Emory Law Journal

The Most Important (and Best) Supreme Court Opinions and Justices
Frank B. Cross,
James F. Spriggs II Frank B. Cross is Herbert D. Kelleher Centennial Professor of Business Law, McCombs School of Business, University of Texas at Austin; Professor of Law, University of Texas Law School; Professor of Government, University of Texas at Austin.James F. Spriggs II is Sidney W. Souers Professor of Government and Professor of Political Science, Washington University in St. Louis; Professor of Law (by courtesy), Washington University in St. Louis School of Law.

Introduction

Identifying the most important cases decided by the Supreme Court is more than an interesting parlor game; the process illuminates the function of the law. The Court issues scores of opinions annually, some of which go on to assume great importance in future years, while many others languish in desuetude. Some opinions may appear to be important (e.g., they are commonly found in constitutional law casebooks), when in fact they have little real impact on the nation’s law. For purposes of this Article, we define importance in legal terms—opinions with greater legal importance are more relevant for deciding legal disputes and thus helping to structure legal outcomes.

The identification of key cases has practical significance for judicial research. When researchers study Supreme Court cases empirically, they commonly treat each case as an equally important data point. In reality, though, one single Supreme Court decision may be vastly more significant than numerous other small cases. 1 See Beverly B. Cook, Measuring the Significance of U.S. Supreme Court Decisions, 55 J. Pol. 1127, 1127 (1993) (“Decisions are not equal in significance . . . .”). There is reason to believe that the dynamics of decision making in especially salient cases may be different than for cases of lesser practical significance. 2 See Isaac Unah & Ange-Marie Hancock, U.S. Supreme Court Decision Making, Case Salience, and the Attitudinal Model, 28 Law & Pol’y 295 (2006) (finding that the influence of Justice ideology on decisions varies with case salience). The “lack of a valid, well-accepted, and ‘ready’ measure of salience” has resulted in significant “voids in our knowledge” of Supreme Court decisions. 3Lee Epstein & Jeffrey A. Segal, Measuring Issue Salience, 44 Am. J. Pol. Sci. 66, 72 (2000).

We enter this void with a study of the citations to past Supreme Court opinions. Citation analysis is “growing mainly because it enables rigorous quantitative analysis of elusive but important social phenomena,” including stare decisis. 4Richard A. Posner, An Economic Analysis of the Use of Citations in the Law, 2 Am. L. & Econ. Rev. 381, 382 (2000). The process “offers substantial promise of improving our knowledge of the legal system.” Id. at 402. Considerable quantitative research has been done on the outcomes of Supreme Court decisions, but the content of opinions has not been much studied. This is a serious limitation because it is the opinion—not the mere outcome—that is the Court’s salient product. 5For a critique of the research focus on outcomes, see Harry T. Edwards & Michael A. Livermore, Pitfalls of Empirical Studies That Attempt to Understand the Factors Affecting Appellate Decisionmaking, 58 Duke L.J. 1895, 1909 (2009) (noting that separate opinions reaching the same outcome may be very different in expressing the law). See also Thomas G. Hansford & James F. Spriggs II, The Politics of Precedent on the U.S. Supreme Court 3 (2006) (arguing that while case dispositions are important, “[t]he legal reasoning . . . can have more far-reaching consequences”); Martin Shapiro & Alec Stone Sweet, On Law, Politics, and Judicialization 98 (2002) (suggesting that “what judges say is even more important than how they vote”); Frank B. Cross et al., The Reagan Revolution in the Network of Law, 57 Emory L.J. 1227, 1234 (2008) (stressing that the significant aspect of a Supreme Court decision lies in the opinion, not the outcome, and that “[o]utcome-based research fails to capture the differential future significance of Supreme Court opinions”). Perhaps the leading empirical researchers of Supreme Court outcomes, Jeffrey Segal and Harold Spaeth, have conceded that it is the Court’s opinion that “constitutes the core of the Court’s policy-making process.” Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited 357 (2002) [hereinafter Segal & Spaeth, Attitudinal Model Revisited]. Past research “focused too narrowly on the disposition of the case.” 6Jack Knight, Are Empiricists Asking the Right Questions About Judicial Decisionmaking?, 58 Duke L.J. 1531, 1532 (2009). Knight notes the need to study “aspects of the opinions accompanying the votes.” Id. at 1533. Studies of case outcomes without consideration of opinion content can lead to very misleading conclusions. 7 See Barry Friedman, Taking Law Seriously, 4 Persp. on Pol. 261 (2006). The article examines decisions on affirmative action and concludes that “looking to outcomes rather than opinions leads to the wrong conclusion of what the court ‘did.’” Id. at 266. Friedman compares Justices Rehnquist and Thomas and notes that their votes appear quite similar but “if one reads the decisions authored by these Justices, it is apparent that the two are quite different in ways that have great significance for the law.” Id. at 267. This Article studies one aspect of opinion content to determine which opinions are most important, and seeks to ascertain why they are so important.

Identifying the most important opinions and the determinants of such importance has considerable legal significance. If the ideology of the Justices drives opinion importance, that fact has implications for decisions about the composition of the Court. If some feature of the opinion itself drives importance, that fact is crucial to our evaluation of the Justices, or Court norms and procedures. Perhaps a larger majority makes an opinion more important. Perhaps the use of more citations in an opinion gives it greater future impact. Perhaps some Justices are simply better at writing opinions of significance. Ascertaining such determinants is central to the evaluation of the Court and its members.

This Article embarks upon the project of identifying which Supreme Court opinions have proved the most legally significant and exploring why. We employ an analysis of citations to opinions. Other legal authors have used citation studies to assess the importance or value of opinions or judges. 8 See, e.g., Montgomery N. Kosma, Measuring the Influence of Supreme Court Justices, 27 J. Legal Stud. 333, 333 (1998) (using citation counts as a proxy for the influence of individual Justices); William M. Landes & Richard A. Posner, Legal Precedent: A Theoretical and Empirical Analysis, 19 J.L. & Econ. 249 (1976) (analyzing an opinion’s influence as a form of capital and measuring its depreciation over time through citation counting); John Henry Merryman, The Authority of Authority: What the California Supreme Court Cited in 1950, 6 Stan. L. Rev. 613 (1954) (the first empirical analysis of a court’s use of precedent, noting the effect of age on citation probability). A number of studies have used citations as a measure of the quality of circuit court judges. See Stephen J. Choi & G. Mitu Gulati, Choosing the Next Supreme Court Justice: An Empirical Ranking of Judge Performance, 78 S. Cal. L. Rev. 23 (2004); David Klein & Darby Morrisroe, The Prestige and Influence of Individual Judges on the U.S. Courts of Appeals, 28 J. Legal Stud. 371 (1999); William M. Landes et al., Judicial Influence: A Citation Analysis of Federal Courts of Appeals Judges, 27 J. Legal Stud. 271 (1998). The practice is also increasingly used in political science research in articles discussed throughout this Article. We build upon these existing analyses with more sophisticated measures and a focus on what makes Supreme Court opinions more or less important in the law.

The first Part of the Article sets out our criteria for identifying the most important Supreme Court opinions: the frequency of citation by subsequent judges and Justices. Citations are a facially clear measure of the importance of opinions, at least within the law itself. They are commonly used in research and offer an available measure for quantitative analysis. Like any empirical proxy used to represent a concept of interest, the use of citations to study discrete aspects of law is imperfect, but the primary criticisms of their use, such as the “settled law” phenomenon, do not invalidate the measure. 9 See infra Part I.B. Our analyses, for example, show that our measures of case importance correspond to perceptions of case importance.

In the second Part, we quantify the most important Supreme Court opinions. We identify the opinions with the most citations at the Supreme Court, circuit court, and district court levels. This produces very different lists, revealing different dimensions of importance depending on the level of the judiciary. We also provide an additional list for the Supreme Court using a more sophisticated measure of importance available from analysis of the full network of citations at the Court. This enables us to identify the most overrated and underrated opinions of the Court.

Having various measures at different levels of the judiciary to assess importance, the third Part of this Article analyzes what makes an opinion more or less important. We analyze the role of characteristics of the case itself, the age of the precedent, the role of ideological factors, various opinion characteristics, and control variables. Through multiple regression analysis, we discover that all these characteristics are relevant for understanding importance, though particular results are not always as expected according to prevailing theories. Features of the opinion itself that appear to matter include, for instance, its length and the number of citations it contains.

The fourth Part examines the associations of opinion importance and the Justice who authored the opinion. With the ability to control for specific case characteristics from the preceding Part, we examine whether opinions authored by different Justices have greater future citation power, whether at the Supreme Court or lower court levels. Our analysis shows that a few Justices appear to author particularly influential opinions.

The fifth and final Part assesses whether the importance of the Supreme Court’s opinion, significant in itself, may also be considered a measure for the best opinions. While the notion of the “best” opinion is inevitably a subjective one, our quantitative empirical analysis provides a reasonable guide for opinion quality. While no study can provide conclusive answers in itself, we provide the first quantitative analysis of opinion importance and quality, upon which we hope others will build.

I. Criteria for Identifying the Most Important Supreme Court Opinions

Some efforts have been made to identify the most important opinions of the Supreme Court. The Oxford Guide to Supreme Court Decisions summarizes those considered to be the most important (the “Oxford list”). 10 The Oxford Guide to United States Supreme Court Decisions (Kermit L. Hall ed., 1999) [hereinafter The Oxford Guide]. The book describes its contents as a discussion of “the 440 most important cases in the Court’s history.” Id. at vii.Congressional Quarterly publishes a Guide to the U.S. Supreme Court that lists cases it considers to be of landmark status. 11 Joan Biskupic & Elder Witt, Congressional Quarterly’s Guide to the U.S. Supreme Court (3d ed. 1997). This source has been used in research as a guide to the most important decisions of the Court. See, e.g., Jeffrey A. Segal & Harold J. Spaeth, The Influence of Stare Decisis on the Votes of United States Supreme Court Justices, 40 Am. J. Pol. Sci. 971 (1996). These rankings of case importance are based on the assessments of legal experts, considering each case’s “historical and/or social significance, its importance to the development of some area of the law, its impact on the development of American government, and relatedly, its prevalence in legal textbooks.” 12James H. Fowler & Sangick Jeon, The Authority of Supreme Court Precedent, 30 Soc. Networks 16, 20 (2008). These lists of cases have been used in academic research as a screen for the most important decisions. 13 See, e.g., Segal & Spaeth, supra note 11. The Congressional Quarterly compilation has a distinguished list of compilers, but they did not indicate their criteria for inclusion, and the list may have a bias for constitutional decisions. 14 See Epstein & Segal, supra note 3, at 69. The list is also merely binary, categorizing cases as major or not, without any other differentiation among individual opinions.

One of the most accepted measures for case importance in social science is New York Times front-page coverage of a Supreme Court opinion when issued. 15The initial case for the reliability of the measure is found in Epstein & Segal, supra note 3. The measure has been used in numerous subsequent articles, including Michael A. Bailey et al., Signals from the Tenth Justice: The Political Role of the Solicitor General in Supreme Court Decision Making, 49 Am. J. Pol. Sci. 72 (2005); Vanessa A. Baird, The Effect of Politically Salient Decisions on the U.S. Supreme Court’s Agenda, 66 J. Pol. 755 (2004); Paul M. Collins Jr., Towards an Integrated Model of the U.S. Supreme Court’s Federalism Decision Making, 37 Publius 505 (2007); James H. Fowler et al., Network Analysis and the Law: Measuring the Legal Importance of Precedents at the U.S. Supreme Court, 15 Pol. Analysis 324 (2007). This standard contrasts with both the Oxford and Congressional Quarterly lists because it is a contemporaneous (rather than retrospective) standard. Any identification of the most important cases in history should take advantage of how those cases were used over time. The contemporaneous New York Times measure could still have accuracy as a predictor of future importance, and the future significance of a decision may be obvious. Use of this measure may be distorted though, as front-page coverage is surely contingent on the day’s other news, and the measure may have an ideological or geographical bias. 16 See Forrest Maltzman & Paul J. Wahlbeck, Salience or Politics: New York Times Coverage of the Supreme Court (Apr. 3–6, 2003) (unpublished manuscript) (on file with authors) (discussing the geographic bias). Maltzman and Wahlbeck also suggest that coverage is affected by the number of votes in the majority and whether the Chief Justice wrote the opinion, among other biasing factors. Id. As with the accepted compilations of case importance, however, this standard is a binary one that simply puts cases in the categories of significant or not significant—without further differentiation.

Another possible standard for importance is inclusion in major law school constitutional law casebooks or political science texts. 17 See, e.g., Cook, supra note 1 (discussing assorted measures of Supreme Court opinion significance). This measure has an obvious bias for constitutional decisions, excluding all others, and suffers other deficiencies as well. The authors of casebooks, though expert, provide a small sample of commentators. In addition, they may choose cases that are pedagogically useful rather than those with the greatest importance. 18For a review of the constitutional canon in casebooks, see J.M. Balkin & Sanford Levinson, The Canons of Constitutional Law, 111 Harv. L. Rev. 963 (1998). Balkin and Levinson address various features, beyond opinion importance, which go into the text of casebooks, including the significance of the particular audience. Id. at 976. They observe that some cases may be included precisely because they are “wrongly decided or, even if the correct result is reached, offer styles of reasoning that the authors wish to question or criticize.” Id. at 982. The choices may also be influenced by the ideology of the authors. Id. at 998.

Another suggestion has been to use the number of law review notes received by a case as a cue for significance. 19 See Dennis Haines, Rolling Back the Top on Chief Justice Burger’s Opinion Assignment Desk, 38 U. Pitt. L. Rev. 631 (1977). While one might prefer law students to journalists as a resource for case significance, this measure too has the lack of historical perspective and possible geographical and ideological biases. Nor does it appear facially valid, as Brown v. Board of Education was treated in many fewer law review notes than Fuentes v. Shevin, yet the former case is generally considered far more significant. 20 See Saul Brenner & Harold J. Spaeth, Stare Indecisis: The Alteration of Precedent on the Supreme Court, 1946–1992, at 25 (1995). Some have suggested treating as important cases headlined on the cover of the advance sheets of the Lawyer’s Edition of the U.S. Supreme Court Reports, 21Harold J. Spaeth, Distributive Justice: Majority Opinion Assignments in the Burger Court, 67 Judicature 299 (1984). but this measure is too expansive, including nearly all the decisions rendered by the Court. 22 Brenner & Spaeth, supra note 20, at 25.

Others have argued for measuring salience based on the number of amicus briefs filed at the Court, but this tool contains a substantial bias by case type, likely reflects other case traits (such as legal complexity and ambiguity), and is difficult to use as a historic measure due to the lack of much amici activity until the mid-1900s. 23Epstein & Segal, supra note 3, at 69; see Paul M. Collins, Jr., Amici Curiae and Dissensus on the U.S. Supreme Court, 5 J. Empirical Legal Stud. 143, 151–53 (2008). Moreover, any measure using amici could only reflect the state of the case as it approached the Court, not the resultant opinion; such a measure is thus incapable of capturing change in the legal importance of an opinion over time. While the presence of numerous amici is surely meaningful (and we will use this in our analysis), it is not an ideal measure for the importance of Supreme Court decisions.

Some would suggest that “activist” decisions of the Supreme Court are the most important. Certainly some decisions regarded as activist (Brown or Roe v. Wade or Miranda v. Arizona) clearly seem quite significant. However, the notion of activism is quite vague and “often in the eye of the beholder.” 24 Stefanie A. Lindquist & Frank B. Cross, Measuring Judicial Activism 1 (2009). A decision striking down a federal statute might seem quite significant, but federal statutes vary considerably in their practical significance. There are many different criteria for judicial activism, which makes it difficult to isolate such cases. 25 See id.

In a popular book, Bernard Schwartz has produced a list of the “top ten” greatest Supreme Court opinions. 26 Bernard Schwartz, A Book of Legal Lists: The Best and Worst in American Law 48 (1997). His choices in order are as follows:INSERT TBL While the Schwartz list involves obviously important decisions, it measures “greatness,” which includes some normative judgment beyond mere significance, though Schwartz focuses on influence as a measure of greatness. However, Schwartz provides no explanation for the method he used to identify the greatest cases. 27Schwartz concedes that the list is “as personal as a sports writer’s choices for an all-star team.” Id. at 66. He stresses the value of an important Supreme Court opinion, suggesting that the “mind boggles at how different our system would be if these cases had not been decided as they were.” 28 Id. at 67.

The notion of importance is somewhat ambiguous and multidimensional (e.g., legal, political, social, etc.); a case might be politically quite significant but legally unimportant (perhaps Bush v. Gore). Some decisions, for instance, have a significant effect on public attitudes on political issues, independent of any legal consequence. 29 See generally Roy B. Flemming et al., One Voice Among Many: The Supreme Court’s Influence on Attentiveness to Issues in the United States, 1947–92, 41 Am. J. Pol. Sci. 1224, 1224 (1997) (noting that “politically significant” decisions affect “systemic attention to the underlying issues”). Our focus is on legal significance rather than political or societal significance.

A. Citation Measures for Important Cases

Citations to prior Supreme Court decisions are the primary source of authority for today’s opinions of the Court. 30 See Glenn A. Phelps & John B. Gates, The Myth of Jurisprudence: Interpretive Theory in the Constitutional Opinions of Justices Rehnquist and Brennan, 31 Santa Clara L. Rev. 567 (1991) (providing data showing that prior Court precedents are more commonly invoked than other authorities in the Court’s opinions). Reliance on prior opinions is the foundation of stare decisis, which is central to our law. Precedents are “viewed as the principal asset of a judicial system,” so that the higher their quality, “the better the judicial system may be said to be.” 31Jonathan R. Macey, The Internal and External Costs and Benefits of Stare Decisis, 65 Chi.-Kent L. Rev. 93, 106 (1989). It is difficult to assess the significance of an opinion in the abstract, as “the meaning and value of precedent depends on how subsequent Justices conceived it.” 32 Michael J. Gerhardt, The Power of Precedent 109 (2008). Citation counts have been used as a measure of “influence on the law.” 33Kosma, supra note 8, at 338. Judge Posner, for example, used citations to assess the significance of Justice Cardozo. 34 Richard A. Posner, Cardozo: A Study of Reputation 80–90 (1990). Hence, we use citations as a tool for estimating an opinion’s importance.

Our measure of the most legally important cases in the history of the Supreme Court depends upon the number and pattern of citations received by a case. Citations function as the “currency of the legal system,” so that their measure represents a central measure for the legal system. Our measures of legal importance include the number of citations received by a case both at the Supreme Court and lower court levels, plus a network measure of the “legal relevance” of each of the opinions, based on the connections (both direct and indirect) between it and other opinions in the network of citations at the Supreme Court level.

Frequency of citation is a reasonable standard for measuring case importance. Citations “set forth the authority on which a case rests.” 35Lawrence M. Friedman et al., State Supreme Courts: A Century of Style and Citation, 33 Stan. L. Rev. 773, 794 (1981). If a Supreme Court opinion is never cited, that suggests that its content is not useful in the resolution of subsequent litigation. Such an opinion could hardly be considered an important one. 36 See Richard A. Posner, Judges’ Writing Styles (and Do They Matter?), 62 U. Chi. L. Rev. 1421, 1424–25 (1995) (stating that “[e]ven a brilliant analysis of yesterday’s legal problems is unlikely to hold much current interest,” while the greater opinion “can be pulled out and made exemplary of law’s abiding concerns”). Conversely, if an opinion is frequently cited, that very fact suggests that it provides valuable governance or information.

There is considerable variance in the rate at which Supreme Court opinions are cited by later Supreme Courts. Of the 26,616 opinions released by the Supreme Court from 1791 to 2005, about 14% of them were never cited by a majority opinion of the Supreme Court, and an additional 10.9% were cited by the Court only once. 37These data were derived from Fowler et al., supra note 15. In addition, the average case received 7 citations by majority opinions of the Court over its life, with a standard deviation of 9.5 citations. A small number of cases, however, received a lot of citations, with, for instance, about 1% of cases receiving at least 65 citations. The cases with fewer citations (especially those with none) are plainly of lesser importance, as future Courts have found them to be largely irrelevant to their work. The Supreme Court, though, is only the tip of the judicial iceberg. Even if the Supreme Court rarely cites an opinion, it might still be legally very important if it is frequently used by lower courts, who decide the overwhelming majority of disputes. We incorporate lower courts in our analysis.

Citation rates for opinions are certainly influenced by the content of ensuing litigation. An opinion written in an area of the law that sees little litigation is less likely to be cited than one in a more litigated field, simply on grounds of relevance. This fact is relevant to case importance. If a legal question is so rare that it does not often arise in disputes, it probably is not an important one. 38A possible exception to this position would be an opinion that so clearly settled the law that disputes did not arise precisely because of the power and clarity of the opinion. This possibility is discussed below at note 63 and accompanying text.

The structure of the law is often characterized as a path-dependent system. 39 See generally Oona A. Hathaway, Path Dependence in the Law: The Course and Pattern of Legal Change in a Common Law System, 86 Iowa L. Rev. 601 (2001) (describing path-dependence theory and applying it to common law decision making). Opinions are to some degree dependent on earlier opinions that they cite. From an economic perspective, this path dependence represents an efficiency adaptation, as subsequent opinions follow earlier opinions because it is less costly to do so. 40 See id. at 606–09, 627–35. The procedure has other benefits as well, because subsequent judges can use the information provided by the earlier holding. Ronald Dworkin has analogized stare decisis to a chain novel, in which succeeding authors build upon what was written before, in hopes of producing the best overall story. 41This theory is discussed in Stefanie A. Lindquist & Frank B. Cross, Empirically Testing Dworkin’s Chain Novel Theory: Studying the Path of Precedent, 80 N.Y.U. L. Rev. 1156 (2005). In a chain novel, the importance of a particular chapter depends critically on the degree to which its foundation is used by the authors of later chapters. 42 See id. A character who appears in the second chapter but is never again mentioned has little importance in a novel.

Given this path-dependent structure of precedent, the importance of an opinion is associated with its subsequent frequency of use as a citation by later opinions. One would expect that “the repeated use of precedents reinforces their own significance.” 43 Id. at 1170. Michael Gerhardt has stated that the “more courts and other institutions approvingly cite precedents, the more their value increases.” 44 Gerhardt, supra note 32, at 192. Similarly, Justice Alito has declared that “when a precedent is reaffirmed, that strengthens the precedent,” 45 See Court in Transition: When a Precedent Is Reaffirmed, That Strengthens the Precedent, N.Y. Times, Jan. 11, 2006, at A26 (quoting Justice Alito’s confirmation hearing testimony). and Illinois chief justice Walter Schaefer has stated that, “[a]long with quality, quantity [of citation] too is significant,” as a “settled course of decision is more compelling than an isolated precedent . . . .” 46Walter V. Schaefer, Precedent and Policy, 34 U. Chi. L. Rev. 3, 11 (1966).

One important reason for the Court to rely on precedent is to grant greater legitimacy to its decisions. The Court is often criticized for activism—making ideological political decisions—and this perception harms its legitimacy. Justices themselves have written that “the Court’s legitimacy depends on making legally principled decisions” that rely on precedent. 47Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 866 (1992). Individuals consider Supreme Court decisions legitimate because of the perception that they are based on “case-relevant information” and not “political pressures and public opinion.” 48Tom R. Tyler & Gregory Mitchell, Legitimacy and the Empowerment of Discretionary Legal Authority: The United States Supreme Court and Abortion Rights, 43 Duke L.J. 703, 786 (1994). If the Court’s decisions were seen as political, it would become “more vulnerable to retaliation from the political branches.” 49Thomas M. Merrill, A Modest Proposal for a Political Court, 17 Harv. J.L. & Pub. Pol’y 137, 139 (1994). There is evidence for these propositions. A recent study used an experimental research design to show that attributes of Court opinions that connote the neutral and principled character of decision making influence individuals’ perceptions of those decisions. 50 See James R. Zink et al., Courting the Public: The Influence of Decision Attributes on Individuals’ Views of Court Opinions, 71 J. Pol. 909 (2009). In particular, opinions that overrule precedent (rather than follow it) and cases decided by minimum-winning coalitions (rather than unanimously) are generally held in lower regard by the public. 51 Id.

The reliance on precedent provides legitimacy for the Supreme Court’s opinions. Political scientists have argued that even if the Justices wanted to be uncontrolled policymakers, they would be constrained by legitimacy. 52 Lee Epstein & Jack Knight, The Choices Justices Make 45 (1998). They must “make accommodations over the interpretation of precedent because they believe that doing so enhances the probability that society will consider the resulting decision legitimate.” 53 Id. Ample evidence supports this position. 54Justice Stevens has declared that following precedent “obviously enhances the institutional strength of the judiciary.” John Paul Stevens, The Life Span of a Judge-Made Rule, 58 N.Y.U. L. Rev. 1, 2 (1983). Insofar as legitimacy is a concern, it is likely that relying on well-established precedents, often used by the Court, has greater value than relying on obscure precedents that have not been previously embraced. 55 Hansford & Spriggs, supra note 5, at 23 (contending that “for legitimacy reasons, the justices are more likely to rely on those precedents possessing greater legal weight” as reflected in their repeated citation).

