disagreement with his views on constitutional principles rather than his professional qualifications - by four members of the Standing Committee who were identified in the press as political opponents of the Reagan Administration. See Williams, supra note 403, at 264-66 (discussing ABA Committee's rating of Judge Bork, with citations to Senate reports and reports in various newspapers and legal periodicals); see also Bork, supra note 255, at 292-93 (describing ABA Committee's split vote as "extremely damaging to [his] nomination since the judgment was nominally about professionalism" and attributing this action to political opponents on Committee). Because of alleged politicization of the ABA, the Republican chair of the Senate Judiciary Committee recently terminated the Standing Committee's official role in confirmation proceedings, although the Clinton Administration intends to continue sending names of potential nominees to the Committee for evaluation. See Harvey Berkman, Hatch to ABA: You're Out. ABA: So What, Nat'l L.J., Mar. 3, 1997, at A6.
490. See Slotnick, supra note 403, at 393.
491. In addition, the ABA Committee's recent elimination of the "Exceptionally Well Qualified" rating may be a healthy development by preventing rated judges from thinking too highly of themselves as jurisprudential leaders. See American Bar Association, The ABA Standing Committee on Federal Judiciary: What It Is and How It Works 7 (1991) (describing three current ratings for judicial nominees).
492. But see William G. Ross, Participation by the Public in the Federal Judicial Selection Process, 43 Vand. L. Rev. 1, 67 (1990) ("[T]here is no merit to the argument that the ABA should simplify its ratings system and deem each candidate to be merely 'Qualified' or 'Unqualified.' The existing gradations help to distinguish exceptional candidates from marginal ones.").
493. However, we at least are willing to present a hypothesis for further investigation. After all, "statistics is a methodology and not a substitute for creative thinking." Frederick D. Herzon & Michael Hooper, Introduction to Statistics for the Social Sciences 7 (1976).
494. See Laura E. Little, Loyalty, Gratitude, and the Federal Judiciary, 44 Am. U. L. Rev. 699, 737 (1995) (writing, in context of whether federal judges feel obligation of gratitude toward benefactors responsible for placing judge on bench, that "a poor rating by the ABA can doom a candidate, but a favorable rating provides no guarantee of confirmation" and thus "a nominee who is eventually confirmed is, in most instances, likely to view a favorable ABA rating as one of many hurdles in the process, rather than a benefit requiring repayment in some form"). Whether this view is accurate or reflects the sentiments of judges receiving the higher ABA ratings, rather than a mere favorable Qualified rating, requires further empirical study.
495. In a very recent study, Professors William Landes, Lawrence Lessig, and Michael Solimine measured the influence of individual federal court of appeals judges by citations to their published opinions and found that judges who had received unqualified ratings were significantly less influential with their peers. See William M. Landes, Lawrence Lessig & Michael E. Solimine, Judicial Influence: A Citation Analysis of Federal Courts of Appeals Judges, 27 J. Legal Stud. 271, 325 (1998).
496. Posner, What Do Judges Maximize, supra note 52, at 20.
497. Cohen, supra note 209, at 16.
498. See Cohen, supra note 21, at 187.
499. See id.; see also Karle & Sager, supra note 165, at 444 (concluding, in statistical study of sentencing under Guidelines, that they have not measurably affected plea bargain rates).
500. See Cohen, supra note 21, at 187, 193.
501. See id. at 193-94.
502. The CASELOAD variable was significant at the 99% probability level on the dependent variable for "Practical v. Theoretical" reasoning, as reported in Table 11, and positively correlated with theoretical reasoning. In our view, this result is likely meaningless, as there is little reason why a judge's workload would influence the style of reasoning. At most, one could hypothesize that practical reasoning, which demands attention to detail and an evaluation of the factual setting, is more time-consuming and thus that a busy judge would prefer a theoretical dismissal of a claim. On the other hand, the constitutional question in the Sentencing Guidelines cases was largely abstract and divorced from the facts, such that practical reasoning might have been the path of least resistance here. Moreover, one could just as easily assume that a judge with a heavier workload would lack the time or patience to engage in high-level conceptual reasoning and feel compelled by circumstances to render a decision that is firmly and practically grounded. In any event, whether or not a valid explanation can be given for the significance of CASELOAD on this dependent variable, it would be distinct from and shed no light on the influence of workload on the judge's resolution of the constitutional challenge to the Guidelines.
