Measuring the effects of specialization with circuit split resolutions

Stanford Law Review

Volumn 63, Issue 5, 2011, Pages 1145-1176

  Hansford, Eric ^a  

^a STANFORD LAW SCHOOL   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 79957850724     PISSN: 00389765     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Note

References (149)

1. The regional circuits include the First through Eleventh Circuits and the D.C. Circuit. This encompasses all federal courts of appeals except for the Federal Circuit, which is considered a fully specialized court.
2. Richard A. Posner, Will the Federal Courts of Appeals Survive Until 1984? An Essay on Delegation and Specialization of the Judicial Function, 56 S. Cal. L. Rev. 761-778 (1983).
3. Lawrence Baum, Probing the Effects of Judicial Specialization, 58 Duke L.J. 1667-1671 (2009).
4. S. Jay Plager, The United States Courts of Appeals, the Federal Circuit, and the Non-Regional Subject Matter Concept, 39 Am. U. L. Rev. 853-60 (1990);
5. Richard L. Revesz, Specialized Courts and the Administrative Lawmaking System, 138 U. Pa. L. Rev. 1111-1123 (1990).
6. Edward K. Cheng, The Myth of the Generalist Judge, 61 Stan. L. Rev. 519-27 (2008) ("[T]his Article therefore uses the term 'specialized' to denote any court or judge that deviates from the generalist ideal.").
7. If judicial specialization leads to fewer decisionmakers, specialization may also end up increasing the uniformity and consistency of decisions. But specialization itself does not drive this effect.
8. See, e.g., Baum, supra note 3;
9. Henry J. Friendly, Reactions of a Lawyer-Newly Become Judge, 71 Yale L.J. 218-29 (1961);
10. Posner, supra note 2; Revesz, supra note 4;
11. Diane P. Wood, Generalist Judges in a Specialized World, 50 SMU L. Rev. 1755 (1997).
12. Some predict that specialization improves judicial performance primarily in complex subject matters. See, e.g., Revesz, supra note 4, at 1117 ("[P]roponents of specialization argue that specialized courts are more likely to make correct decisions in complex areas."); Note
13. Wood, supra note 7, at 1766 ("[Specialization] would ensure that the adjudicators were knowledgeable in the subject matters presented to them...."). Others think specialization improves judicial performance in all areas.
14. Rochelle Cooper Dreyfuss, Specialized Adjudication, 1990 BYU L. Rev. 377-378
15. "Most important, the court's expertise should enable it to craft better opinions, especially in fields where a small number of cases are now distributed rather thinly among the regional courts. Since generalist judges are confronted with the specialty subject matter infrequently, they lack the motivation, experience, and time to develop an understanding of the law. They decide the occasional case based upon a cursory understanding of policy and receive limited feedback on how well they fared. Thus, [a] specialized court's sustained involvement with a field would facilitate superior decisionmak-ing.".
16. But see Posner, supra note 2, at 780 ("A person who does only one job may perform better than an abler person who divides his time among several jobs, none of which he learns to do really well. But I wonder how transferrable this insight is from the industrial, technical, and academic fields where it is conventionally articulated to appellate judging.").
17. See Posner, supra note 2, at 779.
18. See id. at 784-85.
19. See Baum, supra note 3, at 1681.
20. Lawrence Baum, Specialization and Authority Acceptance: The Supreme Court and Lower Federal Courts, 47 Pol. Res. Q. 693 (1994);
21. Glynn S. Lunney, Patent Law, the Federal Circuit, and the Supreme Court: A Quiet Revolution, 11 Sup. Ct. Econ. Rev. 1 (2003);
22. R. Polk Wagner & Lee Petherbridge, Is the Federal Circuit Succeeding? An Empirical Assessment of Judicial Performance, 152 U. Pa. L. Rev. 1105 (2004);
23. Robert M. Howard, Comparing the Decision Making of Specialized Courts and General Courts: An Exploration of Tax Decisions, 26 Just. Sys. J. 135 (2005) (discussing the U.S. Tax Court).