In addition to providing external legitimacy to opinions, reliance on precedent may also be used by Justices as a means of providing greater authority to their own opinions. Greater fealty to precedent may give Justices “greater influence” and make them “more influential both on and off the Court.” 56 Gerhardt, supra note 32, at 368. One theory of the use of precedent suggests that it is a tool for judges to project power via their own opinions. Even the most willful judge would follow prior decisions, by this theory, in order to protect “the precedential significance of his own decisions.” 57Landes & Posner, supra note 8, at 273. This has been modeled via game theory. See Erin O’Hara, Social Constraint or Implicit Collusion?: Toward a Game Theoretic Analysis of Stare Decisis, 24 Seton Hall L. Rev. 736, 745–49 (1993) (explaining how judges agree to follow each other’s precedents to avoid nonproductive competition); Eric Rasmusen, Judicial Legitimacy as a Repeated Game, 10 J.L. Econ. & Org. 63, 67 (1994) (arguing that stare decisis enhances judges’ power vis-à-vis future judges). While Justices may well independently value decision making according to stare decisis, 58 See, e.g., Lawrence Baum, The Puzzle of Judicial Behavior 61 (1997) (“[I]t pleases judges to carry out what they conceive as the judge’s role.”); Tracey E. George, From Judge to Justice: Social Background Theory and the Supreme Court, 86 N.C. L. Rev. 1333, 1355–57 (2008) (discussing this role theory and its influence on judges). Baum argues that ideological preferences are constrained “because decision makers want to reach results that they can accept as correct.” Baum, supra, at 65. the legitimacy and power-protection theories add reasons for the power of precedent in the Court and the consequent path dependence of precedent.

This path-dependency effect of precedent has been clearly demonstrated empirically. An empirical analysis examined the “vitality” of precedents, meaning the relative frequency with which they were “positively” versus “negatively” interpreted in subsequent opinions of the Court. The authors found, for instance, that an opinion was more likely to be positively interpreted in a given year if it had a higher level of legal vitality, even after controlling for a whole host of additional variables. 59 See generally Hansford & Spriggs, supra note 5 (analyzing the Supreme Court’s use of precedent from 1946 to 2000 and showing that the interplay of the Justices’ ideological orientations and the current legal authority of a precedent combined to influence how the Court legally interpreted the precedent). Other factors also mattered, most notably the ideological distance of the Court from the precedent, but precedential vitality was consistently a significant determinant of subsequent legal interpretations.

There is “often decisional leeway in determining whether a precedent governs a case.” 60 Id. at 22. Many of the cases cited in briefs by litigants do not appear in the Court’s subsequent opinion. 61 See Frank B. Cross, Chief Justice Roberts and Precedent: A Preliminary Study, 86 N.C. L. Rev. 1251, 1274 (2008) (surveying cases decided by Chief Justice Roberts and finding that the “opinions cited, on average, less than half the cases found in both the petitioners’ and respondents’ briefs”); James F. Spriggs II & Thomas G. Hansford, The U.S. Supreme Court’s Incorporation and Interpretation of Precedent, 36 Law & Soc’y Rev. 139 (2002) (modeling the conditions under which the Court positively or negatively interprets the precedents cited in litigant and amici briefs). A citation measure reflects the evaluations of sitting Justices and judges about the importance of precedents for the disputes they resolve. While it may not reflect all aspects of an opinion’s importance, a citation metric measures the importance of those opinions within the law itself, surely an important standard. In Cardozo’s words, these are the cases that “count for the future.” 62 Benjamin N. Cardozo, The Nature of the Judicial Process 164–65 (1921).

If citations are a measure for case importance, one must decide: citations by whom? The Supreme Court is the ultimate arbiter of American law, so Supreme Court citations are surely relevant. The Supreme Court sets the ground rules for all decisions and has an obvious influence on lower courts. However, it is those lower courts that resolve most of the disputes in our legal system. Consequently, citations by lower courts are also relevant criteria for any measure of case importance. The comparative importance of particular levels of our judiciary is a debatable one, and we will report the results for different levels, leaving it to the reader to evaluate their relative significance.

B. Validity of Citation Use

While citations are an obvious measure of the legal significance of a case in the corpus of stare decisis, they might be disputed as a misleading measure of case importance. Use of citations as a measure of significance is subject to a variety of challenges, which we address in this section. While no measure is perfect, the citation metric is widely used and valid for measuring the significance of cases.

The first and most common criticism of citation usage is that it fails to capture dispositive rulings that conclusively resolve legal issues. Some decisions may settle the law in a given area, setting out such clear directions that future cases in its ambit do not even arise. Such a decision could be one of very great practical importance, defining the law for primary actors, who follow it faithfully. However, such a case would appear insignificant in any measure based on citations because the lack of subsequent litigation would correspond to a lack of citations. 63 See Kosma, supra note 8, at 339 (noting the criticism that “a precedent may set such a clear legal standard that subsequent cases settle rather than go to trial and appeal, resulting in few citations to the case in later opinions”).

The facile answer to this criticism is that we are measuring for legal significance, not overall political or societal significance. 64 See, e.g., Landes et al., supra note 8, at 274 (noting that citation counts could understate the significance of a precedent “that is so effective in clearing up an unsettled area of law that future disputes settle rather than go to trial”). The authors suggested that such cases were “rare,” however. Id. While the “settled case” phenomenon is theoretically problematic for any citation measure, its existence is questionable. Under the operation of precedent at the Supreme Court, a decision rarely if ever truly settles the law in a fashion that halts future litigation. Even if a case did so, it might still assume importance in citations as precedent for other legal matters.

Under the strictest concept of stare decisis, a decision only resolves the dispute on the precise facts before the Court and is debatably analogous to other groups of facts. In practice, the language of opinions may functionally resolve many other circumstances that differ from those before the Court. The language choices in the opinion largely control how far beyond the instant facts its power stretches. For example, if the court sets a rule, it ostensibly governs many differing factual circumstances, but if it sets a standard, it leaves the resolution of those cases unclear. 65Much has been written on the different implications of rules and standards. See, e.g., Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 Duke L.J. 557 (1992) (modeling this effect on primary behavior); Kathleen Sullivan, The Justices of Rules and Standards, 106 Harv. L. Rev. 22 (1992) (discussing the Supreme Court’s use of rules and standards). Therefore, a clear rule might be said to settle a large number of cases and might appear falsely weak in a salience measure based on citations.

The notion that any opinion, including one with a clear rule, settles the law so that it receives few future citations misunderstands the operation of precedent. In their precedential opinions, judges “set boundaries in fact spaces.” 66Charles M. Cameron, New Avenues for Modeling Judicial Politics 45 (Oct. 4, 1993) (unpublished manuscript), available at http://www.princeton.edu/~ccameron/NewAvenues.pdf. They define the sets of cases governed by the opinion’s holdings. A bright-line rule may have the functional effect of setting broader boundaries than other types of opinions. But in each case, even with the bright-line rule, there will inevitably be legal questions at the margins. While the core of the rule may be settled and may not yield litigation, the marginal applications will. Moreover, given the nature of stare decisis, the rule’s holding will surely be cited as an analogy to other cases, well outside the core holding of the original opinion.

The creation of a rule in an attempt to settle the law also invites a series of future legal challenges even within its apparent core. Consider Miranda v. Arizona. 67384 U.S. 436 (1966). In Miranda, the Court set down an unusually clear requirement that statements by a criminal defendant in police custody would be admissible only if the defendant were informed of his or her rights in a very specific way. Yet the opinion left open many questions to be clarified by future decisions. What is the definition of custody or interrogation? 68 See, e.g., Berkemer v. McCarty, 468 U.S. 420 (1984); Rhode Island v. Innis, 446 U.S. 291 (1980); Brewer v. Williams, 430 U.S. 387 (1977); Orozco v. Texas, 394 U.S. 324 (1969). What about spontaneous statements made by a defendant, unprovoked by questioning? What if the police used good faith? 69 See United States v. Leon, 468 U.S. 897 (1984). What if public safety requires prompt police action? However much the Court might wish to conclusively resolve a legal question, leaving no possible future disputes, the nature of the case-or-controversy requirement and opinion writing means that this is virtually impossible.

It is difficult to identify a single case that so settled the law that it rendered future citations unnecessary. Some cases are claimed to be “superprecedents,” settling the state of the law 70Superprecedent is a term that has been recently used to describe a Supreme Court opinion that is so entrenched in our law and politics that it is beyond challenge. See Michael J. Gerhardt, Super Precedent, 90 Minn. L. Rev. 1204, 1207 (2006). The term was coined to refer to an opinion that “would be so effective in defining the requirements of the law that it prevents legal disputes from arising in the first place, or, if they do arise, induces them to be settled without litigation.” Michael Sinclair, Precedent, Super-Precedent, 14 Geo. Mason L. Rev. 363, 364 (2007). such that they “might never be cited in an appellate opinion yet have greater precedential significance than [the] most frequently cited cases.” 71Sinclair, supra note 70, at 364. Yet the cases known to be superprecedents have received numerous citations in later opinions. Marbury v. Madison, 725 U.S. (1 Cranch) 137 (1803). for example, settled the issue of constitutional judicial review quite conclusively, but the case has nevertheless received numerous subsequent citations. 73 Marbury is a leading example of a superprecedent. See Gerhardt, supra note 70, at 1207–08; Sinclair, supra note 70, at 364. As of this writing, Marbury has over 17,000 total citations in the Westlaw database. It has the fourth most citations of any case in Supreme Court history, garnering 209 citations by the U.S. Supreme Court through 2005. See infra Table 1. Landes and Posner affirmed this conclusion about the settled case theory, writing:

[S]uch cases are probably rare. If a case is highly specific, it will hardly qualify as a “superprecedent”; by definition it will control only those infrequent cases that present virtually identical facts to those of the case in which it was originally announced. If it is highly general, and therefore more likely to be an important precedent, it is unlikely to decide—so clearly as to prevent disputes or litigation from arising—the specific form of the question presented in subsequent cases. 74Landes & Posner, supra note 8, at 251.

Some cases may settle some legal issues, but if important, they are still relevant citations for issues on the margin or by analogy to different circumstances. “Even were such an impressive settlement-generating opinion to appear, courts would likely recognize its influence and cite it frequently (perhaps using similar reasoning in a different context) rather than ignore it . . . .” 75Kosma, supra note 8, at 339.

The notion of settled law also embraces a naïve vision of stare decisis and its control over courts. In light of considerable empirical research, the formalistic vision of judges reliably adhering to precedent is no longer a viable one. 76Research on the attitudinal model of decision making has shown systematic patterns of voting by Supreme Court Justices that appear to parallel their political ideological preferences. See Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model (1993) [hereinafter Segal & Spaeth, Attitudinal Model]; Segal & Spaeth, Attitudinal Model Revisited, supra note 5. “Scores” of additional studies have confirmed this effect. Emerson H. Tiller & Frank B. Cross, What Is Legal Doctrine?, 100 Nw. U. L. Rev. 517, 523 (2006). Justices are influenced to some degree by their personal ideological attitudes in their decisions, and they will try to avoid the governance of precedents that they find unappealing. 77Many have expressed skepticism about the influence of stare decisis on the Justices. Henry Monaghan has stated that precedent is merely a “mask hiding other considerations.” Henry Paul Monaghan, Stare Decisis and Constitutional Adjudication, 88 Colum. L. Rev. 723, 743 (1988). Judge Wald suggested that judges simply distinguish away unappealing precedents and “follow those precedents which they like best.” Patricia M. Wald, Changing Course: The Use of Precedent in the District of Columbia Circuit, 34 Clev. St. L. Rev. 477, 481 (1986). Segal and Spaeth found that Justices who dissented from an original opinion did not respect its power, but continued to dissent from future opinions relying on the original opinion. Harold J. Spaeth & Jeffrey A. Segal, Majority Rule or Minority Will (1999); see also Hansford & Spriggs, supra note 5, at 3 (showing that the Justices are more likely to negatively interpret (e.g., overrule) precedent if they are ideologically opposed to it). Litigants will probe even settled precedents, seeking exceptions to their controlling power or expansions of that power, and ideologically influenced Justices will sometimes respond.

In addition to the theoretical dubiousness of the Supreme Court “settling” the law conclusively, it appears that Justices do not even try to fully settle the law. Rather, through a signaling process, they render decisions on particular cases in part to provoke access to additional cases that can be used to develop the ruling in the original case. 78 See Vanessa A. Baird, The Effect of Politically Salient Decisions on the U.S. Supreme Court’s Agenda, 66 J. Pol. 755 (2004). Through its selective certiorari decisions, the Court sets the agenda for change. Research has found that a salient Supreme Court decision produces an increase in circuit court decisions and amicus briefs and increases the Supreme Court’s agenda of potential cases to build upon the original decision’s precedent. 79 Id.

Although commentators have suggested that the settled case effect might bias our instruments, 80 See Kosma, supra note 8, at 339–40; Landes & Posner, supra note 8, at 274–76. none have suggested a specific example of such an opinion. A logical possibility might be the Slaughter-House Cases. 8183 U.S. (16 Wall.) 36 (1873). This group of cases has been seriously criticized for neutralizing the Privileges and Immunities Clause of the Constitution 82 See, e.g., Charles L. Black, Jr., A New Birth of Freedom: Human Rights, Named & Unnamed 55 (1999) (stating that the decision was “probably the worst holding, in its effect on human rights, ever uttered by the Supreme Court”); Lawrence H. Tribe, American Constitutional Law 1303–11 (3d ed. 2000) (discussing the case and criticism); Steven G. Calabresi & Sarah E. Agudo, Individual Rights Under State Constitutions When the Fourteenth Amendment Was Ratified in 1868: What Rights Are Deeply Rooted in American History and Tradition?, 87 Tex. L. Rev. 7, 73 (2008) (“The [opinion] construed the Fourteenth Amendment in an absurdly narrow way, which has been roundly excoriated by courts and scholars.”). and is regarded as one of the most important Court opinions. 83 See, e.g., Steven J. Eagle, Property Tests, Due Process Tests and Regulatory Takings Jurisprudence, 2007 BYU L. Rev. 899, 904 (referring to Slaughter-House as “one of the most important decisions of the nineteenth century”).

By definitively closing off a channel of potential constitutional litigation, the opinion was very important to the law. However, one might expect the importance of the opinion to be obscured by citation studies—its significance would lie in cases that were not litigated because the Slaughter-House Cases created a clear rule. The opinion for these cases thus presents a candidate for the settled law hypothesis. Raw Supreme Court citations to the opinion over time are set out in Figure 1. In Figure 2, we also present a more sophisticated measure of the importance of this case based on a network analytic technique that uses both the direct and indirect relations among cases to determine a “legal relevance” score (i.e., how central a case is in the overall network of Supreme Court jurisprudence). This score is created by counting the number of citations to a case and weighting each citation by how “outwardly legally relevant” the case is (which is a function of the number of precedents cited by the citing case, as weighted by how “inwardly legally relevant” each cited case is). The score is then represented as a percentile; for example, a score of .85 means a case is in the 85th percentile of legal relevance for all cases having been decided by the Court. 84For a discussion of the advantages of legal relevance scores for measuring case importance, see Fowler et al., supra note 15; infra Part II.B.

Figure 1: Cumulative Number of Supreme Court Citations to the Slaughter-House Cases

cross-spriggs-fig1

Figure 2: Legal Relevance Score for the Slaughter-House Cases

cross-spriggs-fig2

Neither graph depicts settled law. As evident in Figure 1, Slaughter-House has received a considerable number of citations over its life, and even more importantly, continues to receive citations in the contemporary time period. In addition, the authority score for the Slaughter-House Cases, while increasing substantially in the early years after its release, remains at about the 93rd percentile of all Supreme Court majority opinions as of 2005. These two Supreme Court measures thus capture the considerable significance of the opinion, even though it might be considered a case that settled an important area of the law.

The settled law bias therefore may not seriously bias citation studies. 85 See Kosma, supra note 8, at 340 (observing that if the number of such cases was small and not unusually distributed, any distortion of results would be “of minor significance”). Moreover, one of our measures discussed below, the authority scores, may counteract any settled law bias to some extent. Throughout this Article though, we will remain alert to the possibility of a bias in the results.

There may be an analogous problem in that an unusually ambiguous opinion might spawn a great deal of litigation, not because of its importance, but because of its lack of clarity. 86 See id. at 339 (“Such an opinion, with marginal or questionable influence, may be overrepresented when counting citations.”). The uncertainty created by such a case would not associate with its importance. When this occurs though, the burst of litigation and associated citations should be brief. The ambiguous opinion, being relatively unhelpful for the resolution of subsequent cases, should be supplanted by a more useful opinion. 87 See id. at 339–40. Kosma notes that if the pattern of citations continues, the original case may have been “more useful than one might have guessed initially.” Id. at 340. Thus, any positive effect from an initial ambiguous opinion should dissipate as it is interpreted.

Some legal issues are settled. Michael Gerhardt notes that “[r]econsideration of many cases is simply off the table.” 88 Gerhardt, supra note 32, at 45. Insofar as “the Court’s precedents frame its choices of which constitutional matters not to hear,” citations may be a poor measure of significance. 89 Id. at 153. As a leading example of such cases, Gerhardt points to the fact that “the Court no longer considers incorporation questions—whether the liberty component of the Fourteenth Amendment due process clause applies the Bill of Rights, in whole or in part, to the states.” 90 Id. at 45. This provides an opportunity to test whether the settled case phenomenon undermines the meaning of citation studies. We will reveal below that such an incorporation opinion, now “settled law,” was the single most important Court decision by one measure. 91 See infra Table 4. Although the opinion settled one narrow legal question, it remained quite important in the network of Supreme Court precedent.

While the settled case phenomenon is the most prominent challenge to use of citations, others are sometimes made. Not every citation is truly a useful precedent for a subsequent decision. Some citations may be trivial or unavoidable. 92Frank B. Cross et al., Warren Court Precedents in the Rehnquist Court, 24 Const. Comment. 3, 4–5 (2007). Some cases may get cited simply because the later court believes they were wrongly decided. 93 Id.; see also Frank B. Cross & Stefanie A. Lindquist, Judging the Judges, 58 Duke L.J. 1383, 1391 (2009) (noting the need to consider negative citations). This in itself is significant though, as critical citations are “also a gauge of influence since it is easier to ignore an unimportant decision than to spell out reasons for not following it.” 94Landes et al., supra note 8, at 273. Judge Posner has noted that even negative citations are “motivated by the authority of the previous case.” Posner, supra note 4, at 385. Moreover, the overruling of prior precedents is quite rare. 95In a typical decade, the Court overrules less than 0.002% of its previous opinions. Jeffrey A. Segal et al., The Supreme Court in the American Legal System 316 (2005). Of the 6,363 Supreme Court cases decided between 1946 and 1999, the Court had overruled only 107 of them by 2001. Hansford & Spriggs, supra note 5, at 81. When a precedent is distinguished or narrowed, such purportedly negative treatment is still some testimony to the influence of that precedent. A study of circuit court citations found that the presence of negative citations did not bias its results. 96 See Choi & Gulati, supra note 8, at 56–57.

The settled case phenomenon and other criticisms thus do not delegitimize citations as a measure of the importance of Supreme Court opinions. Indeed, citations have been widely used for this purpose, 97 See sources cited supra note 8. and for some time. 98A seminal study in this regard is S. Sidney Ulmer, The Use of Power in the Supreme Court: The Opinion Assignments of Earl Warren, 1953-1960, 19 J. Pub. L. 49, 55 (1970) (using citations in the five years following a decision as a measure of importance). An important early study evaluated individual Justices using citations as a measure of “significant influence over the subsequent development of legal doctrines.” 99Kosma, supra note 8, at 333–34. The “measurement of precedential significance by counting citations may prove to hold the key to the problem of evaluating judicial output.” 100Landes & Posner, supra note 8, at 293.

The usefulness of citation analysis has been shown in existing research to have criterion validity. One such measure, the “legal relevance” score discussed above, has been assessed against other measures of case importance. 101Fowler et al., supra note 15. The study found that legal relevance score was a better predictor of future citations in the Supreme Court than were other scales (such as presence on Oxford or Congressional Quarterly lists, appearance on the New York Times front page, or even the raw number of citations to a case). 102 Id. at 342.

A historical use of network analysis on Supreme Court citations provides further validation. One study examined the mix of Supreme Court citation patterns in the Lochner era as contrasted with those found in cases after the 1937 “switch in time.” 103E.A. Leicht et al., Large-Scale Structure of Time Evolving Citation Networks, 59 Eur. Physical J. B 75, 83 (2007). The authors found a striking change in citation patterns associated with this “switch,” concluding that the citation analysis tool reflected a fundamental change in the opinions’ legal thinking. 104 Id. This research demonstrates that use of citations is associated with opinion content and that the importance of citations is associated with the state of the law at any given time. An early study similarly argued that citation choices had “a profound effect on the way the law grows and the shape legal doctrines take.” 105Merryman, supra note 8, at 615.

Citations remain an inexact measure of opinions and their importance. There is surely some randomness or “luck” associated with the receipt of citations. 106 See Daniel A. Farber, Supreme Court Selection and Measures of Past Judicial Performance, 32 Fla. St. U. L. Rev. 1175, 1178 (2005) (suggesting that more citations to an opinion may be a feature of “just plain luck”). This does not defeat the value of research on citations, though. 107 See Posner, supra note 4, at 387 (noting that although quantitative studies of citations inevitably contain much random “noise,” this does not “disable useful statistical analysis”). While citation choices are somewhat “idiosyncratic” in particular cases, Landes and Posner have observed that the “extensive research and writing that lawyers, judges, and law clerks devote to discovering, marshalling, enumerating, and explaining precedents are not costless undertakings, and would not be undertaken if precedent did not enter systematically into the decision of cases.” 108Landes & Posner, supra note 8, at 252. Put otherwise, we have no doubt that, as is the case for all proxy measures, our measures of legal significance contain error. The operative questions, however, are the degree to which this error is random and the overall signal-to-noise ratio in the measures themselves.

Citations are a useful tool for assessing opinions retrospectively. There is no one “correct” citation measure, though. We employ the Supreme Court citation count, the lower court citation count, and the Supreme Court legal relevance scores as separate tools. This enables some cross-check on the findings of any one measure and potentially isolates different types of effects. The following Part describes our methodology for quantifying the precedential significance of opinions from the Court.

II. Quantifying the Most Important Supreme Court Decisions

It has long been recognized that “all decisions are not of equivalent value” to the Court. 109Schaefer, supra note 46, at 7. Some opinions simply involve more important legal or societal issues, at least for purposes of future cases that can cite them. The nature of the opinion itself may have an influence on whether and the degree to which it is cited by later holdings. Yet the presence of those citations is surely a signal to the importance of an opinion. Justice Jackson wrote that the “first essential of a lasting precedent is that the court or the majority that promulgates it be fully committed to its principle.” 110Robert H. Jackson, Decisional Law and Stare Decisis, 30 A.B.A. J. 334, 335 (1944).

While it is no surprise that opinions are not equivalent, existing empirical research has largely treated them as if they were. Opinions are judged simply by binary outcomes, based upon who won or lost the case. These outcomes may be categorized as liberal or conservative, but they treat all liberal (or conservative) outcomes as if they were equal, though this is plainly not the case. In this Article, we seek to differentiate among judicial opinions, based on the significance of the opinion according to its subsequent citations.

A. Data

Our full data set includes all opinions of the United States Supreme Court, including orally argued per curiam decisions, released between 1791 and 2005 (for a total of 26,681 Court opinions). 111 See Fowler et al., supra note 15, for how this list of Supreme Court opinions was created. For each of these cases, we have data for the total number of subsequent opinions of the Supreme Court (majority, concurring, or dissenting) that cited each of these cases, as found using LexisNexis and Shepard’s Citations Service, through calendar year 2005. 112These data are available from Ryan C. Black & James F. Spriggs II, An Empirical Analysis of the Length of U.S. Supreme Court Opinions, 45 Hous. L. Rev. 621 (2008), and Fowler et al., supra note 15. The average Supreme Court case is cited about 9 times over its “life”, 113Fowler et al., supra note 15, at 328 n.12. By “life,” we mean the period of time from when it was decided through 2005. with an interquartile range of 2 and 11 citations. We also have similar data for the total number of circuit court and district court majority opinions that cited each of these opinions of the Supreme Court. Supreme Court opinions are cited by appellate and district court majority opinions an average of 55 and 66 times over their lives, respectively. The interquartile range for the number of citations by appellate is 4 and 45, while this range is 1 and 34 citations for district courts. A small fraction of Court decisions receive a large number of citations, as the top 1% of Court cases are cited by at least 65, 644, and 919 Supreme Court, appellate court, and district court opinions, respectively.

Some limited research has already been done on this issue. One study identified the most legally important cases based on both the citations contained in the opinion and the citations eventually received by the opinion. 114Fowler et al., supra note 15. Another study used network analysis to identify the most important cases through various citation measures. 115 See Seth J. Chandler, The Network Structure of Supreme Court Jurisprudence, 10 Mathematica J. 501 (2007). We build upon these earlier studies in this research, by using additional measures and studying the characteristics of the most important opinions.

B. Leading Cases by Different Metrics

In this section we identify the most important Supreme Court decisions on different citation metrics. The first is simply the number of subsequent citations at the Supreme Court itself. This raw citation count is the conventional measure used in prior research. 116 See sources cited supra note 8. An opinion that received no citations would not be influential in the law. The greater the number of citations received by an opinion is some testimony to its significance to the Court. Table 1 sets out the list of top cases by simple number of citations the opinion has received.