503. See supra Table 4.
504. See supra Table 8.
505. The variable CRIM-CASELOAD was above the 94% probability level on this test. While we would not report this second result without the independent significance finding above the 95% level on the outcome test, this separation of powers test closely tracks and thus confirms findings at the outcome level. The separation of powers challenge to required judicial membership on the Commission was one of the primary objections and was addressed by 170 of the 188 judges who authored or participated in written opinions. See supra Tables 5, 6.
506. To begin with, the possibility of an error exists. As an additional test, we conducted an alternative regression analysis on the outcome variable with still another substitute measure of workload in the form of the raw criminal filings per district judge (that is, without comparison to total filings in the form of a ratio). This substitute criminal caseload variable was likewise both significant and positive. Going even one step further, we conducted yet another regression analysis with the criminal filings variable in which we excluded the CRIME-RATE variable to avoid any possible collinearity between the level of crime in the locality and the judge's criminal caseload - CRIM-CASELOAD remained significant and still positive.
507. Goldman, supra note 28, at 499. Goldman found little relationship between years of judicial experience and judicial voting behavior. See id. at 499-500.
508. See Bowman, Quality of Mercy, supra note 72, at 712 n.121 (stating that, in author's experience as prosecutor, "judges appointed since 1987 are much less resistant to the Guidelines than their predecessors," and that "[r]ecently appointed judges may not unanimously embrace the Guidelines, but many . . . seem to welcome the guidance afforded by a set of sentencing standards").
509. See supra Table 11.
510. As reported in Table 12, SENIORITY is significant at the 95% probability level in Column 2, which was an alternative regression analysis in which the dummy variable REAGAN was substituted for PARTY. With the standard set of variables, reported in Column 1, SENIORITY just barely fell below the 95% probability level (SENIORITY was significant at p<.06). For continuous variables, the odds multiplier represents the effect of a unit increase in the variable. See Clermont & Eisenberg, supra note 291, at 1146. Accordingly, for continuous variables included in this study, such as those representing caseload, seniority of judges, promotion potential, and precedent, determining and explaining the size of the influence is more difficult. Indeed, for the CRIM-CASELOAD ratio, the PROMO-POT factor score, and the PREC-CIR special ratio, identifying the unit and providing an easily accessible explanation is either impossible or more trouble than it is worth. However, for SENIORITY the unit is the familiar one of months and thus the odds multiplier can be translated into an understandable probability. The odds multiplier for SENIORITY on the "Practical versus Theoretical" dependent variable is 1.01 and on the "Originalist versus Non-Originalist" dependent variable is .95. With respect to the Practical dependent variable, 34% of judges adopted practical reasoning, see supra Table 5; thus the odds of adoption of practical reasoning were approximately .52 (.34 divided by .66). As the odds multiplier for SENIORITY on this dependent variable is 1.01, the odds of practical reasoning were increased by the presence of this variable to approximately .53 (1.01 multiplied by .52). Thus, holding all other variables constant, the SENIORITY variable increased the probability of use of practical reasoning from 34% to 35%. With respect to the Originalist dependent variable, 34% of judges adopted originalist interpretation, see supra Table 5; thus the odds of adoption of originalist interpretation were also approximately .52 (.34 divided by .66). As the odds multiplier for SENIORITY here was .95, the odds of use of originalist interpretation were decreased to approximately .49 (.95 multiplied by .52). Thus, the SENIORITY variable decreased the probability of use of originalist interpretation from 34% to 33%. Although these probability changes may seem very small, the unit again is months of seniority, and the sizable nature of these changes can be appreciated by considering how these changes would accumulate as a judge gains in tenure month-by-month.