24. Lawrence Baum, Specializing the Federal Courts: Neutral Reforms or Efforts to Shape Judicial Policy?, 74 Judicature 217 (1991); Dreyfuss, supra note 8.
25. Baum, supra note 3, at 1681.
26. See id. at 1682 ("Even when generalist and specialized courts decide the same types of cases, efforts to compare them can run into measurement problems.");
27. Ellen R. Jordan, Specialized Courts: A Choice?, 76 Nw. U. L. Rev. 745-784 (1981).
28. Baum, supra note 3, at 1681.
29. See 28 U.S.C. §§ 1291, 1294 (2006).
30. Technically they are not true "generalist" courts because the federal courts have limited subject matter jurisdiction.
31. Erwin Chemerinsky, Federal Jurisdiction 259 (4th ed. 2003) ("State judiciaries have general jurisdiction and may therefore hear all causes of action unless there is a statute denying them subject matter jurisdiction. But federal courts have limited subject matter jurisdiction; that is, they are restricted in what cases they may adjudicate and may exercise jurisdiction only if it is specifically authorized.").
32. But colloquially, most consider the federal courts of appeals (other than the Federal Circuit) general-ists, given that they have authority over diverse subject matters. See, e.g., Ruth Bader Gins-burg, An Overview of Court Review for Constitutionality in the United States, 57 La. L. Rev. 1019, 1021 (1997) ("With some notable exceptions... federal courts are not specialized tribunals; typically, they are generalist courts, and none of their members sit, as continental judges do, in sections divided by subject matter.").
33. See 28 U.S.C. §§ 1291, 1295.
34. Chief Justice Roberts wrote an article highlighting the circuit's unique caseload.
35. John G. Roberts, What Makes the D.C. Circuit Different? A Historical View, 92 Va. L. Rev. 375 (2006).
36. For example, because the District of Columbia has no federal prisons in its borders, prisoner petitions, "which make up a notable portion of the docket nationwide on other courts of appeals," represent only a small part of the D.C. Circuit's work. Id. at 376. On the other hand, "about two-thirds of the cases before the D.C. Circuit involve the federal government in some civil capacity, while that figure is less than twenty-five percent nationwide." Id. at 377. The D.C. Circuit also has exclusive jurisdiction over certain kinds of appeals. See, e.g., 26 U.S.C. § 7482(b)(3) ("In the case of any decision of the Tax Court in a proceeding under section 7478, such decision may only be reviewed by the Court of Appeals for the District of Columbia.").
37. The D.C. Circuit was initially included in the data collected but had to be dropped because its mix of cases diverged so significantly from that of any other generalist circuit. See discussion infra Part III.A.
38. Admin. Office of the U.S. Courts, Federal Judicial Caseload Statistics: March 31, 2009, at 32 tbl.B-7 (2009), available at http://www.uscourts.gov/uscourts/Statistics/FederalJudicialCaseloadStatistics/2009/tables/B07Mar09.pdf.
39. See id.
40. And slightly misleading since they do not correct for the size of the circuit.
41. See Baum, supra note 3, at 1672 ("I refer to 'judges' rather than 'courts' because the judge is the appropriate unit to consider. To the extent that specialization by case type affects what courts do, it is primarily because individual judges do work that has only a limited range in its subject matter.").
42. See Cheng, supra note 5, at 547 ("Using opinion writing as a measure of specialization tendencies is a reasonable choice, but it cannot capture all of the underlying behavior. For example, the metric necessarily misses the influence that a nonwriting expert might have on the ultimate opinion, whether at conference or during the opinion writing process.").
43. The subject categories used to construct Yearly Cases per Judgeship come from coding used by the Administrative Office of the Courts. See infra notes 111-15.
44. Courts of appeals sit in panels of three, so an average judge in the circuit will decide three times the cases reflected by the variable.
45. Kenneth J. Arrow, The Economic Implications of Learning by Doing, 29 Rev. Econ. Stud. 155-56 (1962).
46. The final analysis excluded the D.C. Circuit, see infra Part III.A, so Partially Specialized became the three highest and eight lowest.