Table 1:

Top 25 Cases by Supreme Court Citation Numbers

Case

Citation Numbers

  • McCulloch v. Maryland (1819)

355

  • Gibbons v. Ogden (1824)

273

  • Boyd v. United States (1886)

218

  • Marbury v. Madison (1803)

209

  • Osborn v. President (1824)

206

  • Miranda v. Arizona (1966)

196

  • Ashwander v. Tennessee Valley Authority (1936)

189

  • Cantwell v. Connecticut (1940)

186

  • Erie Railroad Co. v. Tompkins (1938)

185

  • Cohens v. Virginia (1821)

174

  • Yick Wo v. Hopkins (1886)

166

  • Ex parte Young (1908)

166

  • NAACP v. Button (1963)

162

  • Gideon v. Wainwright (1963)

160

  • Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc. (1984)

160

  • Brown v. Maryland (1827)

156

  • NAACP v. Alabama (1958)

155

  • New York Times Co. v. Sullivan (1964)

152

  • Mapp v. Ohio (1961)

148

  • Thornhill v. Alabama (1940)

147

  • Minnesota Rate Cases (1913)

147

  • Schneider v. State (1939)

146

  • Cooley v. Board of Wardens (1852)

144

  • Brown v. Board of Education (1954)

144

  • Johnson v. Zerbst (1938)

143

The top of the list is dominated by older cases, which have had more opportunities to be cited, given their age. The top four cases include the classic warhorses of the early Court (McCulloch, Gibbons, and Marbury), which were also in the top four opinions of Schwartz’s subjective list. 117 See Schwartz, supra note 26. The very high ranking of Boyd v. United States 118116 U.S. 616 (1886). might be deemed a surprise. Boyd did make the Oxford list of decisions 119 The Oxford Guide, supra note 10, at 30. but was not mentioned in Lucas A. Powe’s recent comprehensive history of the Court. 120 Lucas A. Powe, Jr., The Supreme Court and the American Elite, 1789-2008 (2009). However, Justice Black called the opinion “among the greatest constitutional decisions.” 121Schmerber v. California, 384 U.S. 757, 776 (1966) (Black, J., dissenting). Justice Brandeis said that Boyd was a “case that will be remembered as long as civil liberty lives in the United States.” Olmstead v. United States, 277 U.S. 438, 474 (1928) (Brandeis, J., dissenting). Being the origin of the exclusionary rule, it readily merits a high standing.

The highest-ranking modern case is Miranda. 122384 U.S. 436 (1966).Gideon, Chevron, Mapp, Brown, and other well-known post-World War II decisions also appear on the list. Thurgood Marshall described Thornhill v. Alabama 123310 U.S. 88 (1940). as “one of the most important decisions of the Supreme Court,” 124Thurgood Marshall, Mr. Justice Murphy and Civil Rights, 48 Mich. L. Rev. 745, 748 (1950). and it appears on our list. This gives some facial validity to the use of this metric. It seems to provide at least a rough approximation of cases that are generally considered to be significant. Citation numbers at the Supreme Court appear to be a reasonable operationalization of case importance.

Looking at the Supreme Court, however, may not be the best guide to case importance. Its decisions are but the tip of the iceberg. While the Supreme Court decides fewer than one hundred cases per year, the lower federal courts resolve thousands of disputes. The circuit courts “are largely left to themselves” to develop “legal rules in unsettled areas of law.” 125 David E. Klein, Making Law in the United States Courts of Appeals 51 (2002).

Perhaps the circuit courts are the crucial level of the federal judiciary. Because the Supreme Court issues so few decisions in a given year, it cannot address most of the legal topics litigated annually. The circuit courts represent the “court of last resort” for most. 126 Frank B. Cross, Decision Making in the U.S. Courts of Appeals 2 (2007). The circuit courts defer to factual findings, so their opinions explicate the law, not simply individual case facts. 127While district courts offer more opinions than do the circuit courts, their rulings are typically “heavily fact based and jurisdictionally limited in effect, and they do not set the significant legal precedents that make up the law.” Id. These are the courts that fundamentally “create U.S. law” and “play by far the greatest legal policymaking role in the United States judicial system.” 128 Id. Yet the circuit courts rely on Supreme Court opinions in making their decisions. Therefore, circuit court citations offer an important tool for measuring the importance of Supreme Court opinions. Table 2 sets out the most cited Supreme Court opinions at the circuit court level.

Table 2:

Top 25 Cases by Circuit Court Citation Numbers

Case

Citation Numbers

  • Strickland v. Washington (1984)

8,827

  • Anderson v. Liberty Lobby, Inc. (1986)

8,194

  • Anders v. California (1967)

6,748

  • Jackson v. Virginia (1979)

6,064

  • McDonnell Douglas Corp. v. Green (1973)

5,531

  • Glasser v. United States (1942)

5,489

  • Apprendi v. New Jersey (2000)

5,446

  • Miranda v. Arizona (1966)

4,509

  • Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc. (1984)

4,096

  • United States v. Olano (1993)

4,069

  • Brady v. Maryland (1963)

3,957

  • Erie Railroad Co. v. Tompkins (1938)

3,707

  • Anderson v. Bessemer City (1985)

3,669

  • Harlow v. Fitzgerald (1982)

3,539

  • United States v. Booker (2005)

3,518

  • Cohen v. Beneficial Industrial Loan Corp. (1949)

3,483

  • Universal Camera Corp. v. NLRB (1951)

3,393

  • Terry v. Ohio (1968)

3,334

  • Slack v. McDaniel (2000)

3,205

  • Texas Department of Community Affairs v. Burdine (1981)

3,072

  • United States v. U.S. Gypsum Co. (1948)

3,013

  • Chapman v. California (1967)

2,918

  • Bivens v. Six Unknown Agents (1971)

2,896

  • Matsushita v. Zenith (1986)

2,889

  • INS v. Elias-Zacarias (1992)

2,843

While the circuit court list contains several cases that are widely regarded as being seminal (e.g., Miranda, Chevron, Erie) and that also appear on the Supreme Court rankings, this list also contains cases that are less prominent. The top case on the list, Strickland v. Washington, dealt with the ability to obtain a writ of habeas corpus due to the ineffectiveness of counsel at trial. 129466 U.S. 668 (1984). The same issue was also the basis for the third and fourth cases on the list. 130The third case on the list, Anders v. California, 386 U.S. 738 (1967), deals with habeas corpus standards, as does the fourth case, Jackson v. Virginia, 443 U.S. 307 (1979). None of these opinions appear on the Oxford list or the Congressional Quarterly list of important decisions, and they were never covered on the front page of the New York Times. 131 See Biskupic & Witt, supra note 11; The Oxford Guide, supra note 10. The large number of citations received by these cases appears to be attributable to the frequency of prisoner petitions for relief, often brought on a pro se basis. The results imply that habeas corpus is by far the most important subject for Supreme Court decisions, which seems questionable.

The second case, Anderson v. Liberty Lobby, Inc., involved the proper standards for granting summary judgment, a common procedural tool. 132477 U.S. 242 (1986). The fifth case, McDonnell Douglas Corp. v. Green, was the seminal holding on the burden of proof under Title VII, a commonly litigated provision. 133411 U.S. 792 (1973). These cases rank high on the list because of the frequency with which these legal issues are adjudicated. While this might provide some testimony to their importance, the citations may be attributed simply to the fact that these decisions are the governing Supreme Court standard for the pertinent area of law, not due to any specific characteristics of the Court’s opinion itself. 134 See Thomas A. Smith, The Web of Law, 44 San Diego L. Rev. 309, 347 (2007) (noting that the summary judgment opinions are “cited so often because the federal courts . . . handle many motions for summary judgment”). Smith argues that it does “not follow” that these opinions would necessarily be the “most important” or “most authoritative” opinions. However, he also notes that it would be hard to deny that they provide “law that is at the core of what federal courts do.” Id.Any leading Supreme Court decision on these questions could well be highly cited. 135This claim may be a bit too strong because a truly unworkable decision on these topics would prove unhelpful to lower courts and presumably result in another Court decision creating a better precedent. However, even a moderately good decision, workable at the lower court level, would not necessarily need overruling or revision. This reflects a possible defect in this metric as a measure of opinion quality or significance. A case that is cited often for routine matters may be less significant than one cited less often but that is outcome determinative in the subsequent opinion.

The seventh case, Apprendi v. New Jersey, may reflect a slightly different effect that could be distinguished from case significance. This opinion was highly disruptive of existing law, holding that factual determinations underlying criminal sentencing require a jury’s determination. 136530 U.S. 466 (2000). See Stephanos Bibas, Apprendi in the States: The Virtues of Federalism as a Structural Limit on Errors, 94 J. Crim. L. & Criminology 1, 11 (2003) (noting the decision’s “disruptive” effect in upsetting settled criminal procedures by potentially requiring the costly “reopening [of] . . . hundreds of thousands of cases”). As a result, large numbers of existing sentencing decisions became subject to challenge. Perhaps such disruption is a fair measure of significance, but it too is colored simply by the very large number of criminal sentencing cases. Once its scope is settled, it may become much less significant in the long run.

The citation rates for circuit court cases appear to be substantially a feature of lower court litigation patterns and procedural rules. This is surely a measure of the relevance of the opinions to practice at the circuit court level. It is not necessarily a perfect measure of the importance of opinions, though, as general procedural standards are highly important at this level, and not all their decisions are of equivalent importance.

The role of precedents in the district courts should also be evaluated. District courts obviously decide more cases than any other level of the federal judiciary. While the decisions of district courts are based substantially upon the case facts, the courts must apply the law to those facts, and Supreme Court opinions constitute an important source of that law. 137District courts are bound to apply the law of the circuit within which they operate. Hence, their citations may be influenced by the filter of that circuit court. Hence, the citation rates of district courts have significance, and Table 3 sets out the Supreme Court opinions most cited by federal district courts.

Table 3:

Top 25 Cases by District Court Citation Numbers

Case

Citation Numbers

  • Anderson v. Liberty Lobby, Inc. (1986)

65,629

  • Matsushita v. Zenith (1986)

29,539

  • Conley v. Gibson (1957)

23,467

  • McDonnell Douglas Corp. v. Green (1973)

15,336

  • Celotex Corp. v. Catrett (1986)

11,661

  • Monell v. Department of Social Services (1978)

9,579

  • Adickes v. S. H. Kress & Co. (1970)

9,397

  • Scheuer v. Rhodes (1974)

8,854

  • Texas Department of Community Affairs v. Burdine (1981)

8,564

  • Strickland v. Washington (1984)

8,282

  • United Mine Workers of America v. Gibbs (1966)

8,056

  • Richardson v. Perales (1971)

7,743

  • Harlow v. Fitzgerald (1982)

7,183

  • Thomas v. Arn (1985)

7,096

  • Haines v. Kerner (1972)

6,950

  • Hishon v. King & Spalding (1984)

6,743

  • Erie Railroad Co. v. Tompkins (1938)

6,575

  • International Shoe Co. v. Washington (1945)

6,310

  • Estelle v. Gamble (1976)

5,696

  • United States v. Diebold (1962)

5,304

  • St. Mary’s Honor Center v. Hicks (1993)

5,116

  • Foman v. Davis (1962)

4,965

  • Klaxon Co. v. Stentor Electric Manufacturing Co. (1941)

4,791

  • Neitzke v. Williams (1989)

4,386

  • Board of Regents v. Roth (1972)

4,301

The district court list shares characteristics with the circuit court list. The leading two cases deal with the standards for summary judgment, 138Anderson v. Liberty Lobby, Inc., 477 U.S. 242 (1986); Matsushita Electric Indus. Co. v. Zenith Radio Corp., 475 U.S. 574 (1986). as does the fifth case. 139Celotex Corp. v. Catrett, 477 U.S. 317 (1986). The McDonnell Douglas decision on Title VII standards is fourth, 140411 U.S. 792 (1973). and a subsequent opinion on burdens of proof in these actions is ninth. 141Texas Dep’t of Cmty. Affairs v. Burdine, 450 U.S. 248 (1981). Other opinions on the list were important for governing significant areas of the law that arise before district courts, such as the standards for § 1983 actions 142 Monell v. Dep’t of Soc. Servs., 436 U.S. 658 (1978); Adickes v. S. H. Kress & Co., 398 U.S. 144 (1970). and sovereign immunity. 143Scheuer v. Rhodes, 416 U.S. 232 (1974).

The list of top cases cited by the Supreme Court conforms more closely to popular views on case importance than do the circuit court or district court lists. In this sense, the lower court lists may lack what is called facial validity. But the popular views of opinion importance may themselves be derivative of the Supreme Court’s assessment of importance. The lower courts remain a valuable measure of the significance of opinions within the body of United States law.

The definition of importance is at issue here. Perhaps the fact that the opinions on summary judgment or habeas corpus are so often cited by lower courts is clear evidence of their importance within the law. As noted above, the frequent citations can simply be a product of the types of cases most often litigated rather than anything related to the opinion itself, and the citations may be perfunctory. We will remain agnostic on the value of the lower court citation counts and report results for each of the court levels.

To the raw citation counts for the Supreme Court, we add a more sophisticated calculation, which uses a network methodology to generate legal relevance scores. Network studies are increasingly used throughout the sciences to measure various phenomena. 144Network research has been used to study “food webs, electrical power grids, cellular and metabolic networks, the World-Wide Web, the Internet backbone, the neural network of the nematode worm Caenorhabditis elegans, telephone call graphs, . . . and the quintessential ‘old-boy’ network, the overlapping boards of directors of the largest companies in the United States.” Steven H. Strogatz, Exploring Complex Networks, 410 Nature 268, 268 (2001). The most common use of networks probably involves social interconnections, such as patterns of Facebook friendships.

The network of citations of Supreme Court opinions is somewhat different. While two people may befriend one another, two cases cannot. The later case may cite an earlier opinion, but the earlier case cannot cite the later one, being not yet in existence. This feature makes the law a time-directed network, where links between cases can go in only one direction. 145 See Cross et al., supra note 5, at 1237. Our research enables an evaluation of an opinion based on the number of cases that cite that opinion plus the significance of those citing cases (based on the citations they receive).

Some have suggested that “the web of citations from one case to another is a critical component of the network of rules that comprise ‘the law.’” 146David G. Post & Michael B. Eisen, How Long Is the Coastline of the Law? Thoughts on the Fractal Nature of Legal Systems, 29 J. Legal Stud. 545, 545 (2000). The “extent and nature of a precedent’s network of citations” thus influences the “strength of its constraining power.” 147Michael J. Gerhardt, The Irrepressibility of Precedent, 86 N.C. L. Rev. 1279, 1289 (2008). Gerhardt suggests that the “clarity of [a precedent’s] significance and meaning . . . depends on the consistency and uniformity with which the Court and other public authorities have cited it.” Id. While the precise meaning of the network of citations is yet obscure (due to limited research), the connections plainly contain information for understanding the operation of stare decisis and identifying key historic opinions.

We use the legal relevance score for opinions building on mathematical tools developed for internet searches. 148 See generally Jon M. Kleinberg, Authoritative Sources in a Hyperlinked Environment, 46 J. Ass’n Computing Machinery 604 (1999) (proposing and testing a mathematical formula for determining what constitutes an authoritative internet site). See Fowler et al., supra note 15, for a description of the use of this measure for the legal network. This provides a measure of the “centrality” of a particular event (in our study, an opinion) in a broader network (the full corpus of citations). This captures the degree to which a precedent is embedded in the Court’s jurisprudence, using the citation patterns with which it has been applied. It correlates with raw citation numbers, because more citations provide more scores it can accumulate, but it also incorporates important information in the indirect linkages among cases. That is, if Case C cites Case B and Case B cites Case A but Case C does not cite Case A, Case A nonetheless gains some importance through the indirect citation linkage between it and Case C as gleaned through Case C’s citation of Case B.

This legal relevance score is arguably a better measure of opinion significance than available alternatives. 149Fowler et al., supra note 15, at 324–25. First, the legal relevance score captures both the direct and indirect connections in the network and thus improves upon a measure that includes only direct citations. In legal terms, the progeny of a Supreme Court opinion is a consequence of the opinion, and this relationship is captured by the legal relevance scores. Roe v. Wade, 150410 U.S. 113 (1973). for example, is a consequence of the Court’s opinion in Griswold v. Connecticut. 151381 U.S. 479 (1965). Cases that cite Roe on abortion rights will often not cite Griswold. Yet the earlier opinion in Griswold had a role in even those cases that did not directly cite it. The legal relevance score captures this indirect effect.

The legal relevance score captures both the number of citations received by an opinion and the significance of the citing cases (as measured by the number of citations their cited cases receive). 152Fowler et al., supra note 15, at 335. While raw citation counts can change over time (though they can only increase, not decrease), the legal relevance score metric is more dynamic and can either increase or decrease and tends to change more rapidly than raw citation counts. 153 Id. It serves as a measure of the significance of these cases as of the date of our measurement (calendar year 2005).

For an example of the effect of these legal relevance scores, consider the opinions on abortion rights. An earlier study found that Roe actually had fewer direct citations in the Supreme Court than did Webster v. Reproductive Health Services 154492 U.S. 490 (1989). and Thornburgh v. American College of Obstetricians & Gynecologists. 155476 U.S. 747 (1986); Fowler & Jeon, supra note 12, at 21. Yet the latter two opinions were the progeny of Roe and may not have existed absent the earlier opinion in Roe. Because they cited and relied upon Roe in their decisions, Roe gets some credit for their citations and has a higher legal relevance score than do the later decisions. Intuitively, Roe seems the more important decision, and the legal relevance score therefore seems to better capture the importance of the opinions. The top cases for legal relevance scores are set forth in Table 4. We note that the legal relevance scores are measured as a percentile. For example, Cantwell is above the 99th percentile on the score.

Table 4:

Top 25 Cases by Legal Relevance Score

Case

Legal Relevance Score

  • Cantwell v. Connecticut (1940)

1

  • Schneider v. State (1939)

.9999625

  • NAACP v. Button (1963)

.999925

  • Thornhill v. Alabama (1940)

.999888

  • New York Times Co. v. Sullivan (1964)

.999850

  • NAACP v. Alabama (1958)

.999813

  • Lovell v. Griffin (1938)

.999775

  • Speiser v. Randall (1958)

.999738

  • West Virginia Board of Education v. Barnette (1943)

.999700

  • Hague v. Commission for Industrial Organization (1939)

.999663

  • Shelton v. Tucker (1960)

.999625

  • Whitney v. California (1927)

.999588

  • Chaplinsky v. New Hampshire (1942)

.999550

  • Near v. Minnesota (1931)

.999513

  • Roth v. United States (1957)

.999475

  • Ashwander v. Tennessee Valley Authority (1936)

.999438

  • Pierce v. Society of Sisters (1925)

.999400

  • Buckley v. Valeo (1976)

.999363

  • Thomas v. Collins (1945)

.999325

  • Stromberg v. California (1931)

.999288

  • Boyd v. United States (1886)

.999250

  • United States v. O’Brien (1968)

.999213

  • McCulloch v. Maryland (1819)

.999175

  • Martin v. Struthers (1943)

.999138

  • Kovacs v. Cooper (1949)

.999101

Many of the cases regarded as most important appear high on this list, though the top two cases, Cantwell v. Connecticut and Schneider v. State, might seem surprising. Cantwell incorporated the First Amendment’s protection of religious free exercise as applied to the states. 156310 U.S. 296 (1940). It therefore may be the foundation for the large number of cases evaluating the constitutionality of state actions with respect to religion. Its place on the list is some evidence of how the use of legal relevance scores avoids the settled law bias discussed above—Cantwell settled the incorporation question, but it remains important in legal relevance scores because of the importance of its progeny, even if today’s cases may not directly cite Cantwell.

Schneider was an early freedom of speech opinion, striking down a local ordinance that barred persons from distributing handbills door-to-door and on public streets. 157308 U.S. 147 (1939). It created the public forum doctrine for free speech that has been the subject of much subsequent litigation that reached the Court. Although its direct citation numbers are not near the top of the historic list, its progeny effect is captured in the legal relevance score, which vaults it to second on our list. These cases show how legal relevance scores are an important measure of an opinion’s importance, independent of direct citations.

The highest legal relevance scores tend to be more recent decisions than the raw citations list, though McCulloch still checks in at number 23. This is because the classic cases may have been transcended in their importance by intervening decisions, and these scores reflect the contemporary importance of individual opinions. The legal relevance scores are ever changing, as described in the following section. While the legal relevance scores capture the importance of an opinion’s progeny cases, this too fades over time. We believe that the Supreme Court legal relevance scores are the best measure for case importance, but others may disagree, and we will report our analysis for the raw citation scales as well.

These lists are in no sense a list of the “best” Supreme Court opinions. “Best” is a subjective standard, and our criteria do not attempt to measure it. Rather, we measure importance in the law. Insofar as “winners write history,” the most important cases do have some quality. However great an opinion may be, if it lies fallow and uncited, that opinion is not making much of a difference in the law. Hence, it is worthwhile to assess the determinants that make a case more important in the corpus of stare decisis.

C. Change in Importance over Time

An opinion’s receipt of citations obviously varies over time. As Supreme Court terms pass, many additional opinions are rendered, which increases the opportunities of a case for citations. Older cases plainly have more opportunities, as more opinions have been rendered in which they may be cited. The number of citations per opinion also has increased dramatically over the Court’s history. 158Fowler et al., supra note 15, at 333. The opportunity to receive citations is therefore greater in recent years, which may enhance the scores of relevant recent opinions. Regardless of this effect, there is reason to believe that the significance of opinions diminishes over the years.

The seminal article examining the use of citation measures focused on the effect of time on an opinion’s receipt of citations. 159Landes & Posner, supra note 8. In this research, Landes and Posner propounded a capital investment analogy to the creation of precedents. 160 Id. at 262. Judges devote effort to the creation of precedential opinions as an investment in their holdings. 161 See id. at 264. The precedential value of these holdings depreciates over time, like other capital investments, as competing opinions enter the market for citations and the information content of the original decision “declines over time with changing circumstances.” 162 Id. at 263.

Landes and Posner found that the precedential impact of decisions declined by 2%–7% per year. 163 Id. at 280. Supreme Court precedents depreciated more slowly than those of circuit courts, which was ascribed to their broader generality. 164 Id. at 282–83. The authors theorized that precedent produces information relevant for deciding future cases, and the value of this information “declines over time with changing circumstances.” 165 Id. at 263. A study of state supreme courts found similar evidence of the depreciation of precedents. Friedman et al., supra note 35, at 807. This depreciation effect, though, is not uniform, and some cases may have significance that continues for decades or even centuries. 166Landes & Posner, supra note 8, at 263 n.19.

The change in case importance over time has been studied with quantitative analyses of legal relevance scores. 167 See Fowler et al., supra note 15. For an example, consider Townsend v. Sain, which found a plausible case for habeas corpus for a prisoner sentenced to death based on a confession obtained while he was under the influence of drugs, including a possible “truth serum.” 168372 U.S. 293 (1963). The pattern of citations to Townsend is displayed in Figure 3.

Figure 3: Total Citations to Townsend v. Sain

cross-spriggs-fig3

Figure 4: Legal Relevance Score of Townsend v. Sain

cross-spriggs-fig4

The opinion saw a steady rate of citations and a burst of use even twenty years after it was rendered. After about twenty-five years, though, its value as a precedent flattened out considerably. This conclusion is buttressed by the change in the legal relevance scores for this case as seen in Figure 4.

Landes and Posner have suggested that an opinion’s significance declines over time as it is supplanted by more useful contemporaneous opinions, and this is surely true (though the legal relevance scores capture its residual effect to some degree). 169Landes & Posner, supra note 8, at 263. Other factors may also influence the change in importance. Past opinions will be cited only to the extent that they are relevant to the legal issues addressed in the latter opinion. As the Court’s agenda changes, different cases will be more relevant and cited more often. 170Unsurprisingly, the choice of citations is closely related to the precedents’ legal relevance to the cases taken by the Court. Spriggs & Hansford, supra note 61, at 143. For example, our history has seen a dramatic change in the legal topics of cases taken by the Court. In the first half of the twentieth century, the Court focused on economic questions, while the latter half of the century saw a shift to jurisprudence centering on individual and civil rights. 171 See Biskupic & Witt, supra note 11, at 322 (noting that for the first 150 years of its history, “the Supreme Court exerted its greatest influence on the states of the Union through its decisions on matters of economic interest”). The era has also seen the adoption of intervening constitutional amendments and statutes that inevitably shift the cases taken by the Court. The Court’s opinions will therefore cite different patterns of prior opinions. This intuitive effect has been confirmed by empirical research showing that the issue area of citations corresponded to the issue area of the underlying precedent. 172 Hansford & Spriggs, supra note 5, at 62.

Consider, as an example, the difference in the importance between cases that overrule precedent and those that do not. Overruling cases are generally considered to be salient and important decisions in the Court’s overall jurisprudence. Due to their causing an abrupt change to the legal status quo, one would expect that they would be more central cases in the network of law than cases that do not overrule precedent. We offer a systematic test of this hypothesis below, and here we simply report the average legal relevance between these two types of cases. Figure 5 shows that for nearly the entire range of the age of precedent, cases that overrule precedent are more important than cases that do not. We also see that overruling cases acquire importance more quickly than their non-overruling counterparts, and this legal status advantage does not disappear until an overruling case is nearly fifty-seven years old.

Figure 5: Average Legal Relevance Score for Overruling and Non-Overruling Cases

cross-spriggs-fig5

D. Overrated and Underrated Opinions

Some cases may be perceived as being of great importance when in fact they will have very little impact. Our top twenty-five lists above correspond roughly to perceptions of the importance of cases, but there are exceptions. Some cases perceived as highly significant, either contemporaneously or even retrospectively, have had relatively little value, as measured by citations. The legal significance of other opinions, as measured by citations, has been overlooked by the legal expert evaluations.

Consider Boyd v. United States 173116 U.S. 616 (1886). as an opinion that may have been historically underrated. It did not appear on Schwartz’s list of top opinions, 174 See Schwartz, supra note 26. though it scores very high on total citations and on legal relevance score (given its age). It was a seminal decision on the Fourth and Fifth Amendments and held that constitutional protections for the security of persons and property should be liberally construed. 175 Boyd, 116 U.S. at 616. It has not been entirely overlooked, as it has received over 2,000 citations in law reviews and been recently described as “[a] crucial opinion early in the Court’s doctrinal development.” 176Adam M. Samaha, Originalism’s Expiration Date, 30 Cardozo L. Rev. 1295, 1324–25 (2008). Yet it is not generally recognized as one of the key Supreme Court opinions, though its citation history suggests it should be so regarded.