511. See supra Table 2.
512. Epstein, supra note 51, at 831-32.
513. Id. at 836.
514. Id.
515. Even for the select few under consideration for a Supreme Court appointment, the ability to enhance promotion prospects through decisionmaking is most uncertain. Fifth Circuit Judge Patrick Higginbotham wrote an opinion highlighting criticism by commentators of Roe v. Wade, 410 U.S. 113 (1973), but ultimately stating that he was constrained by Supreme Court opinions to strike a restrictive Louisiana abortion statute. See Margaret S. v. Edwards, 794 F.2d 994, 995-99 (5th Cir. 1986). Another member of the panel who separately concurred was critical of Higginbotham's use of a circuit opinion to suggest disapproval of a Supreme Court precedent. See id. at 999-1000 (Williams, J., specially concurring). Ironically, when Higginbotham's name was floated for a possible Supreme Court nomination, some anti-abortion groups opposed his nomination on the grounds that he had authored an opinion striking the statute, ignoring his indirect statement of opposition to Roe. See Neal Devins, Through the Looking Glass: What Abortion Teaches Us About American Politics, 94 Colum. L. Rev. 293, 323 (1994) ("[P]ro-life interests scuttled the planned nomination of federal appeals court Judge Patrick Higginbotham to the Supreme Court because he recognized - albeit reluctantly - Roe to be the law of the land in one of his opinions.") (book review). If Higginbotham was attempting to signal his hostility to Roe as a means of gaining promotion to the Supreme Court, the effort failed. See also Posner, What Do Judges Maximize, supra note 52, at 5 (questioning impact of any particular decision on prospects for promotion from court of appeals to Supreme Court, saying "[s]ome decisions have no impact at all on those prospects and in the case of almost all the remaining decisions the impact is unpredictable - the decision may offend as many influential people as it pleases").
516. See Deborah J. Barrow, Gary Zuk & Gerard S. Gryski, The Federal Judiciary and Institutional Change 80 (1996); Posner, The Federal Courts, supra note 155, at 350.
517. See Cohen, supra note 21, at 192.
518. Id. at 188.
519. See id. at 193-94. Subsequently, in a longitudinal study of district court rulings in the context of criminal antitrust sanctions, Cohen also found that promotion potential explained a significant portion of the variance in corporate criminal antitrust penalties. See Cohen, supra note 209, at 27. Thus, Cohen has successfully replicated his findings in another context.
520. See supra notes 86-92, 317-23 and accompanying text.
521. See supra note 334 and accompanying text (discussing Sentencing Guidelines constitutionality opinions of Justice Scalia and Judge Kozinski).
522. There were other district court judges serving in 1988 who subsequently were promoted to the courts of appeals, but who had not issued any ruling on the constitutionality of the Sentencing Guidelines and thus are outside the scope of this study.
523. At the time Cohen conducted his study, five district judges who had ruled on the Guidelines had been elevated to the courts of appeals, and four of these five had upheld the constitutionality of the Commission, thus supporting his hypothesis. See Cohen, supra note 21, at 189. The fact that the judges who were elevated relatively soon after the Sentencing Guidelines Crisis were disproportionately favorable to the Guidelines may indicate that these decisions played a more significant role in promotion decisions at that time, but was of declining interest to those involved in judicial discretion over time.
524. See supra Part III.A.6.
525. See supra Table 4.
526. See supra Tables 8, 9.
527. PROMO-POT is also significant at the 95% probability level for an alternative analysis on the general "Separation of Powers-Branch Location" dependent variable, when the variable CRIM-CASELOAD is added to the set of independent variables. This is further evidence of the consistency of influence of the PROMO-POT variable.
528. See supra notes 87-89 and accompanying text (describing Department of Justice position).
529. See supra Table 7, Column 2.
530. See supra notes 85-92 and accompanying text (describing model briefs presented by Department of Justice and Sentencing Commission).
531. Cohen did not control for political party in his promotion potential analysis. See Cohen, supra note 21, at 193.
532. Because our theory suggests that the primary positive interaction is between judges appointed by Democratic presidents and promotion potential, we reversed the PARTY coding for this interaction term. Whereas we had coded Republican-appointed judges as "1" for the PARTY variable, we coded Democrat-appointed judges as "1" for the INTERACT variable so that both PARTY and PROMO-POT would be pointing in a positive direction for purposes of this construction.
533. Unless we believe that a particular variable (in this case PARTY) systematically distinguishes judges across background variables, then all judges should be included in the same equation. Because we have no basis for concluding that Democrat-appointed judges are systematically distinguishable from Republican-appointed judges on other independent variables, a division of these judges into separate regression equations is not theoretically appropriate. Nonetheless, because the single equation approach confirms that the statistical significance of PROMO-POT is destroyed by the introduction of the INTERACT term, our use of the separate equations approach for supplemental investigation may have some probative value. Indeed, the interesting and strikingly contrasting results for Republican-and Democrat-appointed judges - in a manner directly consistent with our hypothesis - suggest this approach had commonsense worth, however theoretically contraindicated.