47. I also tried dividing the circuits into the single highest and eleven lowest, the two highest and ten lowest, and the four highest and eight lowest. The results were consistent with those in Part IV and followed a predictable pattern: partial specialization had a stronger effect when designating the single top or top two circuits as "partially specialized," and a weaker effect when designating the top four circuits as "partially specialized." The difference was statistically significant when dividing the single top circuit from the other eleven circuits. Across the data, though, the most natural breaking point fell after the top three, so I used that division.
48. The one exception was Labor, in which the Seventh Circuit, with 7.55 Yearly Cases per Judgeship, was designated partially specialized and the Second Circuit, with 7.13 Yearly Cases per Judgeship, was not. In the Labor circuit splits the Second Circuit performed identically to the Seventh Circuit, so the coding does not affect my results. See Table 2 for complete summary statistics.
49. See sources cited supra notes 7-16.
50. Gary S. Becker & Kevin M. Murphy, The Division of Labor, Coordination Costs, and Knowledge, 107 Q.J. Econ. 1137-1141 (1992) (considering factors that affect the "degree of specialization").
51. See infra note 103 and accompanying text.
52. See Sup. Ct. R. 10(a);
53. Robert L. Stern, Eugene Gressman, Stephen M. Shapiro & Kenneth S. Geller, Supreme Court Practice 226 (8th ed. 2002)
54. "The Supreme Court often, but not necessarily, will grant certiorari where the decision of a federal court of appeals, as to which review is sought, is in direct conflict with a decision of another court of appeals on the same matter of [law]. One of the prime purposes of the certiorari jurisdiction is to bring about uniformity of decisions on these matters among the federal courts of appeals. Hence a square and irreconcilable conflict of this nature ordinarily should be enough to secure review, assuming that the underlying question has substantial practical importance." (emphasis omitted) (citation omitted).
55. See, e.g., Forest Grove Sch. Dist. v. T.A., 129 S. Ct. 2484, 2490 (2009) ("Because the Courts of Appeals that have considered this question have reached inconsistent results, we granted certiorari to determine whether § 1412(a)(10)(C) establishes a categorical bar to tuition reimbursement for students who have not previously received special-education services under the authority of a public education agency." (footnote omitted)); Note
56. Bridge v. Phoenix Bond & Indem. Co., 553 U.S. 639, 646 (2008) ("We granted certiorari to resolve the conflict among the Courts of Appeals on 'the substantial question' whether first-party reliance is an element of a civil RICO claim predicated on mail fraud." (citations omitted) (quoting Anza v. Ideal Steel Supply Corp., 547 U.S. 451, 461 (2006)); Logan v. United States, 552 U.S. 23, 30 (2007) ("We granted certiorari to resolve a split among the Circuits as to whether § 921(a)(20)'s exception for 'civil rights restored' should be interpreted to include civil rights retained at all times." (citation omitted)).
57. See, e.g., Dean v. United States, 129 S. Ct. 1849, 1852-53 (2009) (citing two sole opposing circuits as the "conflict among the Circuits").
58. See, e.g., Irizarry v. United States, 553 U.S. 708, 713 n.1 (2008) (identifying a split with five circuits on one side and six on the other).
59. See Symposium, Empirical Measures of Judicial Performance, 32 Fla. St. U. L. Rev. 1001 (2005).
60. Stephen Choi & Mitu Gulati, A Tournament of Judges?, 92 Calif. L. Rev. 299-305 (2004). It also lists "[c]aseload [p]erformance" (efficiency) and "[i]ndependence" as positive characteristics of judges. Id. at 309-10.
61. Lawrence B. Solum, A Tournament of Virtue, 32 Fla. St. U. L. Rev. 1365-83 (2005).
62. Ruth Bader Ginsburg, The Obligation to Reason Why, 37 U. Fla. L. Rev. 205-206 (1985).
63. Frank B. Cross & Stefanie Lindquist, Judging the Judges, 58 Duke L.J. 1383-1403 (2009).
64. Stephen J. Choi & G. Mitu Gulati, Choosing the Next Supreme Court Justice: An Empirical Ranking of Judge Performance, 78 S. Cal. L. Rev. 23-30 (2004).