For the vast majority of opinions, our citation evaluations conform to general expectations. Cases that make the Congressional Quarterly or Oxford lists of important decisions have an average legal relevance score in the 89th percentile for all opinions; those not on the lists have a mean score in the 47th percentile, a statistically significant difference. However, there are individual cases where the importance assessments differ.

In this section, we compare the lists of cases perceived as important with those that have proved most important by our authority scores. For perceived importance, we use the lists compiled in Congressional Quarterly and Oxford. For each of these cases, we examined their authority scores as of 2005. Twelve cases tied for the lowest authority scores are among those included on the Congressional Quarterly and Oxford lists of important cases (the most overrated opinions), presented in Table 5.

Table 5:

Least Significant CQ/Oxford Cases

Case

Legal Relevance Score

  • Vacco v. Quill (1997)

.07256

  • Perpich v. Department of Defense (1990)

.07256

  • McDonald v. Smith (1985)

.07256

  • United States v. Ptasynski (1983)

.07256

  • Maryland v. Louisiana (1981)

.07256

  • County of Imperial v. Munoz (1980)

.07256

  • Train v. Campaign Clean Water, Inc. (1975)

.07256

  • Travis v. United States (1967)

.07256

  • United States v. Guy W. Capps, Inc. (1955)

.07256

  • Escanaba & Lake Superior Railroad Co. v. United States (1938)

.07256

  • Chapman & Dewey Land Co. v. Bigelow (1907)

.07256

  • United States v. Libellants of the Schooner Amistad (1841)

.07256

Although these opinions were identified in the books’ lists as important ones, none has particular renown. Our measures confirm their relative insignificance. The quantitative legal relevance score of the table is a percentile measure, so these cases are in the 7th percentile, meaning that over 92% of the Court’s opinions proved more significant within the set of legal relevance scores. The results of our legal relevance score findings seem reasonable, as these opinions are relatively obscure ones.

These findings must be qualified by the fact that it is a picture of the most important cases as of 2005, when our calculations were made. As the above section noted, the legal relevance scores change over time, sometimes dramatically. For example, according to legal relevance scores, United States v. Libellants of the Schooner Amistad (the famous Amistad decision) 17740 U.S. (15 Pet.) 518 (1841). was a very important case in the past. Yet by 2005, it had sunk to being one of the least important cases on legal relevance scores. Our calculations are of relatively contemporary significance.

For the most underrated cases, we looked for the highest legal relevance scores for cases that did not make it onto the Congressional Quarterly or Oxford lists. Table 6 displays this list of cases.

Table 6:

Most Significant Cases Not on Oxford/CQ Lists

Case

Legal Relevance Score

  • Winters v. New York (1948)

.99869

  • Police Department of Chicago v. Mosley (1972)

.99850

  • Prince v. Massachusetts (1944)

.99839

  • McGowan v. Maryland (1961)

.99828

  • Niemotko v. Maryland (1951)

.99794

  • Grayned v. City of Rockford (1972)

.99779

  • Sweezy v. New Hampshire (1957)

.99708

  • Monroe v. Pape (1961)

.99576

  • Ginsberg v. New York (1968)

.99554

  • Ferguson v. Skrupa (1963)

.99532

  • Screws v. United States (1945)

.99505

  • Pickering v. Board of Education (1968)

.99498

Our list of underrated cases by the Congressional Quarterly and Oxford lists contains at least a few cases the reader will recognize—probably at least Grayned and Monroe. Other cases on the list are less well known. Some may be familiar with Winters v. New York, which found unconstitutionally vague a state law prohibiting the sale of “obscene” magazines accounting criminal deeds, 178333 U.S. 507 (1948). but we doubt they would appreciate how very significant it is in the network of cases today. Police Department of Chicago v. Mosley dealt with the public forum doctrine and time, place, and manner restrictions on speech, and it has been very important in that recurring area of the law. 179408 U.S. 92 (1972).Prince v. Massachusetts found that the government properly had broad authority to protect children even from their parents. 180321 U.S. 158 (1944).McGowan v. Maryland is exceedingly important for upholding the constitutionality of Sunday sales prohibitions, even though religious in foundation, so long as they had a secular purpose. 181366 U.S. 420 (1961). Had it come out differently, Establishment Clause jurisprudence could be far different. All of our top underrated cases were in the top 1% of all Supreme Court opinions for citation influences.

There is a clear correspondence between the cases that are widely appreciated as important and those that receive the most citations and have the highest legal relevance scores. There is some divergence between perception and citation reality, though, as illustrated by our list of overrated and underrated cases. Public perceptions sometimes fail to appreciate the legal importance of some opinions, which we can capture through a study of citation frequency.

III. Determinants of an Opinion’s Legal Importance

Identifying the most important cases in the Supreme Court’s history is interesting, but this identification tells us little about the more interesting question: Why are certain cases more important? To the extent that these factors are within the control of the Justices, the identification of determinants could have great importance in evaluation of the Court. In this Part, we identify the determinants of more important Supreme Court opinions. We examine case characteristics (such as the issue and legal areas of the opinion), the opinion’s age as of the time of our calculations, various ideological features of the opinion, the nature of the majority coalition, and certain characteristics of the opinion (such as the number of citations and length of the opinion). The necessary data are not available for the full history of the Court though; thus, much of this analysis is limited to opinions rendered since 1946, for which the full case data is available.

There is surely some randomness to the significance of an opinion. Cantwell v. Connecticut is apparently important because it was the first decision to incorporate the First Amendment’s freedom of religion, 182 See 310 U.S. 296 (1940). and this area of the law became quite important at the Court. Schneider v. State is high on our list because it was an early decision on free speech law, 183 See 308 U.S. 147 (1939). which has become very significant at the Court. Nevertheless, opinion quality is relevant to our measures. Had these cases come out the other way or contained different legal analysis, their importance might be much less.

The significance of Supreme Court opinions is not foreordained by the case facts or legal questions addressed, however. In 1984, there was an expectation that Berkemer v. McCarty 184468 U.S. 420 (1984). would be a very important case, but it produced a relatively insignificant opinion and has been largely forgotten. 185Nancy Staudt et al., On the Role of Ideological Homogeneity in Generating Consequential Constitutional Decisions, 10 U. Pa. J. Const. L. 361, 361–62 (2007). This illustrates the key point that “it is the opinion of the Court, and not its bottom-line judgment, that determines the consequentiality of the decision.” 186 Id. at 370 n.39. “The intrinsic quality of the precedent relied upon is significant in determining its fate.” 187Schaefer, supra note 46, at 10.

We attempt to piece out the various factors that may drive the importance of a Supreme Court opinion and discover to what extent the opinion itself matters (as opposed to immutable external circumstances). We consider intrinsic case characteristics, age, features of the opinion itself, and control variables to assess the determinants of opinion importance by our measures.

A. Case Characteristics

Certain intrinsic case characteristics may determine the significance of an opinion for future citations, independent of the opinion itself. Some topics are simply more important for the Supreme Court or lower courts. The Supreme Court sets its agenda through certiorari decisions. If it takes cases of a given type, prior opinions of that type will receive more citations. Lower courts have their agendas set by litigants, and those decisions will also influence citation rates. In addition, certain legal groundings for opinions may produce more subsequent citations. To measure the characteristics of the case, we rely on Harold Spaeth’s data on historic opinions of the Court. 188For documentation, see Harold J. Spaeth et al., The Supreme Court Database (Aug. 26, 2010), http://scdb.wustl.edu/data.php. This source is so accepted in political science that it “would certainly be unusual for a refereed journal to publish a manuscript whose data derived from an alternate source,” and the same is true of law reviews. Lee Epstein et al., The Political (Science) Context of Judging, 47 St. Louis U. L.J. 783, 812 (2003). This source is the “greatest single resource of data on the Court.” Michael Heise, The Past, Present, and Future of Empirical Legal Scholarship: Judicial Decision Making and the New Empiricism, 2002 U. Ill. L. Rev. 819, 848.

1. Issue Area of Opinion

Decisions in some issue areas are sure to have greater future citation impact than others. The most obvious reason for this effect is the nature of the Court’s agenda. A decision interpreting the Bankruptcy Code, for example, will most commonly be cited in other bankruptcy decisions. If the Court does not accept certiorari in additional bankruptcy cases, the original decision in that area is unlikely to be much cited.

The available data breaks the Court’s cases into thirteen issue areas. 189The case types in the United States Supreme Court Judicial Database are categorized as the following: attorneys, civil rights, criminal procedure, due process, economic activity, federalism, First Amendment, judicial power, privacy, federal taxation, interstate relations, unions, and miscellaneous. Spaeth et al., supra note 188. The nature of the categories may not be entirely transparent, and more detailed descriptions of the categorization can be found in the codebook for the database. The Court’s relative attention to these areas has varied over time. In 1946, the Court decided 48 cases (out of 140 total) that were categorized as economic issues. 190 Lee Epstein et al., The Supreme Court Compendium 80 (3d ed. 2003). This number steadily declined to only 10 cases (out of 77 total) in 2001. 191 Id. at 85. During this time period, the Court took many more cases involving the Bill of Rights and civil liberties. 192 Id. The change in the nature of the citing cases is sure to influence the cases cited in an opinion.

To separate out the effect of case type, we isolate civil liberties cases. These include those categorized in the Supreme Court Database as involving First Amendment issues, due process, rights of criminal defendants, privacy, and civil rights. 193 See Spaeth et al., supra note 188. We expect these cases might receive more citations because they have been prominent on the Court’s agenda in recent years. The power of this issue-area effect is measured by the dummy variable Civil Liberties.

In addition to the Court’s agenda, one might expect different devotion to precedent in different case areas. Some have argued that the Court should give its greatest deference to stare decisis in its economic opinions because people have adapted to them through private ordering. 194 See Stevens, supra note 54, at 2. The Court has urged that “decisions affecting the business interests of the country should not be disturbed except for the most cogent reasons, certainly not because of subsequent doubts as to their soundness.” 195Nat’l Bank v. Whitney, 103 U.S. 99, 102 (1880). If this is indeed the case, one might expect precedent to be more powerful in economic decisions.

Businesses adapt to the law. They can write contracts based on their understanding of the law or possibly to avoid its application to their circumstances. Thus, they have a reliance interest in the prevailing law, the use of which is the basis for their private ordering of their actions. Such private ordering was the very first reason given for stare decisis by Hart and Sacks in their classic work on the legal process. 196 Henry M. Hart, Jr. & Albert M. Sacks, The Legal Process: Basic Problems in the Making and Application of Law 587 (1958); see also Lewis F. Powell, Jr., Stare Decisis and Judicial Restraint, 47 Wash. & Lee L. Rev. 281, 286 (1990) (declaring that reliance “is especially important in cases involving property rights and commercial transactions”).

Hence, the power of stare decisis is said to be at its “acme in cases involving property and contract rights, where reliance interests are involved.” 197Payne v. Tennessee, 501 U.S. 808, 828 (1991). This position has a pedigree in the earliest opinions of the Court. 198 See Polly J. Price, A Constitutional Significance for Precedent: Originalism, Stare Decisis, and Property Rights, 5 Ave Maria L. Rev. 113 (2007) (reviewing the history of giving strong stare decisis to decisions involving property rights); Lee J. Strang & Bryce G. Poole, The Historical (In)Accuracy of the Brandeis Dichotomy: An Assessment of the Two-Tiered Standard of Stare Decisis for Supreme Court Precedents, 86 N.C. L. Rev. 969 (2008). The logic of the position is not inexorable, however. People order their lives in reliance on the law in areas other than economics. Justice Marshall suggested that “stare decisis [was] in many respects even more critical in adjudication involving constitutional liberties than in adjudication involving commercial entitlements.” 199 Payne, 501 U.S. at 852–53 (Marshall, J., dissenting). For example, the reliance interest in abortion decisions has been emphasized. See Michael J. Gerhardt, The Pressure of Precedent: A Critique of the Conservative Approaches to Stare Decisis in Abortion Cases, 10 Const. Comment. 67 (1993). Nevertheless, the Court has been emphatic about the importance of stare decisis in economics cases.

We isolate economic precedents using the categorization of the Supreme Court database. These cases include the area of economics, plus cases involving federal taxation and labor. If an opinion falls within these areas it is coded as the dummy variable Economic for our analysis. This enables us to isolate any unique power of precedent for economic decisions, as often hypothesized.

2. Legal Area of Opinion

In addition to the issue area of the case, some legal areas may also influence future citations. The Court addresses matters of constitutional interpretation, statutory interpretation, the review of administrative agencies, admiralty common law, and other broad legal categories. Some legal areas of cases might be expected to yield more citations than others, which we measure by several categories.

One of the most common hypotheses is that the Court will give greater weight to statutory opinions than those in constitutional law. 200This position is generally traced to Justice Brandeis’s dissenting opinion in Burnet v. Coronado Oil & Gas Co., 285 U.S. 393, 406–10 (1932) (Brandeis, J., dissenting) (arguing for separate standards for constitutional and statutory precedents). Justice Powell explained: “The idea has long been advanced that stare decisis should operate with special vigor in statutory cases because Congress has the power to pass new legislation correcting any statutory decision by the Court that Congress deems erroneous.” 201Powell, supra note 196, at 287; see also Payne, 501 U.S. at 828 (noting that precedent has less power in constitutional cases because their “correction through legislative action is practically impossible”). One author argued for a rule in which stare decisis in statutory cases is absolute, though he recognized that this was not the case. 202 See Lawrence C. Marshall, “Let Congress Do It”: The Case for an Absolute Rule of Stare Decisis, 88 Mich. L. Rev. 177 (1989). Nevertheless, it is believed that the “Supreme Court has long given its cases interpreting statutes special protection from overruling.” 203Amy Coney Barrett, Statutory Stare Decisis in the Courts of Appeals, 73 Geo. Wash. L. Rev. 317, 317 (2005); see also Thomas R. Lee, Stare Decisis in Historical Perspective: From the Founding Era to the Rehnquist Court, 52 Vand. L. Rev. 647, 731 (1999) (discussing the historical development of this perspective). A study found that the Court is less likely to overrule statutory precedents, which seems consistent with this stated legal norm. 204James F. Spriggs, II & Thomas G. Hansford, Explaining the Overruling of U.S. Supreme Court Precedent, 63 J. Pol. 1091 (2001). The Court is said to defer “more to statutory than to constitutional precedents.” 205 Gerhardt, supra note 32, at 84. It has declared that in constitutional cases “stare decisis concerns are less pronounced.” 206Harris v. United States, 536 U.S. 545, 557 (2002); see also Agostini v. Felton, 521 U.S. 203, 235–36 (1997) (listing cases respecting this differentiation). Constitutional precedents are subject to only a “weak presumption of correctness.” Rafael Gely, Of Sinking and Escalating: A (Somewhat) New Look at Stare Decisis, 60 U. Pitt. L. Rev. 89, 110 (1998). The distinction reportedly has persisted since the first half of the nineteenth century. Earl M. Maltz, Some Thoughts on the Death of Stare Decisis in Constitutional Law, 1980 Wis. L. Rev. 467, 467.

In fact, Congress does sometimes reverse judicial precedents interpreting statutes. 207The classic investigation of this practice is found in William N. Eskridge, Jr., Overriding Supreme Court Statutory Interpretation Decisions, 101 Yale L.J. 331 (1991). Others have also empirically investigated the practice. See, e.g., Beth Henschen, Statutory Interpretations of the Supreme Court: Congressional Response, 11 Am. Pol. Q. 441 (1983) (examining the practice in the areas of antitrust and labor law); Joseph Ignagni & James Meernik, Explaining Congressional Attempts to Reverse Supreme Court Decisions, 47 Pol. Res. Q. 353 (1994) (considering determinants of when Congress overrides a judicial opinion). The Court may view the lack of congressional revision of a prior opinion to be evidence of its finding that the opinion got the interpretive question right and defer to the democratic processes. Alternatively, the Court may be adhering to stare decisis to avoid a congressional overruling of its decision. If Congress approved of the prior interpretation, a judicial modification might simply be reversed by the legislature, making the Court’s ruling a vain one, which would make the Justices less likely to attempt such a modification. 208 See Robert D. Cooter & Tom Ginsburg, Comparative Judicial Discretion: An Empirical Test of Economic Models, 16 Int’l Rev. L. & Econ. 295, 296 (1996) (noting that “prudent judges” would not push their ideology to the point where Congress would need to overturn their interpretation of the law).

Constitutional decisions, by contrast, cannot be reversed by the legislature. If the Court gets a constitutional opinion wrong, there is much less possibility for democratic correction. 209This difficulty may be overstated. There are various ways in which Congress may rewrite statutes so as to accomplish its goal independent of a constitutional decision of the Supreme Court. See Amy L. Padden, Overruling Decisions in the Supreme Court: The Role of a Decision’s Vote, Age, and Subject Matter in the Application of Stare Decisis after Payne v. Tennessee, 82 Geo. L.J. 1689, 1717–18 (1994) (describing some methods). In addition, Congress has a variety of ways of pressuring the Court over constitutional decisions, which may be simpler and more effective than the conventional rewriting of a statute. See Frank B. Cross & Blake J. Nelson, Strategic Institutional Effects on Supreme Court Decisionmaking, 95 Nw. U. L. Rev. 1437, 1459–71 (2001) (reviewing the research on these approaches). Consequently, the Court may show less deference to its prior constitutional opinions. This common invocation has been disputed as a descriptive matter, with claims that “the historical practice of the Court was to treat constitutional precedents the same as precedents in other legal areas.” 210Strang & Poole, supra note 198, at 974. A study of opinions in early cases in which the Supreme Court overruled a precedent found that there appeared to be no lessening of stare decisis for constitutional precedents. 211 Id. at 1015–25. Yet more recent Supreme Court decisions have applied the distinction. 212 See Payne v. Tennessee, 501 U.S. 808, 828 (1991). Given the widespread belief that stare decisis is weaker in constitutional opinions, we create a dummy variable for those cases, called Constitutional. This enables a test of whether constitutional opinions have less importance in the citation network.

One might expect a similar effect in other areas of law as well. Judge Posner has argued that decisions applying common law are less ideological and more grounded in stare decisis. 213 See Richard A. Posner, How Judges Think 82–87 (2008). Because these decisions are judge-made law, though, one might expect less deference to their holdings by subsequent judges. 214However, this effect is not clear because legislatures may override both common law and statutory decisions by statute. In general, “common law precedents enjoy a presumption of correctness stronger than that applied to constitutional cases, but not as constraining as that enjoyed by statutory precedents.” 215Gely, supra note 206, at 109. The common law is an evolving process, and judges might feel more authorized to depart from precedents that only reflect prior judicial holdings, rather than congressional action.

This theory about the greater force of statutory opinions (or those of other legal areas) has only been hypothesized, though, and never subjected to empirical scrutiny. Our methods enable a quantitative empirical test of this hypothesis. We also create the dummy variable Other Cases to capture the effect of those cases that are neither constitutional nor strictly matters of statutory interpretation. This leaves statutory cases as the baseline. The effect of Constitutional and Other Cases will appear as compared with the baseline of statutory interpretation decisions.

3. Legal Complexity

Another factor that could influence the number of citations received by a case is simply the number of legal issues considered in the opinion, as found in the Supreme Court Database. If an opinion addresses more legal issues, it covers more territory and consequently offers more potential for citation in future cases. An opinion interpreting two statutes plausibly has twice as many opportunities for future citation as does an opinion interpreting only a single statute.

This variable is captured by counting the number of legal issues and the number of legal provisions at issue in a case in the Supreme Court Database. The accuracy of the legal issue coding for the Database has seen challenge. 216 See Carolyn Shapiro, Coding Complexity: Bringing Law to the Empirical Analysis of the Supreme Court, 60 Hastings L.J. 477 (2009) (reporting errors in coding for legal issues). Nevertheless, the variable has been used and found useful in significant research. 217 See Forrest Maltzman et al., Crafting Law on the Supreme Court: The Collegial Game (2000); Timothy R. Johnson et al., The Influence of Oral Arguments on the U.S. Supreme Court, 100 Am. Pol. Sci. Rev. 99 (2006). The criticisms do not suggest any systematic bias that would undermine the accuracy of the use of the complexity variable; rather, they suggest the existence of random errors that would appear as statistical noise. 218The reevaluation of the coding found that the database undercounted legal issues overall, especially in issue areas such as judicial power and government structure and operations. See Shapiro, supra note 216, at 519–20. These errors would have the effect of producing statistical noise that interfered with finding a true effect. For our research we use the variable Complexity to measure the number of discrete legal issues and provisions in a majority opinion.

B. Age of Opinion

As discussed above, some research has demonstrated that the importance of citations declines over time, consistently and significantly. Older opinions may be less relevant to contemporary case facts, or they may simply have been superseded by intervening opinions. However, very recent opinions will have had little opportunity for citation, so they would appear to score low on citation measures, even if they might eventually prove to be very significant. As a result, the importance of opinions in the network of precedent will depend in part upon how old they are.

The role of the age of the precedent is not an unambiguous one. As Landes and Posner demonstrated, 219Landes & Posner, supra note 8, at 281. precedents appear to depreciate in importance over time. Other studies show that lower court citation or treatment of Supreme Court precedents declines after a certain amount of time. 220 Hansford & Spriggs, supra note 5, at 116–17 figs.7.1 & 7.2 (graphically displaying this effect). Absolute citations remain high for about twenty years but then precipitously decline in frequency. Id.; see also Spriggs & Hansford, supra note 61; Ryan C. Black & James F. Spriggs II, The Depreciation of Precedent on the U.S. Supreme Court 24–25 (June 12, 2009) (unpublished manuscript), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1421413. However, some older precedents “might be more institutionalized and thus possess greater vitality.” 221 Hansford & Spriggs, supra note 5, at 24. When opinions receive repeated positive citations, that fact strengthens their position for future citations. Our lists of the most important precedents above show some older opinions that score quite highly.

At present, a set of studies has found a clear empirical regularity regarding precedent age, showing older cases are generally less likely to be cited. These studies considered the likelihood of citation based on the age of the precedent and the age of the precedent squared. This is the conventional approach for examining a quadratic, non-linear relationship between variables. They found that the age variable was negatively related to the probability of a case being legally interpreted in a year, and the age-squared variable had a positive association. 222 Id. at 64. A somewhat older opinion was less likely to be cited but a much older opinion had a slightly higher probability of citation (or the likelihood of citation flattened out). The much older citations apparently had a deeper grounding in intervening opinions, which strengthened their power. While the age of an opinion does not doom its importance, it does generally lead to a diminishment in a case’s significance. The effect of age may be structured by citation history, and some opinions may maintain importance over the decades because they are repeatedly used.

To evaluate the effect of time on the importance of an opinion, we use two variables, Age (the number of years since its issuance) and Age-Squared (the square of the number of years since its issuance). This is a standard approach to assess a quadratic relationship, which appears as a U-shaped curve. The use of the two variables enables us to assess the possibility that the significance of an opinion increases (or decreases) for a certain amount of time, whereupon it then decreases (or increases).

C. Ideological Factors

There is now an enormous amount of information demonstrating that the Justices of the Supreme Court are influenced by ideology in reaching their decisions. 223 See Segal & Spaeth, Attitudinal Model, supra note 76; Segal & Spaeth, Attitudinal Model Revisited, supra note 5. Considerable additional evidence supports the basic findings of these books. See, e.g., Daniel R. Pinello, Linking Party to Judicial Ideology in American Courts: A Meta-Analysis, 20 Just. Sys. J. 219 (1999) (presenting a meta-analysis of the effect of ideology at various court levels and finding it most profound in the Supreme Court). In the extreme, this might make any citation studies wholly irrelevant because decisions were not based on precedents but instead on the individual preferences of the Justices, who “assemble diverse precedents into whatever pattern” they find convenient. 224 Mark Tushnet, Red, White, and Blue: A Critical Analysis of Constitutional Law 191–92 (1988). There appears to be an “inherent tendency of judges to manipulate the doctrine politically.” 225Christopher P. Banks, Reversals of Precedent and Judicial Policy-Making: How Judicial Conceptions of Stare Decisis in the U.S. Supreme Court Influence Social Change, 32 Akron L. Rev. 233, 235 (1999); see also Padden, supra note 209, at 1689 (suggesting that stare decisis “is often manipulated by liberals and conservatives alike when precedents are viewed as unappealing”). In such a case, the distribution of citations might simply be random or correspond exclusively to the ideological proclivities of the Justices.

Some political scientists have gone so far as to argue that precedent is essentially meaningless at the Supreme Court level. 226This research was originally published as Segal & Spaeth, supra note 11, and subsequently, in more detail, as Spaeth & Segal, supra note 77. One study examined cases in which Justices dissented from an original decision and found that they generally continued to dissent from its extensions in future cases. 227 Spaeth & Segal, supra note 77. However, these findings have been reexamined and questioned in follow-up analyses that considered different cases and Justices. 228For a summary of these analyses, see Charles A. Johnson, Follow-Up Citations in the U.S. Supreme Court, 39 W. Pol. Q. 538 (1986). Moreover, this study examined only case outcomes and did not consider the effect of precedent on the nature of the opinion in the case. 229One study of the early opinions of Chief Justice Roberts found that he viewed prior opinions not as “straightjackets that dictate[d] his decisions” but as “boundaries that shape[d] the nature of his opinions.” Cross, supra note 61, at 1276. Additional empirical foundation comes from research suggesting that a benchmark opinion significantly shaped the subsequent opinions issued by the Court. Herbert M. Kritzer & Mark J. Richards, Jurisprudential Regimes and Supreme Court Decisionmaking: The Lemon Regime and Establishment Clause Cases, 37 Law & Soc’y Rev. 827 (2003); Mark J. Richards & Herbert M. Kritzer, Jurisprudential Regimes in Supreme Court Decision Making, 96 Am. Pol. Sci. Rev. 305 (2002); Kevin M. Scott, Reconsidering the Impact of Jurisprudential Regimes, 87 Soc. Sci. Q. 380 (2006).