534. See supra Part V.B (discussing party variables).
535. See Cohen, supra note 21, at 193.
536. See id. at 195.
537. In a subsequent study, in the context of antitrust criminal sanctions by district judges, Cohen added political party components to the promotion potential factor, and found that the promotion potential effect was more pronounced for Democrat judges than for Republican judges. See Cohen, supra note 209, at 16-27.
538. See, e.g., Bowsher v. Synar, 478 U.S. 714 (1986) (ruling that executive function of determining how to implement budget reductions could not be conferred upon legislative branch officer); Immigration & Naturalization Serv. v. Chadha, 462 U.S. 919, 951-59 (1983) (invalidating on separation of powers grounds "legislative veto," by committee or house of Congress, that would overturn executive decision or rule without passage of legislation by both houses); see also Fried, supra note 82, at 133, 158 (describing early successes of Reagan Administration's separation of powers struggle in Chadha and Bowsher). For more on the formalist view of separation of powers, see supra note 87 and accompanying text.
539. For recent empirical research on the influence of Supreme Court precedent on lower federal court decisionmaking, see Charles A. Johnson, Law, Politics, and Judicial Decision Making: Lower Federal Court Use of Supreme Court Decisions, 21 L. & Soc'y Rev. 325, 338-39 (1987) (finding that lower courts pay attention to Supreme Court decisions in decisionmaking, especially when facts, issues, or litigants are generally similar between cases);
540. Donald R. Songer, Jeffrey A. Segal & Charles M. Cameron, The Hierarchy of Justice: Testing a Principal-Agent Model of Supreme Court-Circuit Court Interactions, 38 Am. J. Pol. Sci. 673, 690-94 (1994) (finding that courts of appeals were highly responsive to changing policy trends in Supreme Court on criminal search and seizure, although there was still room for pursuing their own policy preferences).
541. 487 U.S. 654, 685-97 (1988).
542. See Mistretta v. United States, 488 U.S. 361 (1989); see also Fried, supra note 82, at 170 (describing Mistretta as "another nail in the coffin of a rigorous view of the separation of powers"); id. at 161-71 (discussing reasoning of Mistretta decision).
543. Prior to the confirmation of the Supreme Court's departure from a formalist approach to separation of powers in Mistretta, leading scholarly experts expressed doubt and uncertainty about whether the Morrison decision evidenced a significant shift in the direction of separation of powers jurisprudence. See, e.g., Stephen L. Carter, The Independent Counsel Mess, 102 Harv. L. Rev. 105, 105-07 (1988) (acknowledging that "the method that the majority employed in Morrison is markedly different from the one that the Justices have recently used to decide separation of powers cases" but suggesting that, "far from signalling a change in direction," Morrison "might herald the Supreme Court's resolution to stay the course" in sustaining legitimacy of independent federal agencies); Krent, supra note 87, at 1310-22 (suggesting that inconsistency of Morrison's approval of federal prosecutor independent of executive control with structural approach to separation of powers was "more apparent than real" given power of Congress to specify means that executive must employ to enforce criminal laws, and adhering to position that Sentencing Commission's formulation of sentencing policy exceeded constitutional powers of judiciary and constituted "a glaring departure from the case or controversy limitation on judicial action"). But see Theodore Y. Blumoff, Illusions of Constitutional Decisionmaking: Politics and the Tenure Powers in the Court, 73 Iowa L. Rev. 1079, 1082-83 (1988) (stating that Morrison demonstrated "a new willingness to reappraise [the Court's] separation of powers decisions" and arguing that stricter separation of powers rule "undermines the political branches' ability to fashion creative solutions to evolving problems").
544. See Gubiensio-Ortiz v. Kanahele, 857 F.2d 1245 (9th Cir. 1988), vacated sub nom. United States v. Chavez-Sanchez, 488 U.S. 1036 (1989).
545. The proportion of district court rulings on the Sentencing Guidelines was remarkably consistent across time, with 128 (60.4%) judges invalidating the Guidelines and 84 (39.6%) upholding the Guidelines prior to the Morrison decision in late June 1988, and with 51 (62.2%) striking the Guidelines and 31 (37.8%) approving the Guidelines after Morrison. In his earlier study, Cohen used only those Sentencing Guidelines decisions that were rendered between January and July of 1988. See Cohen, supra note 21, at 190. He suggests that termination of his study at that interim point was justified by the intervention of the Supreme Court's Morrison decision. See id. at 184 & n.4. Thus, he apparently believed that the precedential effect of Morrison, presumably suggesting the Court was amenable to a judicial branch Sentencing Commission, would control subsequent district court rulings. In retrospect, Cohen was correct that Morrison presaged the Mistretta decision, but was mistaken in believing that Morrison would influence district court rulings.