65. Cross & Lindquist, supra note 43, at 1403; see also sources cited infra notes 57-63.
66. This aim seems least realistic because the hardest questions of law probably have no "objectively correct" answer. If two canons point in opposite ways for a hard question of statutory construction, it will be arbitrary to pick between the two without another guiding principle (e.g., how the Supreme Court would decide the issue, or enacting policy preferences).
67. The literature on judicial specialization supports the claim that these are the three most plausible aims. Lawrence Baum identifies two possible meanings of "[g]etting decisions right." Baum, supra note 3, at 1676. He says that in "most" discussions, "it implicitly refers to applying the law to the facts properly." Id. (emphasis added). Baum does not define "properly," but presumably he means either a purely objective assessment of the application in the abstract (the first aim I list) or an assessment according to a certain goal of the application (such as the goal of the second aim I list-to do what the Supreme Court would do). Baum also says "[g]etting decisions right" could mean making decisions "that best reflect[] a judge's conception of good public policy." Id. This is the third aim I list.
68. Decisions by Vote Split, SCOTUSblog (July 7, 2010), http://www.scotusblog.com/wp-content/uploads/2010/07/Final-Charts-070710-41.pdf.
69. Cross & Lindquist, supra note 43, at 1403.
70. Id.
71. Id.
72. See id. ("[C]ircuit courts may be regarded as agents of the Supreme Court, so it seems appropriate to consider the evaluations of their principal." (footnote omitted)).
73. See, e.g., Solum, supra note 41, at 1382-83.
74. See infra text accompanying notes 57-63.
75. The downward deference story may seem more compelling in the few cases of a dominant "expert" circuit, but not many examples come to mind beyond the Second Circuit in securities law, the D.C. Circuit in administrative law, and the Fifth Circuit in immigration law. Of these three examples, only the Second Circuit in securities law is included in my sample (immigration law is criminal, and the D.C. Circuit was excluded).
76. In the one Second Circuit case in a securities circuit split, the Court indeed agreed with the Second Circuit. But if some circuits really are perceived as such "experts" that they receive downward deference from the Supreme Court, it seems that the other circuits would defer to them as well (or reverse positions to conform to the "expert" circuit's opinion), obviating the need for the Supreme Court to resolve a circuit split in the first place.
77. The quality of the local bar in a given subject matter could link agreement rates with partial specialization. Regions that partially specialize in certain subjects probably often have better legal representation, and better legal representation probably makes courts more likely to "get it right." This relationship will not always hold: while the Fifth Circuit partially specializes in immigration law, see supra text accompanying note 22, its immigration lawyers are not particularly celebrated.
78. But it likely does hold for the securities lawyers in the Second Circuit and the administrative lawyers in the D.C. Circuit, the two other cases of extreme specialization. See supra text accompanying notes 20-23. For the subjects studied in this Note, see infra Figure 1, securities law seems to be the only one with dramatically different lawyer quality circuit to circuit. While it appears unlikely that the quality of representation drives the results, I cannot disprove that it might have some influence.
79. Circuit Scorecard, SCOTUSblog (July 7, 2010), http://www.scotusblog.com/wp-content/uploads/2010/07/Final-Charts-070710-10.pdf (using reversal rates as the unit of measurement in a "Circuit Scorecard").
80. Carol J. Williams, U.S. Supreme Court Looks over 9th Circuit's Shoulder, L.A. Times, June 29, 2009, http://articles.latimes.com/print/2009/jun/29/local/me-9th -scotus29 (attempting to explain why the Ninth Circuit has a higher reversal rate than average);
81. Dan Levine, 9th Circuit Racks Up Usual High Reversal Rate in Supreme Court Term, Law.com (June 30, 2008), http://www.law.com/jsp/article.jsp?id=1202422620128.
82. Roy E. Hofer, Supreme Court Reversal Rates: Evaluating the Federal Courts of Appeals, Landslide, Jan.-Feb. 2010, at 8. The Hofer article even assigns each circuit a letter grade (from A to F) based on its reversal rate. See id.