The claim that the Justices are utterly ideological is too strong. 230 See, e.g., Lawrence Baum, Judges and Their Audiences 20 (2006) (“[T]he existing evidence does not establish that Justices are motivated solely (or even overwhelmingly) by policy goals”); Timothy Johnson et al., Oral Advocacy Before the United States Supreme Court: Does It Affect the Justices’ Decisions?, 85 Wash. U. L. Rev. 457, 525 (2007) (reporting that “the attitudinal model is inaccurate in its main theoretical claim” that ideology is the only factor driving the Court’s decisions). While in some cases precedent “appears to be trotted out in defense of decisions that were actually reached on quite independent grounds,” there are others where “the Court actually seems to consider itself bound to adhere to a precedent because of the stare decisis principle.” 231Marshall, supra note 202, at 178. The presence of unanimous opinions, notwithstanding the ideological diversity of the Justices, is testimony to the effect of the law. It is in these cases that outcomes might be “explained by the presence of a very clear precedent.” 232 Mitchell S. G. Klein, Law, Courts, and Policy 112 (1984). An example of this effect might be found in Patterson v. McLean Credit Union, 491 U.S. 164 (1989), where the Court “shocked the legal community” by unanimously supporting the reach of civil rights legislation based on precedent. Marshall, supra note 202, at 178. If very clear precedents are governing the outcome of votes in unanimous decisions, it seems plausible that less clear precedents are at least influencing the language of divided opinions. Political scientists find that Justices adhere to stare decisis in order to preserve the legitimacy of the judiciary, even at the expense of their ideological preferences. 233Jack Knight & Lee Epstein, The Norm of Stare Decisis, 40 Am. J. Pol. Sci. 1018, 1029 (1996); Spriggs & Hansford, supra note 61 (showing that the Court is more likely to positively interpret precedents that have a higher level of legal vitality).

Devotees of ideological determinants of opinions could suggest that unanimous decisions occur only in the presence of an unusually ideologically extreme lower court decision, beyond the ideological positions of even the most extreme Justices of the Supreme Court. It seems implausible that the Court would take so many cases so extreme (or even that they exist), and this theory cannot explain unanimous affirmances. One study of unanimous reversals and the nature of the decisions below found that the cases were not ideological extremes, but instead cases determined by circuit court judges with ideologies paralleling that of the Supreme Court. 234Donald R. Songer & Dona Roy, A Critical Test of the Attitudinal Model 17–18 (Apr. 10, 2005) (unpublished manuscript), available at http://www.allacademic.com/meta/p86158_index.html. The study also found no major difference in the lower court ideologies in cases that saw unanimous liberal and unanimous conservative decisions. Id.

Research indicates that precedent plays some role in Supreme Court decisions, if only as a constraint on Justices’ preferences. The Justices commonly refer to precedent during their internal conference discussions of cases. 235 See Knight & Epstein, supra note 233, at 1027. Stare decisis does not control the Court’s decisions, “but it does structure and influence them.” 236 Lawrence Baum, The Supreme Court 150 (5th ed. 1995). Empirical research also shows that the “[l]aw matters in Supreme Court decision making in ways that are specifically jurisprudential.” 237Kritzer & Richards, supra note 229, at 315. Consequently, we study which opinions have the most influence.

While the role of ideology in Supreme Court decisions may be overstated by some, it plainly exists. The existence of non-unanimous opinions with systematic ideological vote patterns testifies to its role. And “many studies suggest that the interpretation of precedent depends at least in part on the Justices’ policy goals.” 238 Hansford & Spriggs, supra note 5, at 10. The Justices are influenced both by their preferences and by the state of the law, including precedents. Hence, it is important to incorporate ideology into a study of citation effects. 239One study found a very limited ideological effect in the Justices’ choice of citations, but it considered only a small number of opinions. Johnson, supra note 228. We use four ideological variables. These include the ideological direction of the decision associated with the opinion (liberal or conservative), the ideological composition of the Justices sitting on the Court at the time of the opinion, the ideological homogeneity of the majority coalition and the ideological extremity of the author, and the distance between the Court’s ideological composition at the time of the original opinion and the time of the later citing opinion.

1. Ideological Direction of Decision

One possible factor influencing the significance of an opinion is whether it produces a liberal or a conservative result. There is a hypothesis that conservative Justices give greater fealty to stare decisis than do liberal Justices. The liberals are sometimes considered more activist in their decisions. 240Staudt et al., supra note 185, at 373 (referring to the view that “liberals are more likely to render cutting-edge decisions”). If so, a liberal opinion would receive greater respect (and more citations). Judge Posner has suggested a ratchet effect, as conservative Justices have given greater respect to prior liberal precedents without reciprocal deference from liberal Justices. 241 Richard A. Posner, The Federal Courts: Crisis and Reform 217 (1985) [hereinafter Posner, Crisis]. Others have suggested likewise, 242 See, e.g., John C. Eastman, Stare Decisis: Conservatism’s One–Way Ratchet Problem, in Courts and the Culture Wars 127 (Bradley Watson ed., 2002); Lino A. Graglia, The Myth of a Conservative Supreme Court: The October 2000 Term, 26 Harv. J.L. & Pub. Pol’y 281, 284 (2003) (arguing that this ratchet effect was occurring, as a conservative Court was adhering to liberal precedents); Frank B. Cross, Gay Politics and Precedents, 103 Mich. L. Rev. 1186, 1203 (2005) (reviewing Daniel R. Pinello, Gay Rights and American Law (2003)) (addressing hypothesis that “activist liberal precedents breed activism, while contrary conservative precedents have only a weak countervailing effect, because conservative judges feel bound by stare decisis”). and some limited empirical evidence appears to support the hypothesis. 243 See Pinello, supra note 243 (reporting evidence at different court levels showing that pro-gay rights decisions had a much more significant precedential effect than contrary decisions). Some conservatives would suggest that the Court should rely less on constitutional precedent and more on originalism. 244 See, e.g., Steven G. Calabresi, Text, Precedent, and the Constitution: Some Originalist and Normative Arguments for Overruling Planned Parenthood of Southeastern Pennsylvania v. Casey, 22 Const. Comment. 311 (2005) (exploring the tension between various proposals for understanding the Constitution). Thomas Merrill has thus observed that for originalists, the importance of “judicial precedent—the most important tool of the constitutional lawyer—drops from sight.” Thomas W. Merrill, The Common Law Powers of Federal Courts, 52 U. Chi. L. Rev. 1, 69 (1985). On today’s Court, conservative Justices are actually more likely to ascribe to a weaker theory of precedent. 245 See Gerhardt, supra note 32, at 53.

An alternative explanation for greater liberal citation influence over time invokes sociology and history. In at least some major constitutional areas, the nation and the Court have become more liberal over time. This is certainly the case for civil rights. The Warren Court adopted standards of racial equality in decisions like Brown v. Board of Education 246347 U.S. 483 (1954). and Loving v. Virginia. 247388 U.S. 1 (1967). Contemporary conservatives have no respect for segregation, a conservative doctrine of the past, and are happy to embrace the once-liberal precedents. The same effect can be seen in many aspects of the Bill of Rights. For most of the amendments, a decision against the government is regarded as a liberal one (such as defendants’ rights).

The Warren Court issued numerous liberal rulings in support of the constitutional rights of criminal defendants, which have been accepted even by subsequent conservatives. Few today challenge the right to counsel (Gideon), or the application of the Fourth Amendment to wiretaps (Katz v. United States). 248 See Lindquist & Cross, supra note 24, at 146–47 (describing the persistence of decisions such as these). Although Miranda was very controversial, it has been reaffirmed by contemporary conservatives. 249Dickerson v. United States, 530 U.S. 428, 428 (2000). Likewise, “[t]hat the tolerant libertarian view of the free speech clause has such broad support in contemporary America and on a Court that has become increasingly conservative is yet another example of Warren Court activism moving doctrine in a direction that has both persisted and further developed.” 250 Frederick P. Lewis, The Context of Judicial Activism: The Endurance of the Warren Court Legacy in a Conservative Age 38 (1999). It may be that for some historical reason, Warren Court liberal decisions were ahead of the curve and influential for a changing America. We measure for such an effect with a dummy variable called Liberal Precedent. This is coded as “1” if the opinion is a liberal one, so a positive association would show greater precedential power for liberal opinions.

2. Ideological Composition of Court Coalitions

In addition to the decision, we consider the ideological positions of the individual Justices. As noted above, ample research has found that some Justices’ votes are systematically more liberal, while others are more conservative. We provide individualized measures for each Justice’s ideology, using what are known as Martin–Quinn scores, which have become the standard used for research such as this study. 251Some studies in law reviews making use of this measure include Andrew D. Martin et al., The Median Justice on the United States Supreme Court, 83 N.C. L. Rev. 1275 (2005); Theodore W. Ruger, Justice Harry Blackmun and the Phenomenon of Judicial Preference Change, 70 Mo. L. Rev. 1209 (2005); and Paul J. Wahlbeck, Strategy and Constraints on Supreme Court Opinion Assignment, 154 U. Pa. L. Rev. 1729 (2006).

The significance of a decision may be driven by the overall ideological composition of the Court issuing it. Some argue that the key to consequential decisions is the existence of an ideologically homogenous Court. 252 See Staudt et al., supra note 185, at 363. By this theory, a majority coalition with Justices largely in ideological accord is better able to agree on opinion language that will produce a significant effect. If the coalition is more diverse, the Justices would have more difficulty agreeing on consequential language and will therefore produce a less significant opinion.

A study sought to measure the ideological homogeneity hypothesis as a determinant of producing opinions significant enough to appear on the front page of the New York Times. It found that greater ideological diversity in the majority opinion was strongly correlated with a lower probability of producing such a consequential opinion. 253 Id. at 379. This study also found that liberal decisions were more likely to reach its threshold for a consequential opinion. 254 Id. at 381–82.

To test the hypothesis that more ideologically homogenous courts produce more powerful opinions, we use the variable Homogeneity, which is the standard deviation of the Martin–Quinn scores (which measures for individual Justice ideology) for all Justices in the majority opinion coalition of a precedent.

This measure may underestimate the role of the opinion author in drafting the opinion. There is a dispute over the degree to which opinions are driven by the full majority coalition as opposed to the author, which we discuss in detail below. 255 See infra Part III.D. The Homogeneity variable, taken alone, presumes that content is controlled by the full coalition, so that more ideologically cohesive and extreme coalitions will produce more powerful opinions. However, we know from prior research that opinion authors play a key role in structuring opinions, and we therefore included the variable Author Ideological Extremity to account for the author’s ideological position. We measure ideological extremity as the absolute value of the difference between an opinion author’s Martin–Quinn Score and the Martin–Quinn score for the median Justice on the Court in the year the precedent was released.

3. Ideological Distance from Citing Court

Given the evidence of ideology’s influence on Supreme Court outcomes, one might expect a similar effect on citation of prior opinions. Thus, a liberal opinion might be cited less by a subsequent, more conservative Court. The conservative Court might seek out more amenable opinions from different eras, when the Court was more conservative. Thus, we expect that a greater ideological gap between Courts will be associated with fewer citations to the opinions of the earlier, ideologically distant Court, controlling for our other variables.

There is some research on this question that shows a limited effect. A study of the fate of Warren Court precedents in the more conservative Rehnquist Court’s opinions found variance according to opinion author. 256Cross et al., supra note 92, at 15. Opinions by Justice Marshall, and to a lesser degree Justices Fortas and Brennan, saw especially high levels of depreciation during the Warren Court. 257 Id. Opinions written by Chief Justice Warren, however, saw relatively little depreciation.

A larger study of the Supreme Court’s use of precedent clearly confirmed this effect of ideology on the nature of citations in an opinion. The study found that the smaller the ideological distance between a precedent and the composition of the contemporaneous Court, the more likely it was to be positively interpreted and the less likely it would be negatively interpreted or overruled. 258 Hansford & Spriggs, supra note 5, at 64. The ideological position of the precedent was measured based on the ideological voting pattern of the median member of the majority voting coalition behind the opinion. Id. at 59. However, this ideological effect was not the only factor in citation frequency. The study also showed that the effect of the ideological distance of the Court from a precedent was conditional on the legal vitality of a case. 259 Id. at 65. Even when a contemporary Court was ideologically distant from a precedent, the Court was more likely to positively interpret the case if it had greater legal vitality (i.e., the precedent had more often been positively than negatively interpreted in the past).

While ideological distance most likely affects citation frequency, we cannot assess its role because our measures of importance are calculated as of 2005 and thus do not vary over the life of a case. To test this hypothesis, one would need a data set that contained an observation for each case in each year, allowing one to examine how the number of citations to a case (or the legal relevance score) varies as a function of changes in the Court’s ideological distance from a precedent. One piece of research actually examined the role of ideological distance in citation frequency at the Court and showed that the effect of ideological distance matters for only a relatively short time after a case is decided (approximately eight years). 260Black & Spriggs, supra note 220, at 23 (discussing a case’s ideological “depreciation”).

D. Opinion Characteristics

Perhaps the most intriguing question in this study involves the effect of opinion characteristics on future citations. The characteristics of an opinion are within the control of the opinion’s author. Because these factors are at least somewhat controllable, the Justices can use them to give an opinion greater power. Thus, it has been suggested that a Justice’s use of “reason with taut logic” and “persuasive rhetoric” would make future Justices more willing to adopt the Justice’s opinion. 261 Walter Murphy, Elements of Judicial Strategy 98 (1964); see also Jeffrey R. Lax & Charles M. Cameron, Bargaining and Opinion Assignment on the US Supreme Court, 23 J.L. Econ. & Org. 276, 277 (2007) (“[T]he policy impact of a legal opinion depends partly on its persuasiveness, clarity, and craftsmanship—its legal quality . . . .”).

The tautness of an opinion’s logic and the persuasiveness of its rhetoric are difficult to study objectively. A recent study sought to examine the effects of opinion language in administrative law decisions at the circuit court level. 262Robert J. Hume, The Impact of Judicial Opinion Language on the Transmission of Federal Circuit Court Precedents, 43 Law & Soc’y Rev. 127 (2009). However, its variables were not truly language but the type of legal area, the presence of block quotations, and whether the decision was rendered per curiam. 263 Id. at 138–41. It found the expected association for these variables, but the effects were small, and none of the variables truly captured the tautness or persuasiveness of opinion language.

We have no available data that can capture the persuasiveness of the opinion’s language for future Justices or the tautness of its reasoning. However, we can consider several important factors, including the size of the majority coalition behind the opinion, the number of citations to prior decisions in the opinion, and the absolute length of the opinion.

1. Nature of Majority Coalition

The vote margin of an opinion may affect its rate of future citations. In addition to the ideological homogeneity hypothesis, some argue that the number of Justices in the majority, regardless of their ideological positions, will influence the significance of a Supreme Court opinion. The number of Justices joining a majority may be considered relevant to its legal authority.

There has been a “traditional view . . . that an opinion’s precedential authority is directly proportional to the number of Justices that join it.” 264Mark Alan Thurmon, When the Court Divides: Reconsidering the Precedential Value of Supreme Court Plurality Decisions, 42 Duke L.J. 419, 449 (1992). The existing literature generally suggests that “separate opinions and smaller decision coalitions will cause a precedent to be weaker.” 265 Hansford & Spriggs, supra note 5, at 41. Judge Posner has contended that a dissenting opinion “undermines the majority opinion.” 266 Richard A. Posner, The Federal Courts: Challenge and Reform 236 (1999) [hereinafter Posner, Challenge]. Justice Rehnquist has declared that the Court feels greater latitude to overrule those cases “decided by the narrowest of margins.” 267 Payne v. Tennessee, 501 U.S. 808, 828–29 (1991). Hence, one might expect that opinions backed by only a minimum-winning coalition would be weaker and would command fewer future citations.

There is a corresponding belief that unanimous opinions may be more powerful ones. Anecdotally, the Court sought unanimity in some decisions (such as Brown and United States v. Nixon) in order to give the decision greater power. 268 See, e.g., Bob Woodward & Scott Armstrong, The Brethren: Inside the Supreme Court 345 (1979) (discussing Justice Brennan’s view that unanimity was required for the impact of United States v. Nixon); Gabriel J. Chin & Anjali Abraham, Beyond the Supermajority: Post-Adoption Ratification of the Equality Amendments, 50 Ariz. L. Rev. 25, 26 (2008) (“The impact and legitimacy of Brown v. Board of Education flowed in part from the Court’s unanimous rejection of segregation . . . .” (footnote omitted)); Frank B. Cross & Stefanie Lindquist, The Decisional Significance of the Chief Justice, 154 U. Pa. L. Rev. 1665, 1677–78 (2006) (“The ability to produce unanimous decisions in controversial cases, such as Brown and United States v. Nixon, [was] often regarded as vital . . . .” (footnotes omitted)); Micah Schwartzman, Judicial Sincerity, 94 Va. L. Rev. 987, 1023 (2008) (citing Brown as an example of a case “in which unanimity was thought to be essential to the Court’s legitimacy”). Some scholars “contend that it is when the Court speaks in one voice that it best is able to generate consequential precedent.” 269Lee Epstein et al., On the Capacity of the Roberts Court to Generate Consequential Precedent, 86 N.C. L. Rev. 1299, 1306 (2008); see also Bradley C. Canon & Charles A. Johnson, Judicial Policies: Implementation and Impact 168–69 (2d ed. 1999) (contending that unanimous decisions create “final, clear and persuasive policy”). This same effect might be seen to a lesser degree in relatively disparate majorities (e.g., 8–1).

Even short of necessary unanimity, more votes may strengthen an opinion. Walter Murphy suggested that “a 5–4 decision emphasizes the strength of the losing side and may encourage resistance and evasion. The greater the majority, the greater the appearance of certainty and the more likely a decision will be accepted and followed in similar cases.” 270 Walter F. Murphy, Elements of Judicial Strategy 66 (1964). Evan Caminker stated that minimum-winning coalitions “may well command weaker stare decisis respect.” 271Evan H. Caminker, Sincere and Strategic Voting Norms on Multimember Courts, 97 Mich. L. Rev. 2297, 2321 n.73 (1999). Similarly, Chief Justice Rehnquist suggested that less precedential effect should be attributed to decisions resolved “by the narrowest of margins, over spirited dissents challenging the basic underpinnings of those decisions.” 272 Payne, 501 U.S. at 828–29. This position has been criticized as substantially undermining the power of stare decisis. See Padden, supra note 209, at 1713–14.

An alternative view would suggest that decisions with greater dissensus will be the more significant ones. Many prominent cases have been decided with a minimum-winning coalition, on a 5–4 vote of the Court. Frederick Schauer suggested that cases decided unanimously are simply those of relatively little interest to the Court. 273Frederick Schauer, Statutory Construction and the Coordinating Function of Plain Meaning, 1990 Sup. Ct. Rev. 231, 247. Others have argued that “the contention that the Supreme Court does most when it speaks with one voice defies logic” because unanimity necessarily produces narrower opinions among ideologically diverse Justices. 274Epstein et al., supra note 269, at 1306; see also Staudt et al., supra note 185, at 372 (“[A]s each additional Justice agrees to sign on, each presumably with his or her preferences, the decision becomes more and more diluted and thus produces less of an impact than could be achieved by five simpatico Justices.”). The study on ideological homogeneity also found that the number of Justices in the majority was negatively associated with creation of consequential opinions, regardless of ideological homogeneity. 275Staudt et al., supra note 185, at 380. This finding supports Schauer’s theory that large majorities are associated with relatively insignificant cases, which would be presumed to have less effect on future opinions.

The debate over the significance of coalition size for opinion significance thus involves two conflicting factors. On one side, some believe that a greater number of Justices in the majority exogenously gives an opinion greater force and influence. The opposing position does not directly dispute this fact but contends potentially significant issues simply will not command large coalitions. The endogeniety of coalition size therefore means that the size of the majority will correlate with less significant decisions.

To capture the effect of coalition size, we use the two coalitions hypothesized to be most important—the unanimous opinion (professed to be of greater precedential weight) and the minimum-winning coalition (a 5–4 decision, professed to be weaker in precedential influence). This is operationalized with dummy variables for unanimous opinions (Unanimous) and for minimum-winning coalition opinions (MWC). 276These data are from Spaeth et al., supra note 188. The regression will compare both against cases decided by intervening coalition sizes.

2. Citations in Opinion

Another possible factor in the significance of an opinion is how well grounded that opinion is in the law. While the true legal groundedness of an opinion is a subjective measure, the number of citations in the underlying opinion could be a proxy for this factor. A political scientist has recently suggested:

A precedent that is backed with references to cases, statutes, and other materials is likely to appear more important than a precedent that is less well defended. Large quantities of supporting evidence signal to other judges that the outcome endorsed by a court is well grounded in legal authorities. It also suggests that the opinion writer has put a good deal of time and care into the decision. 277Hume, supra note 262, at 132.

Thus, a large number of citations in an opinion may serve as a signal that the opinion has strong legal support and is worthy of particular respect by future Courts.

The meaning of an opinion’s number of citations is not clear. More citations may simply reflect the existence of more meaningful precedents rather than the true degree of the opinion’s grounding in the law. It may be that few citations actually reflect better opinions, as the Justices may have used more discrimination in selecting “the most clearly applicable authority.” 278John Henry Merryman, Toward a Theory of Citations: An Empirical Study of the Citation Practice of the California Supreme Court in 1950, 1960, and 1970, 50 S. Cal. L. Rev. 381, 421 (1977). Conversely, more citations may display the “breadth” of an opinion, making it more relevant to future judges. 279 Id. at 422. Merryman’s study of the California Supreme Court supported this as the justices who cited the most overall authority also cited the most authoritative sources. Id. The absolute number of citations in an opinion is only a rough indicator, but it may capture something of the legal grounding of an opinion.

Justice Cardozo had a practice of citing more cases than his contemporaries, and this reflected his recognition of the “practical necessity for tying forward-looking opinions into the precedential past in order to make them acceptable” to various audiences, including judges. 280Robert A. Leflar, Honest Judicial Opinions, 74 Nw. U. L. Rev. 721, 724 (1979). This greater acceptability could give opinions with more citations greater power for structuring future opinions. We conducted a preliminary study on this effect and found that opinions containing more citations in fact appeared to result in the receipt of more future citations by both the Supreme Court and lower courts. 281 See Johnson et al., supra note 228.

Greater raw citation numbers may be consistent with opinions that are better grounded on the doctrine of stare decisis. Alternatively, they may reflect a desire of the Justices to alter that network. 282 See Cross, supra note 61, at 1276 (suggesting that Chief Justice Roberts views citations as a means for shifting the law, “creating a new path of stare decisis that will direct the course of future rulings”). By their nature, citations are interpretations of the meaning of prior opinions. In a prominent example, Chief Justice Roberts sought to invoke Brown for the now-conservative position of “color blindness,” rejecting affirmative action. 283Parents Involved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701, 746–48 (2007) (plurality opinion). This citation was apparently an attempt to channel the meaning of Brown for future cases. The more an opinion cites cases, the more it engages in this shaping of stare decisis, and one might therefore expect it to have greater impact in the law. We measure this effect with the variable Citations in Prec., which captures the number of citations contained in each of the opinions studied. 284These data come from Fowler et al., supra note 15, at 336–37.

3. Length of Opinion

Opinions plainly come in very different lengths. Different types of courts tend to issue opinions that systematically vary in their length. 285 See Friedman et al., supra note 35, at 775 (comparing similar opinions of different national courts and finding the French opinions shorter than those of American courts, which were in turn much shorter than opinions of English courts). There is also some systematic variation in opinion length of U.S. state supreme courts. Id. at 781–83. A longer opinion might be expected to receive more citations for various reasons. It may simply be that Justices devote more opinion-writing time and effort to more important cases, so that longer opinions merely appear more significant. However, the length of an opinion may have its own direct effect on the importance of that opinion, as it contains more material to be cited and may be associated with a more thoroughly reasoned, and therefore more persuasive, opinion.

There is a hypothesis that opinion length could relate to “precedential significance.” 286Peter H. Schuck & E. Donald Elliott, To the Chevron Station: An Empirical Study of Federal Administrative Law, 1990 Duke L.J. 984, 1003–04. A longer opinion could provide more content and set a stronger precedent. A shorter opinion may be more formalistically deductive and straightforward, while a longer opinion may be more inductive and potentially law changing. 287Friedman et al., supra note 35, at 775–76. More “policy-oriented justifications” may be associated with longer opinions. 288 Id. at 778. An opinion that treats a case as settled law and breaks no new legal ground may be shorter. 289 See Scott Phillips & Ryken Grattet, Judicial Rhetoric, Meaning-Making, and the Institutionalization of Hate Crime Law, 34 Law & Soc’y Rev. 567, 587 (2000). Such explanations could well relate to case significance.

A variety of factors will surely influence an opinion’s length, even beyond the choices of its author. A heavier workload may reduce opinion length, while greater support (such as clerks) may increase it. Dealing with more separate legal issues surely increases length, but this may be unrelated to any significance associated with the precedent. A greater discussion of underlying factual details will lengthen an opinion but in the process reduce its precedential significance by limiting its power to specific facts. The nature of the relationship between an opinion’s length and its significance is therefore unclear. A study of circuit court opinions found those grounded heavily in factual determinations were longer than those grounded in legal interpretation. 290Schuck & Elliott, supra note 286, at 1004. Longer opinions may be written “to limit the scope of the holding and its precedential effect.” 291Cross, supra note 126, at 66. Judge Mikva stated that if “you put too many facts in there that future advocates can distinguish[,] . . . as a precedent it’s not as useful.” 292 William Domnarski, Federal Judges Revealed 193 (2009) (quoting Abner J. Mikva, Honorable Abner J. Mikva Oral History, in Historical Society of the District of Columbia Circuit Oral History Program 210 (1996)). This suggests that longer opinions may have less significance as legal precedents.