546. See Gubiensio-Ortiz, 857 F.2d at 1266.
547. See United States v. Frank, 864 F.2d 992 (3d Cir. 1988), cert. denied, 490 U.S. 1095 (1989).
548. Posner, Federal Courts, supra note 155, at 373.
549. See supra Table 4. Because the PREC-CIR variable measured precedent only in terms of general outcome, that is, by whether decisions held the Guidelines constitutional or unconstitutional, we could apply it only on the general outcome dependent variable. In a future study, we plan to explore precedential influence at the constitutional theory/reasoning level, creating additional precedent variables based on the written opinions.
550. See Cohen, supra note 21, at 194-96.
551. See id.
552. See id. at 195-96.
553. Id. at 196.
554. See Edward L. Rubin, Public Choice and Legal Scholarship, 46 J. Legal Educ. 490, 490 (1996) (describing public choice theory as "pessimistic model of human behavior").
555. See Cohen, supra note 21, at 190 (expecting that district court decisions on Sentencing Guidelines would reflect "an overall trend toward an unconstitutional ruling (as a result of the decrease in discretion) as well as toward a 'follow the leader' mentality as a result of the concern for peer recognition").
556. See William M. Landes & Richard A. Posner, Legal Precedent: A Theoretical and Empirical Analysis, 19 J.L. & Econ. 249, 272-73 (1976) (discussing different motivations judges might have for following precedents); see also Posner, What Do Judges Maximize, supra note 52, at 18 ("[B]y refusing to follow their predecessors' decisions and thus weakening the practice of decision according to precedent, they reduce the likelihood that their successors will follow their decisions.").
557. Glick, supra note 26, at 295.
558. Howard, supra note 30, at xxiii.
559. See Posner, What Do Judges Maximize, supra note 52, at 15.
560. Cf. Landes & Posner, supra note 469, at 292 (stating, in context of proposed economic model for judicial adherence to precedent, that "[t]he concept of precedent is at the heart of the way in which lawyers think about the legal system").
561. Johnson, supra note 453, at 338.
562. Solimine & Wheatley, supra note 269, at 910. See Kent Greenawalt, Private Consciences and Public Reasons 142 (1995): [T]he traditional model posits as a desirable aspiration an ideal that legal decision not depend on the personality of the judge. The aspiration is not fully achievable even if all judges are intelligent, well-trained, and conscientious, but it is worth striving for by emphasizing that bases of legal judgment should be open and available to all.
563. Kent Greenawalt, Private Consciences and Public Reasons 142 ( 1995): [T]he traditional model posits as a desirable aspiration an ideal that legal decision not depend on the personality of the judge. The aspiration is not fully achievable even if all judges are intelligent, well-trained, and conscientious, but it is worth striving for by emphasizing that bases of legal judgment should be open and available to all. Id.;
564. see also Mary Ann Glendon, A Nation Under Lawyers 128 (1994) (stating that classical judges were "openly resigned to the fact that total objectivity is an unattainable goal," but nonetheless insisted upon striving for impartiality and restraining their biases).
565. See Cross, supra note 48, at 326 (considering "the law to be ropes binding a judicial Houdini," consideration of empirical research may help us "understand which brand of rope and which type of knot are the most effective and inescapable").
566. Posner, What Do Judges Maximize, supra note 52, at 19.
567. See Joel Levin, How Judges Reason ix (1992) (arguing that "getting legal concepts and lines of precedent straight and in good order is not
568. See Richard H. Fallon, Non-Legal Theory in Judicial Decisionmaking, 17 Harv. J.L. & Pub. Pol'y 87, 88 (1994) ("American law cannot be reduced to any other discipline, nor can legal analysis be reduced to any other methodology.");
569. Charles Fried, The Artificial Reason of the Law or: What Lawyers Know, 60 Tex. L. Rev. 35, 57 (1981) (arguing that legal reasoning is "distinct method" of "analogy and precedent" that involves "the application of a trained, disciplined intuition where the manifold of particulars is too extensive to allow our minds to work on it deductively").
570. Posner, What Do Judges Maximize, supra note 52, at 24-25.