83. Kevin M. Scott, Supreme Court Reversals of the Ninth Circuit, 48 Ariz. L. Rev. 341-43 (2006).
84. Jerome Farris, The Ninth Circuit-Most Maligned Circuit in the Country-Fact or Fiction?, 58 Ohio St. L.J. 1465 (1997);
85. Richard A. Posner, Is the Ninth Circuit Too Large? A Statistical Study of Judicial Quality, 29 J. Legal Stud. 711 (2000);
86. Scott, supra note 60; see also, e.g., Cross & Lindquist, supra note 43, at 1402-14 (measuring judicial performance of individual judges with reversal rates).
87. See, e.g., Posner, supra note 61, at 716 tbl.4; Scott, supra note 60, at 342.
88. Posner, supra note 61, at 716.
89. 519 U.S. 316 (1997).
90. See Farris, supra note 61, at 1466.
91. See id.
92. See 519 U.S. at 323.
93. Lee Epstein et al., Circuit Effects: How the Norm of Federal Judicial Experience Biases the Supreme Court, 157 U. Pa. L. Rev. 833-77 (2009).
94. See id. at 874 fig.9.
95. Although all the home-circuit biases that the study identifies work against the Ninth Circuit.
96. Williams, supra note 58;
97. John Schwartz, 'Liberal' Reputation Precedes Ninth Circuit Court, N.Y. Times, Apr. 25, 2010, at A33 ("The highest court in the land, [Supreme Court practitioner Tom Goldstein] suggested, so clearly rankles at the views of Judge Reinhardt that litigators hoping to get his decisions overturned will go out of their way to cite him by name in their briefs.").
98. See Cross & Lindquist, supra note 43, at 1407 tbl.1 (showing that, between 1989 and 2000, Judge Reinhardt was reversed fourteen times, while the next most reversed judge was reversed only nine times).
99. See Circuit Scorecard, supra note 57 (showing more than 70% of Supreme Court opinions in the 2009 Term were reversals).
100. See supra note 35 and accompanying text.
101. See Admin. Office of the U.S. Courts, Judicial Business of the United States Courts: 2009 Annual Report of the Director 103 tbl.B-4 (2010), available at http://www.uscourts.gov/uscourts/Statistics/JudicialBusiness/2009/JudicialBusinespdfversion.pdf; see also id. at 104 tbl.B-4A (showing the circuit-by-circuit delays at each stage of the process broken down by type of case).
102. Id. at 103 tbl.B-4. The average circuit took about thirty-two months.
103. Id.
104. See Stern, Gressman, Shapiro & Geller, supra note 35, at 231 ("If it appears that upon a grant of certiorari the Supreme Court might be able to decide the case on another ground and thus not reach the point upon which there is a conflict, the conflict itself may not be a sufficient reason for granting review.... 'Resolution here of the... [issue in conflict among the circuits] can await a day when the issue is posed less abstractly.'" (second omission and alteration in original) (quoting The Monrosa v. Carbon Black Exp., Inc., 359 U.S. 180, 184 (1959))).
105. See Williams, supra note 58 ("[The Ninth Circuit] produces a lot of cutting-edge law, due to industries concentrated in the circuit and the large variation of underlying states and state criminal laws." (quoting Professor David Hoffman)).
106. See id. ("A lot of important policy cases involving interesting and difficult questions come out of the 9th Circuit. The West is known for its experimentation, the initiative process-things that bring constitutional questions to the fore more often." (quoting Professor Jeffrey L. Fisher)).
107. Jesse McKinley, Bush v. Gore Foes Join to Fight Gay Marriage Ban, N.Y. Times, May 28, 2009, at A1.
108. Not every factor would cause statistical bias in identifying a relationship between judicial performance and circuit partial specialization. But some would, and all would make the identification more difficult by adding noise.
109. See infra Table 1.
110. This arises because the Supreme Court's agreement rate in circuit splits with a given circuit reflects only instances when the circuit has disagreed with another circuit. For example, imagine a "good" court that decides questions of a given difficulty correctly 90% of the time, and a "bad" court that decides the questions correctly 60% of the time (little better than a coin flip). These are the "true rates." Both courts will decide the question correctly 54% of the time.