Judges themselves are critical of longer opinions. Judge Pell said that many of the courts’ opinions were “too long.” 293 Id. (quoting Wilbur Frank Pell, Jr., The Oral History of Judge Wilbur F. Pell, Jr., in Circuit Library, U.S. Court of Appeals for the Seventh Circuit 64 (1998)). Judge Mikva said, “[N]othing has done more to harm appellate jurisprudence and law school teaching than the length of opinions.” 294 Id. (quoting Abner J. Mikva, Honorable Abner J. Mikva Oral History, in Historical Society of the District of Columbia Circuit Oral History Program 210 (1996)). Longer opinions may be more ambiguous. 295 See Posner, Challenge, supra note 266, at 147 (suggesting that longer opinions “reduce the opinion’s usefulness as a guide”). A considerable discussion exists comparing rules versus standards. 296 See supra note 65. A rule might be expressed in a straightforward manner, with more brevity, while an opinion setting out a standard may require discussion of all the considerations in its application and how they are to be weighted by subsequent courts, which might be expected to take more pages to set out. Yet a rule would not be expected to have less power in the law than would a more discretionary standard.

Some recent research has begun to examine the meaning of opinion length. A study of circuit court opinions found that reversals were significantly longer than affirmances. 297 Cross, supra note 126, at 65. This offers some support for the thesis that longer opinions may be more important. The study also found that longer circuit opinions produced more citations (both negative and positive), even controlling for whether it was a reversal or an affirmance. 298 Id. at 225–26. As expected, reversals also produced more citations. Id. at 214–15. A separate study of circuit courts likewise found that longer opinions received more citations. 299 Posner, Crisis, supra note 241, at 236 (reporting that longer opinions were more likely to be cited).

While the opinion-length associations for circuit courts may not translate to similar effects at the Supreme Court level, one recent study examined Supreme Court opinion length. 300Black & Spriggs, supra note 112. This study examined the length of opinions throughout the entire history of the Court, with controls for changes in opinion length over time, availability of law clerk support, collegial interactions, case type, and other factors. The authors found that various factors were strongly associated with opinion length, including the amount of bargaining in a case, the size of the majority coalition, its composition, workload considerations, and the complexity and salience of the case decided. 301 Id. at 661.

In addition to evaluating the determinants of longer opinions, the authors also considered the effect of longer opinions on citation rates by lower federal courts. They found that longer opinions were more likely to receive citations (both negative and positive), though the effect was a rather modest one. 302 Id. at 676–79. We therefore use opinion length as a variable (Majority Length) for assessing case importance, along with additional measures. 303We obtained these data from Black & Spriggs. See id. This variable is measured by the total number of words in the majority opinion.

It is plausible that the length of an opinion is truly not a feature of opinions but instead simply a reflection of the nature of the case, perhaps another measure for complexity of the issues presented. Alternatively, a majority opinion could be longer due to the need to respond to the arguments of dissenters. To account for this, we also include a measure of the length of the separate opinions in the case, by number of words, called Separate Length. 304We obtained these data from Black & Spriggs. See id. This variable would be a measure both of the complexity of the issues and the extent of disagreement. With its inclusion, the Separate Length variable better captures the true independent effect of majority opinion length.

4. Footnote Ratio of Opinion

Another feature of opinion writing is the use of footnotes. This feature has been called a “public nuisance of long standing.” 305Ray Forrester, Supreme Court Opinions—Style and Substance: An Appeal for Reform, 47 Hastings L.J. 167, 186 (1995). The reasoning behind the Justices’ use of footnotes is obscure. While some footnotes are just extensive citations that would sit poorly in the text, many opinion footnotes contain text that is part of the opinion. Occasionally, a footnote becomes controversial, as in Microsoft Corp. v. AT&T Corp., where three Justices concurred “as to all [of the majority opinion] but footnote 14.” 306550 U.S. 437, 459 (2007) (Alito, J., concurring).

Perhaps Justices put content in footnotes to downplay or hide the content from contemporary readers. Indeed, footnotes could be hidden “timebombs” that might be used aggressively by future Courts. The most famous footnote in Supreme Court history is footnote 4 of Carolene Products, 307United States v. Carolene Prods. Co., 304 U.S. 144, 153 n.4 (1938). which held that the Court should give more searching review for certain types of legislation, such as laws aimed at “discrete and insular” minorities. This footnote has taken on far greater significance than the main holding in the case. 308J.M. Balkin, The Footnote, 83 Nw. U. L. Rev. 275, 281–82 (1989). Other footnotes have likewise assumed great significance in later cases. 309 See, e.g., Dirks v. SEC, 463 U.S. 646, 655 n.14 (1983) (noting that certain outsiders, including lawyers and auditors, could be considered insiders for insider-trading liability); Ernst & Ernst v. Hochfelder, 425 U.S. 185, 193 n.12 (1976) (providing the basis for a finding that recklessness could satisfy the scienter requirement of section 10(b) of the Securities Exchange Act of 1934).

Some have suggested that footnotes are used to weaken the content of an opinion. One critic observed: “Just think about the last time you read a confident assertion by a judge or law professor, only to be let down by an accompanying note warning, ‘But see . . . .’” 310Adam Freedman, Footnotes: The ‘Insidious Plague’ That Helps Lawyers Explain ‘Fractured’ Reality, N.Y. L.J., Dec. 8, 2008, at S6. Ken Lasson criticized textual footnotes for allowing a writer to take “a strong position in the text while waffling below.” 311 Id. A defender of footnoting suggests that the “footnote will often be an appropriate place for the opinion-writer to set forth his or her doubts about the state of the law or the legal precept being announced.” Edward R. Becker, In Praise of Footnotes, 167 F.R.D. 283, 287 (1996). Perhaps footnotes are the Justices’ way of hedging their bets. If so, opinions with more footnotes might be weaker and receive fewer citations. 312 See Posner, Challenge, supra note 266, at 236 (observing that at the circuit court level “the more footnotes an opinion has (holding the length of the opinion constant), the less likely it is to be cited, perhaps because footnotes make it more difficult for readers to extract a clear holding from an opinion”).

The significance of footnotes may even be contested. At least symbolically, a footnote “is of minor importance.” 313Balkin, supra note 308, at 276. The Bankruptcy Appellate Panel of the Second Circuit wrote—in a footnote—that “federal courts are not to consider the footnotes to an opinion as authority.” 314 Breeden v. Sprague Nat’l Bank (In re Bennett Funding Group, Inc.), 44 Collier Bankr. Cas. 2d (MB) 151, 157 n.7 (B.A.P. 2d Cir. 2000). This is generally not the case for judges, however, and certainly not the case for the Supreme Court, as we have seen how some footnotes have assumed great precedential significance. Nevertheless, the expression of a footnote may carry less persuasive weight than similar language found in the text.

As a rule, footnoting is denounced by commentators and judges. 315 See, e.g., Arthur J. Goldberg, The Rise and Fall (We Hope) of Footnotes, 69 A.B.A. J. 255 (1983) (“Footnotes . . . cause more problems than they solve.”); Abner J. Mikva, Goodbye to Footnotes, 56 U. Colo. L. Rev. 647, 647 (1985) (declaring the footnote to be an “abomination”); In Justice Breyer’s Opinion, A Footnote Has No Place, N.Y. Times, July 28, 1995, at B18 (summarizing Justice Breyer’s antipathy for footnotes). Justice Powell wrote that “[a] frequent and justified criticism of [the Supreme] Court is that opinions . . . are overburdened with footnotes.” 316Memorandum of Briefing Notes from Supreme Court Justice Lewis F. Powell to Law Clerks 20 (Sept. 10, 1984) (on file with the Washington and Lee University Law School). Judge Mikva complained that “footnotes frequently project issues into the case that don’t have to be there.” 317 Domnarski, supra note 292, at 195 (quoting Abner J. Mikva, Honorable Abner J. Mikva Oral History, in Historical Society of the District of Columbia Circuit Oral History Program 211 (1996)).

Others offer support to footnoting. 318 See, e.g., Becker, supra note 311, at 283 (suggesting that “well-conceived and well-crafted footnotes are valuable tools” for judges). Judge Becker contends that the use of footnotes enables a judge to write a more readable opinion for different audiences and provide a fuller understanding of the nuances of the case. Id. at 285–86. Much of the discussion of footnotes in judicial opinions dwells on aesthetic issues, and we hope to give some substance to their evaluation. We use a measure for footnote ratio, which simply represents the number of words in footnotes, divided by the total words of the opinion (FN Ratio). 319We obtained these data from Black & Spriggs, supra note 112. We employ this measure as another independent variable to evaluate whether greater use of footnote content affects the precedential power of an opinion.

E. Additional Controls

In addition to the above determinants, we consider additional control variables, which could influence the power of an opinion in future citations and correlate with our other independent variables of interest. The first of these is the number of amici who join the case before the Supreme Court, called Amici. 320We obtained these data from Hansford & Spriggs, supra note 5, at 62. Previous research has used this variable for different purposes, including the study of amici themselves. There is ample evidence that amicus briefs have an influence on the Court. More amici are associated with greater dissensus among the Justices. 321Paul M. Collins, Jr., Amici Curiae and Dissensus on the U.S. Supreme Court, 5 J. Empirical Legal Stud. 143 (2008).

The primary use of the variable measuring the number of amici is as a test of the salience of the case. If a case is more legally significant, more parties are likely to expend the resources necessary to file an amicus brief. Researchers have therefore used this measure to assess the legal or political salience or complexity of a case. 322 See, e.g., Maltzman et al., supra note 217, at 45–46 (using this measure for political salience of a case); Virginia A. Hettinger et al., Comparing Attitudinal and Strategic Accounts of Dissenting Behavior on the U.S. Courts of Appeals, 48 Am. J. Pol. Sci. 123, 126 (2004) (same).

Our use of amici as a control variable thus may allow us to separate out the intrinsic significance of the legal issue to the case, which would have an obvious effect on its future citations. The New York Times front-page coverage measure fails for our purposes because it is a post facto measure of the significance of the opinion itself, not that of the underlying case (before the opinion is rendered). Number of amici is therefore a better measure of the legal salience of the underlying dispute, which is what we need for our control variable (Amici). If this is the case, we expect that Amici should be associated with more future citations.

Our second control variable considers the overruling of past precedents. When a precedent is overruled, its probability of citation would obviously decline, independent of the features of the opinion measured by our variables. To avoid having this confound the accuracy of our results, we create a variable, Overruled, for cases that have been overruled and another, Overruling, for the opinions that overruled them. We expect Overruled cases to have fewer citations, while Overruling opinions might have more citations. 323 See Fowler et al., supra note 15 (using data collected from Shepard’s Citations Service).

A third control variable is a dummy variable for opinions that held a federal law to be unconstitutional, called Unconstitutional. 324We determined whether a case struck down a federal statute as unconstitutional using Spaeth et al., supra note 188. These are the opinions that represent the classic instances of judicial activism. 325Cass Sunstein thus notes that “it is best to measure judicial activism by seeing how often a court strikes down the actions of other parts of government, especially those of Congress.” Cass R. Sunstein, Radicals in Robes 42–43 (2005). Political scientists likewise conclude that the “most dramatic instances of a lack of judicial restraint—or, conversely, the manifestation of judicial activism—are decisions that declare acts of Congress . . . unconstitutional.” Segal & Spaeth, Attitudinal Model Revisited, supra note 5, at 413. As prominent activist decisions defining the scope of the Constitution and the boundaries of legislative action, we would expect them to receive more citations.

For a fourth control, we identify whether the opinion was issued unsigned and per curiam, called Per Curiam. 326We identified per curiam opinions using Spaeth et al., supra note 188. We identified per curiam opinions using Spaeth’s variable labeled “mow,” which lists the author of each majority opinion. These opinions tend to be brief, more perfunctory, and less controversial. Existing research shows that per curiam opinions are generally less likely to be cited or legally interpreted than signed opinions. 327 Hansford & Spriggs, supra note 5, at 64; see also Frank B. Cross et al., Citations in the U.S. Supreme Court: An Empirical Study of Their Use and Significance, 2010 U. Ill. L. Rev. 489, 539. The role of the per curiam opinion has changed over time, shifting from procedural decisions to those on the merits, and later admitting of dissensus among the Justices. 328 See Laura Krugman Ray, The Road to Bush v. Gore: The History of the Supreme Court’s Use of the Per Curiam Opinion, 79 Neb. L. Rev. 517, 521–30 (2000). Some per curiam opinions may be quite significant. 329For example, Brandenburg v. Ohio set a major First Amendment precedent via per curiam opinion. 395 U.S. 444 (1969) (overturning the Ohio Criminal Syndicalism statute, overruling Whitney v. California, 274 U.S. 357 (1927), and articulating a new test—the “imminent lawless action” test—for judging so-called seditious speech under the First Amendment). Nevertheless, on balance, we expect per curiam opinions to be relatively less significant dispositions at the Court, associated with fewer future citations.

F. Results

This section presents the results of our analysis of what factors cause Supreme Court opinions to be relatively more or less powerful in terms of future citations. We use four dependent variables, each of which captures a different aspect of the cumulative significance of a case as of 2005. Our data set thus includes a single observation for each Supreme Court case decided between the 1946 and 2004 terms of the Court (n = 6,661 cases). Three of our dependent variables are citation counts, consisting of the total number of citations to an opinion by majority opinions of, respectively, the Supreme Court, the circuit courts, and the district courts. 330Whereas in Part II we listed the most significant cases in terms of citations to a case from all subsequent opinions (majority, concurring, and dissenting), the dependent variables in the following models are counts of citations to a case in majority opinions only. Our fourth dependent variable is the legal relevance score for the Supreme Court citations discussed above.

The legal relevance score is a reasonably continuous variable (it is, as explained above, a percentage), for which we use OLS linear regression. Because the other measures are count variables, the statistical analysis is a negative binomial regression. Table 7 reports the coefficients for the independent variables with conventional designations of statistical significance. 331The “*” represents statistical significance at the .05 level (two-tailed test). The number of observations (i.e., cases) is 6,661. We first indicate which variables achieve statistical significance, meaning we can reasonably conclude that the observed association between the dependent variable and them are not due to random covariation. More importantly, we then discuss the magnitude of the relationship between those variables with statistical significance and our dependent variables. Ultimately, of course, we are most interested in the extent to which the factors we examine lead Court opinions to be more or less legally important.

Table 7:

Determinants of Influence

Legal Relevance Score

Supreme Court Citation

Circuit

Court Citation

District Court Citation

Civil Liberties

.1002*

.2026*

.4211*

.0918

Economic

-.0558*

-.1644*

-.3950*

-.7545*

Constitutional

.0726*

.0100

-.3038*

-.7634*

Other Cases

.0808*

.0393

-.3029*

-.6644*

Complexity

.0254*

.0699*

-.0319

-.1012*

Age

.0221*

.1200*

.0846*

.1162*

Age-Squared

-.0003*

-.0014*

-.0013*

-.0019*

Liberal Precedent

.0124*

.0341

-.0370

-.0091

Coalition Homogeneity

-.0208*

-.0081

.0582

.1020

Author Ideological Extremity

.0003

-.0163

-.0275

-.0052

Unanimous

-.0264*

-.0683*

-.0944

-.0035

MWC

-.0013

-.0584

-.0739

.1197

Citations

.0050*

.0276*

.0190*

.0226*

Majority Length

.000003*

.00003*

.00008*

.00009*

Separate Length

.000005*

.00002*

.00003*

.00003*

FN Ratio

.0154

-.1001

-1.037*

-1.283*

Amici

.0202*

.0659*

-.0525*

.0465

Overruled

.0690*

.2127*

.2804*

.4297

Overruling

.0358*

.3713*

.5416*

.5212*

Unconstitutional

.0569*

-.1011

-.1313

-.3753*

Per Curiam

-.1934*

-.9458*

-1.222*

-1.181*

Constant

.2107*

-.9089*

3.4759*

3.980*

The comparative results for the four tests reveal some interesting findings. There are a few surprising differences between the raw number of Supreme Court citations and the authority score that builds on those citations. Liberal opinions and rulings on constitutional issues have significantly higher authority scores but not significantly more direct citations. They apparently produce opinions that are cited by cases that are themselves more important.

There are even more differences between the impact of opinions in the Supreme Court and in lower courts. The direction of the effects is occasionally different. Constitutional cases are, perhaps unsurprisingly, more profound at the Supreme Court level. Nevertheless, there are significant commonalities between the factors driving Supreme Court and lower court significance, which indicate that the selectivity of the certiorari process does not produce such great differences. We review our determinants below.

1. Case Characteristics

Case type has a plain effect on the significance of precedent. The civil liberties issue area produces more significant precedents for all but district court citations. The strong effect at the Supreme Court level may simply be an agenda issue, as the Court takes many of these cases. The effect at the circuit court level, though, demonstrates that civil liberties precedents are indeed especially important in the law.

The results for legal area, however, are not as expected. Although precedent is expected to be stronger for economics opinions, precedent consistently has a weaker effect in this area of the law. This could be a reflection of the elusive settled case phenomenon discussed above. Maybe there are fewer cases simply because the parties realize that binding precedent clearly governs their disputes and cannot be effectively challenged.

The relative effect of precedent in statutory interpretation opinions is mixed. Statutory precedents are weaker at the Supreme Court level (as reflected by the significant positive effect for constitutional and other cases). However, these precedents are significantly stronger at the circuit court and district court levels (as reflected by the significant negative effect for constitutional and other cases). For most cases, statutory precedents are more powerfully significant, and the contrary results at the Supreme Court level are probably an artifact of the certiorari selection effect.

2. Age

Various previous studies have identified the importance of an opinion’s age on the power of precedent, and our research confirms the findings of those studies. Age is significantly positive, but age-squared is significantly negative—meaning that older cases generally have a larger number of citations but this effect flattens out for very old cases. These findings are true by all metrics. Comparing the coefficients for the two variables shows that precedents assume significance fairly rapidly but then slowly decline in importance. This expected finding demonstrates the importance of considering the age of an opinion as an important factor when measuring other influences.

3. Ideological Factors

Our observation of the effects of ideology on the importance of opinions is new and more revealing. Liberal opinions have significantly greater legal relevance scores, though they do not receive significantly more citations at any court level (and receive slightly fewer citations from the circuit and district courts). Taking these results together suggests the following: while liberal opinions do not acquire more citations over their lives, they are cited by cases that themselves are more significant at the Supreme Court. We can draw this inference because the difference in the raw citation count at the Court and the legal relevance score results because the latter takes into account the “quality” of the cited and citing cases. 332The legal relevance score is based first, on the quantity of citations, meaning the total number of citations a case has received. Each of those citations is then weighted by the “quality” of the citing case, which is assessed based on the number of citations in that precedent, where each of those citations is weighted by its quality. See Fowler et al., supra note 15, at 330–32. In addition, this presumably does not reflect the hypothesized ratchet effect, which would also show up in the raw citation numbers. It appears that certain liberal precedents, probably including key Warren Court opinions, have become central to our law in the Supreme Court.

The results for the effect of ideological homogeneity are contrary to those hypothesized. More homogenous coalitions produce opinions with less significance by the network measure and no significant difference in terms of raw citations. Author extremity also produces no significant results. Although one might expect more ideological coalitions or authors to produce more dramatic and important opinions, such opinions must gain acceptance by future judges and Justices to have an impact. It appears this is not the case, and the negative network results for ideological homogeneity suggest that these coalitions may be inclined to overplay their hands.

4. Opinion Characteristics

The most interesting findings are associated with characteristics of majority opinions. The results for coalition size are contrary to the general understanding. The results indicate that cases with unanimous coalitions are less significant at the Supreme Court level but not in the lower federal judiciary. The importance of cases with minimum-winning coalitions, however, does not differ from other cases. The oft-hypothesized greater power of a unanimous opinion is not true as a general matter (though it still could be true for individual cases). Nor are the highly controversial cases decided by minimum-winning coalitions more influential. In short, unanimous opinions have less clout at the Supreme Court.

The number of citations contained in an opinion is consistently positive and significant. Perhaps these opinions are better grounded in the existing law, or perhaps they are simply more persuasive by virtue of greater expressed precedential support. The association occurs in every metric, which is strong evidence of an effect.

A similar strong positive result is seen for opinion length, even after controlling for the number of citations in the opinion (which would influence length) and the length of separate opinions (which can reflect case salience). This finding appears to rebut Judge Posner’s suggestion that longer opinions are less useful. 333 See Posner, Challenge, supra note 266, at 146–47; cf. Stephen J. Choi et al., Judicial Evaluations and Information Forcing: Ranking State High Courts and Their Judges, 58 Duke L.J. 1313, 1322 (2009) (“[J]udicial decisionmaking is better when it is conservative and minimalist.”). The reasons for this relationship are not entirely clear. Greater length may simply reflect greater effort by the opinion author. Alternatively, the association may be due to the simple fact that longer opinions contain additional language on which later courts may rely. It may thus be an expression of a Justice’s desire to project greater influence over future development of the law. The length of separate opinions is also consistently positive.

The results for footnote ratio are also telling. Although putting language in a footnote is hypothesized to reduce the influence of an opinion, it does not have this effect at the Supreme Court level. The Justices are not influenced by whether language is in the body or a footnote to the opinion. At the circuit court and district court levels, however, footnote ratio is clearly negative in its effect. The lower courts apparently view footnotes as a signal that language is less important, even though the Supreme Court Justices themselves do not do so. 334This effect is not certain because we do not have a record of citations to the footnotes themselves. However, it is a highly plausible inference, absent a persuasive reason why more language in footnotes would somehow demean the power of the language in the body of the opinion. Finally, cases of greater legal complexity acquire more citations and greater legal relevance at the Court, but they are actually cited less often in the district courts.

The findings for opinion content must be taken with a caveat. It is possible that a given Justice tends to write longer opinions with more citations. That same Justice may write more powerful precedential opinions, for reasons unrelated to the length and citations of the opinion. If so, the true cause would be the effect of the Justice authoring the opinion, not the length of and citations contained in the opinion. We will explore the possibility of such Justice effects below.

5. Controls

The control variables also contain interesting information. Amici is positive and significant at the Supreme Court level under both measures, suggesting that it is a proxy for legal or political significance of the case at the Court. However, it is negative and significant at the circuit court level but positive and marginally significant at the district court level. The latter findings are curious and of unknown meaning.

The finding for overruled cases is interesting—cases that have been overruled have unusually great significance in the number of citations they receive and their authority score. This superficially anomalous finding presumably reflects reverse causation. They are not significant because they were overruled; they were overruled because they were significant (and undesirable to a later Court). The Court apparently will not overrule an ordinary error of precedent, just a major one. In fact, this result is consistent with existing work showing that the Supreme Court is most likely to negatively interpret or overrule cases that are both ideologically distant from the Justices and that possess greater legal vitality. 335 See Hansford & Spriggs, supra note 5, at 84–91. This is confirmed as well by the consistent significance of the overruling variable, in that cases that overrule precedent are more significant at all levels of the federal judiciary.

Cases finding federal laws unconstitutional are significantly positive for Supreme Court legal relevance scores but not for other variables, and significantly negative for district court citations. This is roughly consistent with the findings for constitutional precedents, which are important at the Supreme Court level but not so significant for use by lower courts, given the different types of cases heard by different tiers of the federal judicial system.

Per curiam opinions were consistently less significant at all court levels, as expected. Such opinions tend to be brief with less material to be cited as authority. In addition, the Court may decide cases per curiam simply because they are less controversial or significant in the law.

G. Substantive Import

The statistical significance identified in the above section does not evidence substantive significance—the magnitude of the effect of a given independent variable on variation in the dependent variable. To depict the substantive significance of our independent variables, we create expected citation rates for changes in each of these variables while holding all other variables at their average. Table 8 displays the results of this calculation for the associations we found to have statistical significance in Table 7.

Table 8:

Effect Size of Independent Variables on Citation Rates

Supreme Court Citation

Circuit Court Citation

District Court Citation

Baseline Citation Rate

9.0

268.1

479.4

Not Civil Liberties

7.4

176.3

Economic

7.6

180.8

226.8

Constitutional

198.7

225.2

Other Cases

198.8

246.8

Low Complexity

8.5

521.7

High Complexity

9.5

440.8

Young Age

3.5

168.9

274.2

Old Age

11.1

215.8

309.8

Conservative Case

Coalition Homogeneity

Low Author Extremity

High Author Extremity

Unanimous

8.4

MWC

Low Citations

6.7

217.8

373.4

High Citations

11.9

324.2

602.5

Short Length

8.1

210.8

371.0

Long Length

10.0

341.6

620.9

Short Separate Length

8.7

255.9

456.5

Long Separate Length

9.9

355.2

666.0

Low FN Ratio

229.6

394.6

High FN Ratio

268.1

479.4

Few Amici

8.4

284.2

Many Amici

9.6

254.3

Overruled

13.1

466.7

Overruling

11.2

359.4

762.6

Unconstitutional

332.9

Per Curiam

3.5

80.1

154.6

We calculated the effect sizes for changes in the variables by using stochastic simulations as implemented by the CLARIFY program. 336Michael Tomz et al., CLARIFY: Software for Interpreting and Presenting Statistical Results, Harvard (June 1, 2001), http://gking.harvard.edu/clarify; see also Gary King et al., Making the Most of Statistical Analyses: Improving Interpretation and Presentation, 44 Am. J. Pol. Sci. 347 (2000) (discussing software for statistical analysis, including the authors’ own CLARIFY software). The baseline predicted citation rate was for an authored, liberal, non-economics, statutory, civil liberties opinion, decided neither by a unanimous coalition nor a minimum-winning coalition, which did not overrule precedent or strike a statute and which was not overruled. For specific values for dummy independent variables (such as Civil Liberties) one can compare the “Baseline Citation Rate” to the citation rate for the particular value of that variable (i.e., “Not Civil Liberties”), which then tells one how much the citation rate changes when moving from one category to the other in that variable. For the continuous variables, we estimate the citation rate for one standard deviation above and below the mean for that independent variable (e.g., “Short Length” and “Long Length” for the variable, Majority Length).