111. Both courts will decide the question wrongly 4% of the time. The courts will disagree the other 42% of the time: 36% of the time the good court will get it right while the bad court will get it wrong, and 6% of the time the good court will get it wrong and the bad court will get it right. Within the instances the courts split, the good court will appear to be right 86% of the time (36/42) and the bad court will appear to be right 14% of the time (6/42). The good court's observed rate is only slightly lower than its "true rate," while the bad court's observed rate is much lower.
112. See supra text accompanying note 44.
113. Continuing the example from note 84, imagine three good courts and three bad courts. The courts will all decide the question correctly about 16% of the time and all decide the case wrongly a trivial number of times. The other roughly 84% of the time the courts will split. Within the 84% (the circuit splits), a given good court will appear right 88% of the time and a given bad court will appear right 53% of the time. So observed rates with six courts come much closer to the true rates of 90% and 60%. As the number of courts increases the observed values would continue to approach the true rates.
114. See Stern, Gressman, Shapiro & Geller, supra note 35, at 255 ("It has been reiterated many times that the Supreme Court is not primarily concerned with the correction of errors in lower court decisions.... The Court's aim, rather, is to resolve the conflicts among the lower courts and to determine questions of importance.... The Supreme Court's burden and responsibility are too great to permit it to review and correct every misstep made by the lower courts in the application of accepted principles.").
115. See the Appendix for how the splits were identified. These four Terms were selected because, with the exception of the very start of the 2005 Term, the Supreme Court included the same nine Justices. Consistency minimizes worries that the Court may be deciding cases differently at different times in the sample.
116. Two opinions said that they were resolving circuit splits, but failed to lay out the relevant circuits. See Boyle v. United States, 129 S. Ct. 2237, 2242-43 (2009) ("Petitioner was then resentenced, and we granted certiorari to resolve conflicts among the Courts of Appeals concerning the meaning of a RICO enterprise." (citation omitted)); Zedner v. United States, 547 U.S. 489, 497 (2006) ("We granted certiorari to resolve the disagreement among the Courts of Appeals on the standard for analyzing whether a defendant has made an effective waiver of rights under the Act."). But these cases would have been dropped in later analysis because both concern criminal law.
117. As these omissions highlight, relying on the Supreme Court's identification of circuit splits is probably not the best identification method. The method relies on the Justices to be consistent in identifying the circuits in a split when resolving it. But the Justices may have incentives to present a split in a certain light, or inclinations toward not including the full split (or not including the split at all).
118. The better method would be to look at the legal question the Supreme Court resolved, and then to look back through the circuits (presumably relying on the petition for writ of certiorari) to find which courts had weighed in on the question. This is a time-consuming task that would take more than Westlaw searches. The splits resolved would probably be easiest to identify contemporaneously with the Supreme Court decisions, suggesting an institution like SCOTUS blog would be in the best position to collect this data.
119. See supra note 20 and accompanying text.
120. I tried running some of the tests described in the next Part with the D.C. Circuit included, and nothing changed significantly.
121. See, e.g., Arizona v. Gant, 129 S. Ct. 1710 (2009) (search incident to arrest).
122. See, e.g., Kimbrough v. United States, 552 U.S. 85 (2007) (Federal Sentencing Guidelines).
123. The correlation is 0.012, meaning that a single additional reversal of the D.C. Circuit (for example) would have made the correlation negative.
124. See supra Part II.B.
125. Lee Epstein et al., The Judicial Common Space, 23 J.L. Econ. & Org. 303-312 (2007) (Second and Ninth); Schwartz, supra note 71 (Second, Third, and Ninth).
126. See Epstein et al., supra note 97, at 312 (Fourth and Fifth); Schwartz, supra note 71 (Fifth and Sixth).
127. Compiled from SCOTUSblog end-of-Term "StatPacks." See Circuit Scorecard- OT08, SCOTUSblog (June 29, 2009), http://www.scotusblog.com/wp-content/uploads/2009/06/circuit3.pdf;
128. Circuit Scorecard-OT07, SCOTUSblog, http://www.scotusblog.com/wp-content/uploads/2008/06/scorecard07.pdf (last visited Oct. 30, 2010);
129. Circuit Sco-recard-OT06, SCOTUSblog, http://www.scotusblog.com/archives/ScorecardOT06.pdf (last visited Oct. 30, 2010);
130. SCOTUSBlog Circuits Chart: October 2005 Term, SCOTUSblog, http://www.scotusblog.com/archives/CircuitsFinal.pdf (last visited Oct. 30, 2010).