At the Supreme Court level, some fairly sizeable effects can be found for age, certain types of cases, and opinion characteristics, especially the number of citations contained in the opinion to be cited and opinion length. For instance, a case that cites a relatively small number of precedents acquires about 6.7 subsequent cites over its life, while one that references a larger number of precedents receives approximately 12. Cases that overrule precedent are cited about 13.1 times, while those that do not only get about 9 total citations. All federal courts appear to have a much greater preference for longer Supreme Court opinions.

Although many of our determinants have a modest net effect at the Supreme Court level, a few combined factors together may have a great impact. To illustrate this, we display the expected number of Supreme Court citations to an opinion for three scenarios. Scenario 1 is a new, short, economics per curiam opinion with few internal citations to prior opinions. Scenario 2 represents the average case in our data (which is an older, lengthier, authored opinion in the area of economics, with the average number of citations to prior cases). Scenario 3 is an older civil liberties case, with a longer opinion and relatively more citations to precedent. Figure 6 shows the difference in expected Supreme Court citations.

Figure 6: Predicted Supreme Court Citations (Three Scenarios)

cross-spriggs-fig6

Our per curiam opinion would likely get only a single citation, with the average Court opinion receiving nearly 9 citations, and the longer, heavily cited civil liberties opinion receiving fifteen citations.

Next we consider the expected citations at the circuit court level. One should note that the magnitude of the relationships between various case characteristics and case significance is consistently meaningful. For instance, a Supreme Court case that contains relatively few citations to precedent will receive about 218 subsequent citations in the courts of appeals, while one with a larger number of citations will be cited over 324 times. We also present, in graphical form, the combined influence of several factors for the same three opinion scenarios. The results are presented in Figure 7.

Figure 7: Predicted Circuit Court Citations (Three Scenarios)

cross-spriggs-fig7

The great disparity in citation rates for the three scenarios remains. The absolute difference in citations at the circuit court level, though, is in the hundreds of opinions (each of which has its own progeny effect within the circuit).

Figure 8 performs the same comparison of our three scenarios for a prediction of district court citations.

Figure 8: Predicted District Court Citations (Three Scenarios)

cross-spriggs-fig8

The relative difference remains roughly the same, with the difference measured in hundreds of opinions.

The preceding analysis shows the great significance of various determinants of opinion influence in the legal network. However, one potentially important factor has not yet been considered—the influence of the opinion’s author. The theory and language of the opinion is surely significant to its power as a precedent, and these are at least somewhat within the control of the Justice writing the opinion. The following Part examines the effect of opinion authors.

IV. Individual Justices

When a Justice drafts a majority opinion, he or she has choices in how to write. Those choices may yield opinions of greater or lesser future precedential significance. Justices may have more relative concern for the consequences of their opinions or may have a greater aptitude for writing important opinions. Justices have their own “styles” of opinion writing, which may prove more or less influential. 337 See generally Posner, supra note 36 (discussing the effects of various judicial writing styles). The subsequent impact of an opinion can be influenced by “the care with which the opinion is drafted.” 338Lax & Cameron, supra note 261, at 282. Beyond mere care, we expect that some Justices are especially concerned for the power of their opinions, and they may be expected to draft them accordingly. While some opinions may be drafted to increase their precedential power, others may be drafted in a conscious attempt to avoid having such power. 339 E.g., Bush v. Gore, 531 U.S. 98, 109 (2000) (per curiam) (“Our consideration is limited to the present circumstances . . . .”).

The control of the opinion author is not wholly unconstrained, as the authoring Justice must retain the votes of other Justices in the majority. 340 See Gerhardt, supra note 32, at 62 (suggesting that the “building of coalitions” means that the opinion’s “content is a function of the majority’s preferences”). The opinion may not even reflect the sincere views of its author. In Craig v. Boren, 341429 U.S. 190 (1976). for example, internal records show that Justice Brennan’s preferred position was to hold gender discrimination to a strict scrutiny standard like race. However, he found that this position would not command a majority of the Court and therefore adopted an intermediate scrutiny standard. 342 Epstein & Knight, supra note 52, at 5–10. Much of the content of the opinion was driven by Justices other than its author, Justice Brennan. He conceded that he changed his opinion in other cases as well. 343 See Maltzman et al., supra note 217, at 94.

The opinion speaks for the entire majority coalition, and other Justices may well have had input into its contents and future significance. 344Chief Justice Rehnquist observed that decision making “inevitably has a large individual component,” but that it is “filtered through the deliberative process of the court as a body.” William H. Rehnquist, Remarks on the Process of Judging, 49 Wash. & Lee L. Rev. 263, 270 (1992). In traditional spatial models, the product of the Court is inevitably controlled by the vital fifth vote for a majority opinion. 345 See Pablo T. Spiller, The Choices Justices Make, By Lee Epstein & Jack Knight, 94 Am. Pol. Sci. Rev. 943, 943 (2000) (book review) (“Once the median policy is proposed, no other proposal will beat it, and it becomes the outcome.”). The general theory provides that “Supreme Court opinion authors make strategic calculations about the need to craft opinions that are acceptable to their colleagues on the bench.” Paul J. Wahlbeck et al., Marshalling the Court: Bargaining and Accommodation on the United States Supreme Court, 42 Am. J. Pol. Sci. 294, 294 (1998). The assigned opinion author must attract this vote for a majority opinion, so the median voter has considerable influence on the content of that opinion. Such a Justice might demand a more constrained opinion than that preferred by the assigned author. Some have suggested that it is the median voter on the Court who controls outcomes, so that it may not matter who writes the opinion. 346Lax & Cameron, supra note 261, at 276–77 (“If the Median Voter Theorem applies, the content of every Supreme Court opinion must devolve to the wishes of the median justice; the identity and preferences of the opinion’s author . . . cannot matter.” (citation omitted)). There is documentary evidence that majority-coalition Justices both respond to and demand changes in drafts of majority opinions. 347 See, e.g., James F. Spriggs II et al., Bargaining on the U.S. Supreme Court: Justices’ Responses to Majority Opinion Drafts, 61 J. Pol. 485 (1999). Sometimes, Justices explicitly refuse to join a draft opinion. Id. at 487–88. The vast majority of Justices in the original majority coalition simply join the opinion without challenge, however. Id. at 498 (noting that this is the case for over 80% of the Justices). When accommodations are made to other Justices, they reflect numerous concerns, rather than simply conforming to preferences of the median Justice. See Wahlbeck et al., supra note 345. The Justices share “bargaining statements” seeking to trade changes in the opinion language for their supportive votes. 348 See Epstein & Knight, supra note 52, at 58–79 (discussing bargaining over the opinion). Empirical evidence reveals that cases with minimum-winning coalitions and those with greater bargaining among the Justices are longer, which suggests the greater influence of other coalition members on the characteristics of the opinion. 349Black & Spriggs, supra note 112, at 662.

Nevertheless, there is reason to believe that the opinion remains substantially in the control of the opinion author. There are costs to opinion writing for each of the Justices, 350 See, e.g., Virginia A. Hettinger et al., Separate Opinion Writing on the United States Courts of Appeals, 31 Am. Pol. Res. 215 (2003) (discussing costs of separate opinion writing); Lax & Cameron, supra note 261 (modeling the opinion content based on the cost of writing separately). and those costs empower an assigned opinion author to exercise control over the content of an opinion, even if the median voter might prefer somewhat different language. 351See Lax & Cameron, supra note 261, at 277 (noting that if writing a Supreme Court opinion takes “costly time and effort,” this fact would create a “wedge” that the assigned opinion author could use “to move an opinion away from the median justice’s most preferred policy”). Justices may make opinion sacrifices to avoid the need to draft separately.

This author effect is confirmed empirically. The median voter theorem would suggest that the median would always be part of the majority opinion coalition, when in fact median voters issue a material number of special concurrences. 352Cliff Carrubba et al., Does the Median Justice Control the Content of Supreme Court Opinions? (Sept. 29, 2010) (unpublished manuscript), available at http://www.allacademic.com/meta/p360479_index.html. A more detailed empirical analysis of the probability of Justices joining the majority coalition found that both the opinion author and the median Justice had influence, but the identity of the opinion author was somewhat more powerful. 353Chris W. Bonneau et al., Agenda Control, the Median Justice, and the Majority Opinion on the U.S. Supreme Court, 51 Am. J. Pol. Sci. 890 (2007).

Even if the effect of the opinion author could be obscured to some degree by other majority coalition Justices, our analysis controls for this possibility in two ways. First, we include a variable for the ideological extremity of each opinion author (Extremity), measured as the absolute value of the difference between the opinion author’s ideological position and the ideological position of the median Justice on the Court in the year the precedent was decided. Second, we include a variable for the ideological homogeneity of the Justices in the majority opinion coalition of the precedent (labeled as Homogeneity). As we describe below, we also include a variety of additional control variables that may not be randomly distributed across the Justices (such as opinion length, with some Justices systematically writing longer opinions than others) and which also help explain citation patterns. By controlling for these variables, the results we find for each Justice are likely to be a conservative estimate of a Justice’s influence on the law. Suppose Justice Brennan had an authentically greater opinion-writing effect than his fellow Justices of the era. To the degree that those Justices influenced his opinion, or he theirs, that effect would cause our procedure to understate the power of Justice Brennan’s pen. 354Given the large number of control variables, the remaining errors are likely to be random. See Cross & Lindquist, supra note 93, at 1392 (“[E]rrors are likely to be randomly distributed throughout the judicial population and may therefore be considered statistical ‘noise.’ The presence of this feature tends to make it more difficult to find true statistical significance and consequently may add further confidence to results that find such significance.” (footnote omitted)); see also Staudt et al., supra note 185, at 376 (noting that measurement error “is absorbed into the disturbance term” so that the error causes an underestimate of true effects); Richard A. Posner, The Theory and Practice of Citations Analysis, with Special Reference to Law and Economics 12–13 (Univ. of Chi. Law Sch., John M. Olin Law & Econ. Working Paper No. 83, 1999), available at http://www.law.uchicago.edu/files/files/83.RAP_.Citations-40532.pdf (“Critics of citations analysis often fail to note that if errors in data are randomly distributed with respect to the variable of interest . . . they are unlikely to invalidate the conclusions of the study, provided that the data sample is large.”).

The Justices writing an opinion in an individual case are limited by the materials with which they have to work. Some cases simply deal with relatively unusual issues, unlikely to recur in a way that provides much opportunity for future citation. Other cases deal with dramatic, recurring facts that ensure at least some future citations. This will tend to produce only random noise, not systematically skew comparisons among Justices. Moreover, with enough opinions, the effect of this factor is likely to average out among the Justices.

Justices possess “many devices for reshaping case facts and law, and therefore significance.” 355Cook, supra note 1, at 1128. Thus, an opinion author may be able to turn a “little fish” of a case into a “choice morsel.” Id. Not every case may be transformed into a landmark decision, but there is surely the ability at the margin for a Justice to write a decision that may be more or less significant. Justice Fortas wrote: “If the Chief Justice assigns the writing of the opinion of the Court to Mr. Justice A, a statement of profound consequence may emerge. If he assigns it to Mr. Justice B, the opinion of the Court may be of limited consequence.” 356Abe Fortas, Chief Justice Warren: The Enigma of Leadership, 84 Yale L.J. 405, 405 (1975). There is reason to believe that the choice of opinion author is “highly consequential for the legal choices made by the Court.” 357Forrest Maltzman & Paul J. Wahlbeck, Opinion Assignment on the Rehnquist Court, 89 Judicature 121, 122 (2005).

Writing opinions is the pathway through which Justices can project their influence to other courts and into the future. Judge Posner notes that “precedent projects a judge’s influence more effectively than a decision” itself. 358 Richard A. Posner, Economic Analysis of Law 585 (7th ed. 2007). The decision affects only the parties, but the language of the opinion drives future judicial decisions and the practice of private parties. The opinion in Miranda surely produced a material change in the practices of policing. Other opinions, such as Roe, have likewise had significant societal effects.

While Justices may be interested in giving influence to their opinions and projecting power, they will not necessarily seek to maximize the impact of their holdings. Some Justices may be “minimalists.” 359 See Cass R. Sunstein, One Case at a Time: Judicial Minimalism on the Supreme Court (1999) (setting out the theory of judicial minimalism). They do not base their decisions on grand theories, nor do they establish all encompassing rules to resolve cases. Minimalist decisions tend to be narrow and shallow, rather than wide and deep. 360The minimalist Justice strives to “resolve the problem at hand without also resolving a series of other problems that might have relevant differences.” Sunstein, supra note 325, at 29. Sunstein suggests that minimalism is “the phenomenon of saying no more than necessary to justify an outcome, and leaving as much as possible undecided.” 361 Sunstein, supra note 359, at 3. This difference has previously been captured as a distinction between innovators and interpreters. 362 See J. Woodford Howard, Jr., Courts of Appeals in the Federal Judicial System: A Study of the Second, Fifth, and District of Columbia Circuits 160–62 (1981). Others have characterized the maximalist judge as a “judicial entrepreneur.” 363 See, e.g., Cynthia L. Cates & Wayne V. McIntosh, Retail Jurisprudence: The Judge as Entrepreneur in the Marketplace of Ideas, 11 J.L. & Pol. 709 (1995) (discussing the notion of judicial entrepreneurship).

Sunstein identifies some contemporary Justices as minimalists (Justices Breyer, Ginsburg, Kennedy, O’Connor, and Souter). 364 Sunstein, supra note 359, at 9. Sunstein focuses on the contemporaneous Court, but historic Justices might be similarly categorized. See, e.g., Sheldon Gelman, The Hedgehog, the Fox, and the Minimalist, 89 Geo. L.J. 2297, 2303 n.38 (2001) (reviewing Sunstein, supra note 359) (describing Justice White as minimalist in orientation). Others, such as Justices Scalia and Thomas, he characterizes as “fundamentalists” who “seek to make large-scale changes in constitutional law.” 365 Sunstein, supra note 325, at 26. Because minimalist decisions are more specific (less general), they would be expected to result in fewer future citations. 366 See Landes & Posner, supra note 8, at 268 (describing why more specific precedents will depreciate faster than more general ones). If a Justice leaves things undecided, in his or her minimalism, those undecided matters will not offer opinion language to be cited. By “saying no more than necessary to justify an outcome,” the minimalist leaves less for future Courts to cite. 367 Sunstein, supra note 359, at 3. However, this may not be the case. If a maximalist opinion does not receive respect, it “will not control the future.” 368 Id. at 19. Thus, maximalists “may be quite surprised by the conduct of subsequent courts, which characterize prior language as ‘dicta’ . . . . [that] turns prior decisions into minimalist ones.” Id. at 20. The more dramatic fundamentalist opinions may overreach and produce results unacceptable to future Courts, addressing different sets of facts. 369 See Choi et al., supra note 333, at 1322 (contending that if minimalist decisions indeed produce better law, “minimalist opinions will be cited more, not the creative and expansive ones”).

Some empirical research has suggested that opinion characteristics do indicate minimalist tendencies for some Justices. 370 See, e.g., Robert Anderson IV, Measuring Meta-Doctrine: An Empirical Assessment of Judicial Minimalism in the Supreme Court, 32 Harv. J.L. & Pub. Pol’y 1045 (2009). One author examined the tendency of the Justices to join opinions of other Justices, regardless of the ultimate outcome of the cases. He found that most of the Rehnquist Court Justices, including the Chief Justice, were generally minimalists, with the noteworthy exceptions being Justices Thomas and Scalia. 371 Id. at 1063–68. This generally confirms Sunstein’s hypotheses about the nature of the contemporary Justices. This study considered only opinion joining, though, and not the future significance of opinions. The relative legal effect of minimalism remains untested.

There may be some biases to an evaluation of Justice effects on future citations. The opportunity to write an opinion is not random, of course, but assigned by the Chief Justice or senior Justice of the majority. Hence, “to the degree the chief retains control over particularly important cases, his opinions may be more frequently cited than others.” 372Kosma, supra note 8, at 340. Consequently, there may be a pro-Chief Justice bias in these measures, given the prospect of self-assignment of important decisions. 373The self-assignment effect has been studied, with early research finding that Chief Justices tend to assign themselves important cases as well as those decided unanimously. Elliot E. Slotnick, The Chief Justices and Self-Assignment of Majority Opinions: A Research Note, 31 W. Pol. Q. 219, 225 (1978). These findings were confirmed in Saul Brenner, Strategic Choice and Opinion Assignment on the U.S. Supreme Court: A Reexamination, 35 W. Pol. Q. 204 (1982). The significance of this effect is modified, though, by the Chief Justice’s need to balance other goals, such as Court harmony. See Forrest Maltzman & Paul J. Wahlbeck, A Conditional Model of Opinion Assignment on the Supreme Court, 57 Pol. Res. Q. 551 (2004). Opinion assignments may also be influenced by a desire to punish or reward members of the Court. Cross & Lindquist, supra note 268, at 1673. External perception of the opinion may also influence opinion assignment. See David W. Rohde, Policy Goals, Strategic Choice and Majority Opinion Assignments in the U.S. Supreme Court, 16 Midwest J. Pol. Sci. 652, 677–78 (1972). Moreover, the Chief may need to assign important cases to the most moderate member of the majority coalition in divided decisions in order to hold the majority. See Theodore S. Arrington & Saul Brenner, Testing Murphy’s Strategic Model: Assigning the Majority Opinion to the Marginal Justice in the Conference Coalition on the U.S. Supreme Court, 36 Am. Pol. Res. 416 (2008). The effect will also vary by Chief Justice; one study found that Chief Justice Rehnquist did not self-assign especially important decisions of his Court. Forrest Maltzman & Paul J. Wahlbeck, May It Please the Chief? Opinion Assignments in the Rehnquist Court, 40 Am. J. Pol. Sci. 421, 421 (1996). Opinions by other Justices are also influenced by assignment, but the fact that the assigning Justice chose a particular author for a particularly important opinion is in itself some testimony to that chosen Justice’s importance.

Our study considers only majority opinions, which qualifies the results somewhat. Some Justices may write particularly powerful or influential dissents, 374The second Justice Harlan, for example, has been characterized as a “great dissenter.” Tinsley E. Yarbrough, John Marshall Harlan: Great Dissenter of the Warren Court (1992). This might be ascribed to the fact that he was ideologically out of step with the liberal Warren Court majority. This honorific was also given to Justice Holmes. See William J. Brennan, Jr., In Defense of Dissents, 37 Hastings L.J. 427, 429 (1986); M. Todd Henderson, From Seriatim to Consensus and Back Again: A Theory of Dissent, 2007 Sup. Ct. Rev. 283, 284 n.7. an effect that we cannot capture. A Justice whose ideology (or legal theory) is out of sync with the prevailing majority may frequently be cast into dissent. 375For a quantitative measure of the Justices least likely to join majority opinions each year since 1956, see Peter A. Hook, The Aggregate Harmony Metric and a Statistical and Visual Contextualization of the Rehnquist Court: 50 Years of Data, 24 Const. Comment. 221, 244–46 (2007). When this Justice drafts majority opinions, they may be relatively uncontroversial ones, unable to receive many future citations, no matter how persuasively written. 376Our study controls for this bias somewhat, with the variables for minimum-winning coalition and unanimity. This may cause us to underestimate the potential opinion-writing ability of some Justices, but we still can capture the role of individual Justices in the network of law. While external circumstances may have conspired against some Justices, this does not alter the descriptive analysis of the power of particular majority opinions.

There are some limited stories about Justice effects. Justice Brennan, for example, has been described as “pervasively influential.” 377Bernard Schwartz, Supreme Court Superstars: The Ten Greatest Justices, 31 Tulsa L.J. 93, 148 (1995). These analyses are subjective and anecdotal, however.

One existing study attempted to measure the influence of Justices based on the Supreme Court citations received by their opinions. 378Kosma, supra note 8. It found that Justices Fuller, Waite, Holmes, the first Justice Harlan, and Gray were the most influential in the history of the Court. 379 Id. at 351. Of the then-sitting Court, Justice Rehnquist wrote the most influential opinions in terms of citations. 380 Id. at 353 n.42. Additional data enables us to expand and improve on this research. We can use the legal relevance score to better measure the influence of an opinion, and we add important other variables affecting an opinion’s significance for the more recent era in which these data are available.

Our task is to estimate the average citation frequency and “legal relevance score” for opinions authored by each Justice. To do so, we need to include control variables for factors that are likely to be correlated with these Justice-specific effects. For example, some Justices may write longer opinions than others, use more footnotes than others, or are assigned more salient or complex cases, and our above analysis indicates such factors correlate with citation frequency. If we did not control for these variables, then we would overestimate the influence of a given Justice.

For more recent opinions, roughly post-World War II, the Supreme Court Database, along with a few other prior studies on citation patterns, provides the necessary data to control for other determinants of opinion significance. Specifically, we include all of the variables in Table 7, along with a dummy variable for each Justice but one. This allows us to estimate a “fixed effect” for each Justice while simultaneously controlling for variables that are likely not randomly distributed across the Justices and that correlate with citation patterns. We first consider the association of individual Justices’ opinions with future Supreme Court citations, holding all the other variables constant at their mean (or their mode for a categorical variable). Figure 9 displays the expected citations for each of the Justices of the era, with 95% confidence intervals marked by smaller dots.

Figure 9: Justices and Predicted Supreme Court Citations

cross-spriggs-fig9

Some dramatic differences emerge. Justices Scalia and Thomas have very high predicted citation rates, though numerous recent Justices are also fairly high, while the Warren Court Justices are not high. Justice Marshall manifested the average predicted citation rate of 9.0. The Justices whose citation rates were statistically significantly (p ≤ .05, two-tailed test) greater than the average are Justices Thomas, Scalia, Ginsburg, Souter, Kennedy, O’Connor, Rehnquist, and Powell. The fact that Justices Scalia and Thomas are at the top of the list and Justice Breyer is relatively low is some evidence for minimalism/maximalism hypotheses, but today’s minimalists had higher citation levels than historic maximalists. Justices who were statistically significantly below the mean are Justices Whittaker, Minton, Clark, Burton, Fortas, Harlan, and Reed.

Legal relevance scores may be a better measure of the true significance of an opinion for future citations because they consider progeny effects. We produce predicted values for our authority scores, holding other variables at their average. Figure 10 sets out the estimates for the Justices of the era, with 95% confidence intervals marked by smaller dots.

Figure 10: Justices and Predicted Legal Relevance Scores

cross-spriggs-fig10

The differences among the Justices on the legal relevance score measure are relatively slight, when compared with citations. Only Justices Frankfurter, Murphy, Jackson, Rutledge, and Whittaker are statistically significantly different from the mean. The latter three have relatively few opinions in the data, and the results may be an artifact of a unique set of cases that was included in the analysis. Justice Breyer is low (though not statistically significantly different from the mean), perhaps a reflection of his minimalist decision making (though Justices Souter and Kennedy are often considered minimalists and have high predicted effects). The Justices of the Warren Court, who issued many important opinions, have only middling scores overall. That Court issued a larger number of total opinions per term, which may have included a number of less significant ones that dragged down the average of the Justices of the Court.

The reader should remember that these estimates include all of our other variables, such as opinion length, the number of cases cited in an opinion, and its age. Insofar as the opinion length and case citations are under the control of the opinion author, the figure may understate the significance of the author. It does measure the relative effect of the author’s language, though, on the significance of the opinion for future Courts.

As we discussed above, much of the significance of the law lies in the decisions of lower courts. We replicate our analysis in Figure 11 using the controls to produce the predicted number of circuit court citations for the opinions of each of the Justices.

Figure 11: Justices and Predicted Circuit Court Citations

cross-spriggs-fig11

As seen in Figure 11, Justice Jackson’s high level is quite striking, though the confidence interval is large (he had fewer opinions in the data than most others). Justices Scalia and Powell are quite high among the more modern Justices with more opinions to consider. The Warren Court Justices are again relatively low.

One interesting comparison is the relative positions of Justices in Supreme Court and circuit court citations measures. Justice Ginsburg was quite high for predicted Supreme Court citations (with statistical significance) but remarkably low for predicted circuit court citations (again with statistical significance). Justice Goldberg showed the opposite effect. This suggests that some Justices may write more for the Supreme Court than for lower courts, though in general there is an association between the two effects (e.g., the high level for Justice Scalia on both court level citations).

While there is a great deal of commonality in citation effects among the Justices of this period, some differences are apparent. The Warren Court Justices were not particularly significant in their effects on the citation network. A few Justices stand out for unusually significant effects, including Justice Scalia from the modern era. The contemporary Justices appear to have quite a high citation effect in general, but one must be cautious in drawing this conclusion. They have had a relatively high effect in the short term, but we do not yet know how the era’s opinions will stand the test of time.

V. Are These the Best Cases in the Supreme Court’s History?

Identifying the most legally important cases in the Supreme Court’s history offers important findings, though our definition of important is an internal one within the law, not necessarily overall societal effect. Yet the law provides an important reflection of society. If an opinion has no societal effect, it is unlikely to provoke litigation that would cause it to be cited by later courts. Conversely, a case with a large societal effect will often produce future litigation, applying the opinion or perhaps attempting to expand its scope, which will show up in our citation and network measures.