131. The Ninth Circuit was also affirmed in part and reversed in part 6.2% of the time. See sources cited supra note 99. No other court was affirmed in part and reversed in part. Giving the Ninth Circuit "half credit" for the cases affirmed in part and reversed in part would make its affirmance rate 13.9%.
132. In Figure 1, six of the seven subject matters converge on the same relationship between Supreme Court agreement and Yearly Cases per Judgeship: the circuits that the Court agrees with average higher Yearly Cases per Judgeship. If assigned randomly with equal probability, all or all but one of the seven subject matters would converge on the same relationship 12.5% of the time. They would converge on the above relationship (higher Yearly Cases per Judgeship for the circuit that the Court agrees with) 6.25% of the time. Neither falls below the 5% standard typically required for statistical significance.
133. The null hypothesis is that the standardized variable for circuits with which the Supreme Court agrees and with which it disagrees will be the same. A t-test gives t = 0.83, failing to reject the null hypothesis.
134. The LPM makes fewer parametric assumptions than a probit model, and its coefficients can be interpreted more easily.
135. Jeffrey M. Wooldridge, Introductory Econometrics: A Modern Approach 241-45 (2d ed. 2003) (explaining LPMs);
136. id. at 553-65 (explaining probit models). For extreme values the LPM can predict probabilities outside the unit interval. See id. at 243. But this problem will not arise for common values of the independent variable (including those near the mean). See id. I also use a linear functional form for Yearly Cases per Judgeship. The linear relationship models how an additional case should affect judicial performance because the judge learns and improves with each case.
137. Table 2 lists the standard deviation for each subject.
138. Within my observations the Supreme Court agreed with the Fifth Circuit 64.9% of the time, and with the average circuit 54.0%. See supra Table 1.
139. See supra Part I.A.
140. See also supra note 36 (giving examples of the Supreme Court identifying circuit splits it is resolving).
141. See, e.g., Engquist v. Or. Dep't of Agric., 553 U.S. 591, 596-97 (2008) ("The court below also acknowledged that other Circuits had applied Olech in the public employment context, id., at 993 (citing cases), but it disagreed with those courts.... We granted certiora-ri to resolve this disagreement in the lower courts...." (citing Engquist v. Or. Dep't of Agric., 478 F.3d 985 (9th Cir. 2007))).
142. See Admin. Office of the U.S. Courts, Federal Judicial Caseload Statistics: March 31, 2006, at 34 (2006), available at http://www.uscourts.gov/uscourts/Statistics/FederalJudicialCaseloadStatistics/2006/tables/B07Mar06.pdf.
143. In the final analysis, I ultimately omitted criminal cases. See supra text accompanying notes 93-94.
144. I used the 2000 census populations. See Your Gateway to Census 2000, U.S. Census Bureau, http://www.census.gov/main/www/cen2000.html (last visited Jan. 5, 2011).
145. See Admin. Office of the U.S. Courts, Civil Cover Sheet (2007), available at http://www.uscourts.gov/uscourts/Forms_AndFees/Forms/JS044.pdf.
146. See Cheng, supra note 5, app. B.
147. See Admin. Office of the U.S. Courts, Federal Judicial Caseload Statistics: March 31, 2008, at 31 tbl.B-7 (2008), available at http://www.uscourts.gov/uscourts/Statistics/Federal_JudicialCaseload_Statistics/2008/tables/B07Mar08.pdf;
148. Admin. Office of the U.S. Courts, Federal Judicial Caseload Statistics: March 31, 2007, at 31 tbl.B-7 (2007), available at http://www.uscourts.gov/uscourts/Statistics/Federal_JudicialCaseload_Statistics/2007/tables/B07Mar07.pdf;
149. Admin. Office of the U.S. Courts, supra note 111, at 30 tbl.B-7.