The most important opinions are not necessarily the best opinions of the Court. While precedential usefulness is one aspect of opinion quality, it may be that more specific decisions, employing minimalist decision making and fewer future citations, could be better opinions for particular cases. Nevertheless, opinion importance remains a key factor in the Court’s decisions. The opinion in Brown, for example, has seen considerable criticism regarding its legal reasoning. 381The classic of this genre is Herbert Wechsler, Toward Neutral Principles of Constitutional Law, 73 Harv. L. Rev. 1 (1959) (concluding that the opinion in Brown could not be justified as principled decision making). Some considered the opinion to be “a travesty of judicial craft.” Lani Guinier, Demosprudence Through Dissent, 122 Harv. L. Rev. 4, 134 (2008). The opinion’s reliance on children’s reactions to different racial dolls was both legally questionable and grounded in relatively bad social science. See Richard Kluger, Simple Justice: The History of Brown v. Board of Education and Black America’s Struggle for Equality 355–56 (1976) (addressing criticisms of the study relied upon); Michael Heise, Judicial Decision-Making, Social Science Evidence, and Equal Educational Opportunity: Uneasy Relations and Uncertain Futures, 31 Seattle U. L. Rev. 863, 867–68 (2008). Its importance to the Court, though, is obvious, and it is commonly regarded as one of the Court’s best holdings. 382 See Cross & Lindquist, supra note 93, at 1421. Even after criticizing the opinion in Brown, Wechsler wrote that it had “the best chance of making an enduring contribution to the quality of our society of any [opinion] that I know in recent years.” Wechsler, supra note 381, at 27.

The frequency of citations to a case has been used as a measure of the quality of the opinion in that case. 383 See, e.g., Choi & Gulati, supra note 8. Walter Schaefer, the chief justice of the Illinois Supreme Court, noted that “an opinion which does not within its own confines exhibit an awareness of relevant considerations, whose premises are concealed, or whose logic is faulty, is not likely to enjoy either a long life or the capacity to generate offspring.” 384Schaefer, supra note 109, at 11. On this theory a good opinion will be cited more than a bad one. Daniel Farber noted that a “judge whose opinions are consistently useful to others is probably doing something right, while a judge whose opinions are rarely cited is probably performing badly.” 385Farber, supra note 106, at 1179. Some dispute this hypothesis, suggesting that citations do not fully capture the true judicial virtues. See Cross & Lindquist, supra note 93, at 1391–93. Those virtues are explored in Lawrence B. Solum, The Aretaic Turn in Constitutional Theory, 70 Brook. L. Rev. 475 (2005). The citation measure is surely imperfect, but it has some validity and would appear to capture some of the key judicial virtues, while preserving objectivity. See Cross & Lindquist, supra note 93, at 1393–95. Hence, measures of citations may be considered “indirect indicators of judges’ ability to justify their decisions.” 386Knight, supra note 6, at 1553. Prior research found that individual Justices’ citation numbers correlated significantly with one subjective measure of judicial “greatness.” 387Kosma, supra note 8, at 360–62. While this citation proxy is an imperfect one, it provides something of a guide to quality as well.

A case that is cited more often enhances all the attributes of stare decisis in judicial decision making. It provides useful guidance to improve the Court’s efficiency. 388 See, e.g., Frederick Schauer, Precedent, 39 Stan. L. Rev. 571, 599 (1987) (discussing this efficiency rationale for precedent). Continued citations assist the equality and legitimacy rationales by allowing “courts to strengthen their reputation by promoting the perception that decisions are consistent over time.” 389Gely, supra note 215, at 107. When an opinion receives more citations, it is internally quite valuable to the Supreme Court’s legal system.

Those who prefer minimalist opinions might disagree with this conclusion. They would suggest that the better opinions may be the less definitive ones that produce fewer citations, leaving more open for the discretion of future judges and Justices. This is a theoretically plausible position, but it is not clear that minimalist opinions in fact produce fewer citations or are less important. Our measures do not clearly punish minimalist decision making, at least to any great degree. Most of the Justices regarded as minimalist had high legal relevance scores.

The notion of “best” cases implies a normative component that may seem to fit poorly within an empirical analysis such as this one. A case generally regarded as normatively very poor (e.g., Dred Scott v. Sandford 39060 U.S. (19 How.) 393 (1857) (holding that slaves did not enjoy constitutional protections of citizenship). or Korematsu v. United States 391323 U.S. 214 (1944) (sanctioning the curtailment of civil rights for one targeted racial group). ) might conceivably appear as an important one by our citation measurement. However, our citation count includes an implicit normative evaluation of the opinion by the judges and Justices who used it in subsequent opinions. The Supreme Court has thus abandoned the holdings in Dred Scott and Korematsu, though neither case has been explicitly overruled.

Our measure has some facial validity. The results in Tables 1 and 4 above contain lists of the Court’s opinions generally held in high normative regard. While these lists would not precisely match a subjective assessment of the best cases by individual observers, most observers would find the lists of most important cases to be reasonable ones by normative measures. The lists for lower courts, in Tables 2 and 3, are different and apparently reflect usefulness of an opinion for the cases that arise rather than their particular normative virtue. The Supreme Court standards, though, provide a plausible list of the opinions that are best, as well as those that are the most important for the network of precedent.

We cannot declare that a highly cited opinion is morally “best” in any sense. Perhaps a moral philosopher might identify a heretofore unknown opinion that best satisfied ethical values, a position that we do not consider. Our citation measure, though, reasonably captures the best opinions from a pragmatist’s perspective.

Conclusion

This research reveals the cases that are the most legally important in the history of the Supreme Court and some of the factors that make a case more important. There are surely other approaches to identifying the most important or best decisions issued by the Supreme Court. Our approach offers an empirical analysis, though, which provides some rigor to the measure and avoids the subjectivity associated with many other approaches. The approach yields some significant insights. There is certainly some random variation associated with which majority opinions receive the most citations and carry the most weight in the network of precedent. Citations will depend on the future cases taken by the courts, which depend in part on uncontrollable societal events. Despite this randomness, clear patterns still emerge.

Some types of cases seem to be intrinsically more significant by virtue of their subject matter. They deal with topics that are especially important in the law and more likely to recur. However, there are significant differences between the types of cases that are important at the Supreme Court level and at the lower court level. The same is true for the legal area addressed by precedent, with constitutional cases more important at the Supreme Court level but statutory precedents more powerful at the lower court level. Age has a clear effect on citations.

The effect of ideology is demonstrable at the Supreme Court, but it does not play a substantial role in citation practice. Liberal opinions show slightly more network power within the Court itself, but they do not receive more citations. Contrary to expectations, more ideologically homogenous opinions are actually weaker in their precedential effect. However, there is an ideological effect we do not capture here, which is that Justices who are ideologically distant from a precedent are less likely to cite it, but only for a relatively short period of time after the case is decided. 392 See Black & Spriggs, supra note 220.

The type of case is significant, but it is not the only factor driving the importance of the Court’s opinion. Some metrics of influence are within the control of the opinion author. Various opinion characteristics show consistently significant effects for each of our measures. Longer opinions and those with more citations have relatively more precedential power. This reveals a true opinion effect and indicates that the Justices have some influence over the subsequent power of their opinions. However, unanimous opinions are weaker.

Individual Justices also show differential impact for future citations, presumably because of the way they write. Justice Jackson, highly regarded by many, wrote majority opinions with great power (and our study did not even include powerful concurrences, such as that in Youngstown Sheet & Tube Co. v. Sawyer). On the present Court, Justice Scalia writes opinions that receive especially high citation rates. 393Research shows that for lower court usage, Justice Scalia’s opinions also have a disproportionately high rate of negative citations, though these remain a small fraction of his total citations. See Frank B. Cross, Determinants of Citations to Supreme Court Opinions (and the Remarkable Influence of Justice Scalia), 18 Sup. Ct. Econ. Rev. 177 (2010). Other Justices have unusually low rates. The differentials do not clearly trace the judicial minimalism/maximalism divide, though this may explain some of the differences.

Studies of citation rates can greatly advance our understanding of Supreme Court decision making and opinion writing. Citations are the central metric for assessing the significance of opinions, at least from a legal perspective. With modern data resources and statistical tools, we can evaluate many hypotheses about the Court from a more rigorous quantitative perspective. This research begins that process.

Footnotes

Frank B. Cross is Herbert D. Kelleher Centennial Professor of Business Law, McCombs School of Business, University of Texas at Austin; Professor of Law, University of Texas Law School; Professor of Government, University of Texas at Austin.James F. Spriggs II is Sidney W. Souers Professor of Government and Professor of Political Science, Washington University in St. Louis; Professor of Law (by courtesy), Washington University in St. Louis School of Law.

1 See Beverly B. Cook, Measuring the Significance of U.S. Supreme Court Decisions, 55 J. Pol. 1127, 1127 (1993) (“Decisions are not equal in significance . . . .”).

2 See Isaac Unah & Ange-Marie Hancock, U.S. Supreme Court Decision Making, Case Salience, and the Attitudinal Model, 28 Law & Pol’y 295 (2006) (finding that the influence of Justice ideology on decisions varies with case salience).

3Lee Epstein & Jeffrey A. Segal, Measuring Issue Salience, 44 Am. J. Pol. Sci. 66, 72 (2000).

4Richard A. Posner, An Economic Analysis of the Use of Citations in the Law, 2 Am. L. & Econ. Rev. 381, 382 (2000). The process “offers substantial promise of improving our knowledge of the legal system.” Id. at 402.

5For a critique of the research focus on outcomes, see Harry T. Edwards & Michael A. Livermore, Pitfalls of Empirical Studies That Attempt to Understand the Factors Affecting Appellate Decisionmaking, 58 Duke L.J. 1895, 1909 (2009) (noting that separate opinions reaching the same outcome may be very different in expressing the law). See also Thomas G. Hansford & James F. Spriggs II, The Politics of Precedent on the U.S. Supreme Court 3 (2006) (arguing that while case dispositions are important, “[t]he legal reasoning . . . can have more far-reaching consequences”); Martin Shapiro & Alec Stone Sweet, On Law, Politics, and Judicialization 98 (2002) (suggesting that “what judges say is even more important than how they vote”); Frank B. Cross et al., The Reagan Revolution in the Network of Law, 57 Emory L.J. 1227, 1234 (2008) (stressing that the significant aspect of a Supreme Court decision lies in the opinion, not the outcome, and that “[o]utcome-based research fails to capture the differential future significance of Supreme Court opinions”). Perhaps the leading empirical researchers of Supreme Court outcomes, Jeffrey Segal and Harold Spaeth, have conceded that it is the Court’s opinion that “constitutes the core of the Court’s policy-making process.” Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited 357 (2002) [hereinafter Segal & Spaeth, Attitudinal Model Revisited].

6Jack Knight, Are Empiricists Asking the Right Questions About Judicial Decisionmaking?, 58 Duke L.J. 1531, 1532 (2009). Knight notes the need to study “aspects of the opinions accompanying the votes.” Id. at 1533.

7 See Barry Friedman, Taking Law Seriously, 4 Persp. on Pol. 261 (2006). The article examines decisions on affirmative action and concludes that “looking to outcomes rather than opinions leads to the wrong conclusion of what the court ‘did.’” Id. at 266. Friedman compares Justices Rehnquist and Thomas and notes that their votes appear quite similar but “if one reads the decisions authored by these Justices, it is apparent that the two are quite different in ways that have great significance for the law.” Id. at 267.

8 See, e.g., Montgomery N. Kosma, Measuring the Influence of Supreme Court Justices, 27 J. Legal Stud. 333, 333 (1998) (using citation counts as a proxy for the influence of individual Justices); William M. Landes & Richard A. Posner, Legal Precedent: A Theoretical and Empirical Analysis, 19 J.L. & Econ. 249 (1976) (analyzing an opinion’s influence as a form of capital and measuring its depreciation over time through citation counting); John Henry Merryman, The Authority of Authority: What the California Supreme Court Cited in 1950, 6 Stan. L. Rev. 613 (1954) (the first empirical analysis of a court’s use of precedent, noting the effect of age on citation probability). A number of studies have used citations as a measure of the quality of circuit court judges. See Stephen J. Choi & G. Mitu Gulati, Choosing the Next Supreme Court Justice: An Empirical Ranking of Judge Performance, 78 S. Cal. L. Rev. 23 (2004); David Klein & Darby Morrisroe, The Prestige and Influence of Individual Judges on the U.S. Courts of Appeals, 28 J. Legal Stud. 371 (1999); William M. Landes et al., Judicial Influence: A Citation Analysis of Federal Courts of Appeals Judges, 27 J. Legal Stud. 271 (1998). The practice is also increasingly used in political science research in articles discussed throughout this Article.

9 See infra Part I.B.

10 The Oxford Guide to United States Supreme Court Decisions (Kermit L. Hall ed., 1999) [hereinafter The Oxford Guide]. The book describes its contents as a discussion of “the 440 most important cases in the Court’s history.” Id. at vii.

11 Joan Biskupic & Elder Witt, Congressional Quarterly’s Guide to the U.S. Supreme Court (3d ed. 1997). This source has been used in research as a guide to the most important decisions of the Court. See, e.g., Jeffrey A. Segal & Harold J. Spaeth, The Influence of Stare Decisis on the Votes of United States Supreme Court Justices, 40 Am. J. Pol. Sci. 971 (1996).

12James H. Fowler & Sangick Jeon, The Authority of Supreme Court Precedent, 30 Soc. Networks 16, 20 (2008).

13 See, e.g., Segal & Spaeth, supra note 11.

14 See Epstein & Segal, supra note 3, at 69.

15The initial case for the reliability of the measure is found in Epstein & Segal, supra note 3. The measure has been used in numerous subsequent articles, including Michael A. Bailey et al., Signals from the Tenth Justice: The Political Role of the Solicitor General in Supreme Court Decision Making, 49 Am. J. Pol. Sci. 72 (2005); Vanessa A. Baird, The Effect of Politically Salient Decisions on the U.S. Supreme Court’s Agenda, 66 J. Pol. 755 (2004); Paul M. Collins Jr., Towards an Integrated Model of the U.S. Supreme Court’s Federalism Decision Making, 37 Publius 505 (2007); James H. Fowler et al., Network Analysis and the Law: Measuring the Legal Importance of Precedents at the U.S. Supreme Court, 15 Pol. Analysis 324 (2007).

16 See Forrest Maltzman & Paul J. Wahlbeck, Salience or Politics: New York Times Coverage of the Supreme Court (Apr. 3–6, 2003) (unpublished manuscript) (on file with authors) (discussing the geographic bias). Maltzman and Wahlbeck also suggest that coverage is affected by the number of votes in the majority and whether the Chief Justice wrote the opinion, among other biasing factors. Id.

17 See, e.g., Cook, supra note 1 (discussing assorted measures of Supreme Court opinion significance).

18For a review of the constitutional canon in casebooks, see J.M. Balkin & Sanford Levinson, The Canons of Constitutional Law, 111 Harv. L. Rev. 963 (1998). Balkin and Levinson address various features, beyond opinion importance, which go into the text of casebooks, including the significance of the particular audience. Id. at 976. They observe that some cases may be included precisely because they are “wrongly decided or, even if the correct result is reached, offer styles of reasoning that the authors wish to question or criticize.” Id. at 982. The choices may also be influenced by the ideology of the authors. Id. at 998.

19 See Dennis Haines, Rolling Back the Top on Chief Justice Burger’s Opinion Assignment Desk, 38 U. Pitt. L. Rev. 631 (1977).

20 See Saul Brenner & Harold J. Spaeth, Stare Indecisis: The Alteration of Precedent on the Supreme Court, 1946–1992, at 25 (1995).

21Harold J. Spaeth, Distributive Justice: Majority Opinion Assignments in the Burger Court, 67 Judicature 299 (1984).

22 Brenner & Spaeth, supra note 20, at 25.

23Epstein & Segal, supra note 3, at 69; see Paul M. Collins, Jr., Amici Curiae and Dissensus on the U.S. Supreme Court, 5 J. Empirical Legal Stud. 143, 151–53 (2008).

24 Stefanie A. Lindquist & Frank B. Cross, Measuring Judicial Activism 1 (2009).

25 See id.

26 Bernard Schwartz, A Book of Legal Lists: The Best and Worst in American Law 48 (1997). His choices in order are as follows:

cross-spriggs-fn-26-tbl

27Schwartz concedes that the list is “as personal as a sports writer’s choices for an all-star team.” Id. at 66.

28 Id. at 67.

29 See generally Roy B. Flemming et al., One Voice Among Many: The Supreme Court’s Influence on Attentiveness to Issues in the United States, 1947–92, 41 Am. J. Pol. Sci. 1224, 1224 (1997) (noting that “politically significant” decisions affect “systemic attention to the underlying issues”).

30 See Glenn A. Phelps & John B. Gates, The Myth of Jurisprudence: Interpretive Theory in the Constitutional Opinions of Justices Rehnquist and Brennan, 31 Santa Clara L. Rev. 567 (1991) (providing data showing that prior Court precedents are more commonly invoked than other authorities in the Court’s opinions).

31Jonathan R. Macey, The Internal and External Costs and Benefits of Stare Decisis, 65 Chi.-Kent L. Rev. 93, 106 (1989).

32 Michael J. Gerhardt, The Power of Precedent 109 (2008).

33Kosma, supra note 8, at 338.

34 Richard A. Posner, Cardozo: A Study of Reputation 80–90 (1990).

35Lawrence M. Friedman et al., State Supreme Courts: A Century of Style and Citation, 33 Stan. L. Rev. 773, 794 (1981).

36 See Richard A. Posner, Judges’ Writing Styles (and Do They Matter?), 62 U. Chi. L. Rev. 1421, 1424–25 (1995) (stating that “[e]ven a brilliant analysis of yesterday’s legal problems is unlikely to hold much current interest,” while the greater opinion “can be pulled out and made exemplary of law’s abiding concerns”).

37These data were derived from Fowler et al., supra note 15.

38A possible exception to this position would be an opinion that so clearly settled the law that disputes did not arise precisely because of the power and clarity of the opinion. This possibility is discussed below at note 63 and accompanying text.

39 See generally Oona A. Hathaway, Path Dependence in the Law: The Course and Pattern of Legal Change in a Common Law System, 86 Iowa L. Rev. 601 (2001) (describing path-dependence theory and applying it to common law decision making).

40 See id. at 606–09, 627–35.

41This theory is discussed in Stefanie A. Lindquist & Frank B. Cross, Empirically Testing Dworkin’s Chain Novel Theory: Studying the Path of Precedent, 80 N.Y.U. L. Rev. 1156 (2005).

42 See id.

43 Id. at 1170.

44 Gerhardt, supra note 32, at 192.

45 See Court in Transition: When a Precedent Is Reaffirmed, That Strengthens the Precedent, N.Y. Times, Jan. 11, 2006, at A26 (quoting Justice Alito’s confirmation hearing testimony).

46Walter V. Schaefer, Precedent and Policy, 34 U. Chi. L. Rev. 3, 11 (1966).

47Planned Parenthood of Se. Pa. v. Casey, 505 U.S. 833, 866 (1992).

48Tom R. Tyler & Gregory Mitchell, Legitimacy and the Empowerment of Discretionary Legal Authority: The United States Supreme Court and Abortion Rights, 43 Duke L.J. 703, 786 (1994).

49Thomas M. Merrill, A Modest Proposal for a Political Court, 17 Harv. J.L. & Pub. Pol’y 137, 139 (1994).

50 See James R. Zink et al., Courting the Public: The Influence of Decision Attributes on Individuals’ Views of Court Opinions, 71 J. Pol. 909 (2009).

51 Id.

52 Lee Epstein & Jack Knight, The Choices Justices Make 45 (1998).

53 Id.

54Justice Stevens has declared that following precedent “obviously enhances the institutional strength of the judiciary.” John Paul Stevens, The Life Span of a Judge-Made Rule, 58 N.Y.U. L. Rev. 1, 2 (1983).

55 Hansford & Spriggs, supra note 5, at 23 (contending that “for legitimacy reasons, the justices are more likely to rely on those precedents possessing greater legal weight” as reflected in their repeated citation).

56 Gerhardt, supra note 32, at 368.

57Landes & Posner, supra note 8, at 273. This has been modeled via game theory. See Erin O’Hara, Social Constraint or Implicit Collusion?: Toward a Game Theoretic Analysis of Stare Decisis, 24 Seton Hall L. Rev. 736, 745–49 (1993) (explaining how judges agree to follow each other’s precedents to avoid nonproductive competition); Eric Rasmusen, Judicial Legitimacy as a Repeated Game, 10 J.L. Econ. & Org. 63, 67 (1994) (arguing that stare decisis enhances judges’ power vis-à-vis future judges).

58 See, e.g., Lawrence Baum, The Puzzle of Judicial Behavior 61 (1997) (“[I]t pleases judges to carry out what they conceive as the judge’s role.”); Tracey E. George, From Judge to Justice: Social Background Theory and the Supreme Court, 86 N.C. L. Rev. 1333, 1355–57 (2008) (discussing this role theory and its influence on judges). Baum argues that ideological preferences are constrained “because decision makers want to reach results that they can accept as correct.” Baum, supra, at 65.

59 See generally Hansford & Spriggs, supra note 5 (analyzing the Supreme Court’s use of precedent from 1946 to 2000 and showing that the interplay of the Justices’ ideological orientations and the current legal authority of a precedent combined to influence how the Court legally interpreted the precedent).

60 Id. at 22.

61 See Frank B. Cross, Chief Justice Roberts and Precedent: A Preliminary Study, 86 N.C. L. Rev. 1251, 1274 (2008) (surveying cases decided by Chief Justice Roberts and finding that the “opinions cited, on average, less than half the cases found in both the petitioners’ and respondents’ briefs”); James F. Spriggs II & Thomas G. Hansford, The U.S. Supreme Court’s Incorporation and Interpretation of Precedent, 36 Law & Soc’y Rev. 139 (2002) (modeling the conditions under which the Court positively or negatively interprets the precedents cited in litigant and amici briefs).

62 Benjamin N. Cardozo, The Nature of the Judicial Process 164–65 (1921).

63 See Kosma, supra note 8, at 339 (noting the criticism that “a precedent may set such a clear legal standard that subsequent cases settle rather than go to trial and appeal, resulting in few citations to the case in later opinions”).

64 See, e.g., Landes et al., supra note 8, at 274 (noting that citation counts could understate the significance of a precedent “that is so effective in clearing up an unsettled area of law that future disputes settle rather than go to trial”). The authors suggested that such cases were “rare,” however. Id.

65Much has been written on the different implications of rules and standards. See, e.g., Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 Duke L.J. 557 (1992) (modeling this effect on primary behavior); Kathleen Sullivan, The Justices of Rules and Standards, 106 Harv. L. Rev. 22 (1992) (discussing the Supreme Court’s use of rules and standards).

66Charles M. Cameron, New Avenues for Modeling Judicial Politics 45 (Oct. 4, 1993) (unpublished manuscript), available at http://www.princeton.edu/~ccameron/NewAvenues.pdf.

67384 U.S. 436 (1966).

68 See, e.g., Berkemer v. McCarty, 468 U.S. 420 (1984); Rhode Island v. Innis, 446 U.S. 291 (1980); Brewer v. Williams, 430 U.S. 387 (1977); Orozco v. Texas, 394 U.S. 324 (1969).

69 See United States v. Leon, 468 U.S. 897 (1984).

70Superprecedent is a term that has been recently used to describe a Supreme Court opinion that is so entrenched in our law and politics that it is beyond challenge. See Michael J. Gerhardt, Super Precedent, 90 Minn. L. Rev. 1204, 1207 (2006). The term was coined to refer to an opinion that “would be so effective in defining the requirements of the law that it prevents legal disputes from arising in the first place, or, if they do arise, induces them to be settled without litigation.” Michael Sinclair, Precedent, Super-Precedent, 14 Geo. Mason L. Rev. 363, 364 (2007).

71Sinclair, supra note 70, at 364.

725 U.S. (1 Cranch) 137 (1803).

73 Marbury is a leading example of a superprecedent. See Gerhardt, supra note 70, at 1207–08; Sinclair, supra note 70, at 364. As of this writing, Marbury has over 17,000 total citations in the Westlaw database. It has the fourth most citations of any case in Supreme Court history, garnering 209 citations by the U.S. Supreme Court through 2005. See infra Table 1.

74Landes & Posner, supra note 8, at 251.

75Kosma, supra note 8, at 339.

76Research on the attitudinal model of decision making has shown systematic patterns of voting by Supreme Court Justices that appear to parallel their political ideological preferences. See Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model (1993) [hereinafter Segal & Spaeth, Attitudinal Model]; Segal & Spaeth, Attitudinal Model Revisited, supra note 5. “Scores” of additional studies have confirmed this effect. Emerson H. Tiller & Frank B. Cross, What Is Legal Doctrine?, 100 Nw. U. L. Rev. 517, 523 (2006).

77Many have expressed skepticism about the influence of stare decisis on the Justices. Henry Monaghan has stated that precedent is merely a “mask hiding other considerations.” Henry Paul Monaghan, Stare Decisis and Constitutional Adjudication, 88 Colum. L. Rev. 723, 743 (1988). Judge Wald suggested that judges simply distinguish away unappealing precedents and “follow those precedents which they like best.” Patricia M. Wald, Changing Course: The Use of Precedent in the District of Columbia Circuit, 34 Clev. St. L. Rev. 477, 481 (1986). Segal and Spaeth found that Justices who dissented from an original opinion did not respect its power, but continued to dissent from future opinions relying on the original opinion. Harold J. Spaeth & Jeffrey A. Segal, Majority Rule or Minority Will (1999); see also Hansford & Spriggs, supra note 5, at 3 (showing that the Justices are more likely to negatively interpret (e.g., overrule) precedent if they are ideologically opposed to it).

78 See Vanessa A. Baird, The Effect of Politically Salient Decisions on the U.S. Supreme Court’s Agenda, 66 J. Pol. 755 (2004).

79 Id.

80 See Kosma, supra note 8, at 339–40; Landes & Posner, supra note 8, at 274–76.

8183 U.S. (16 Wall.) 36 (1873).

82 See, e.g., Charles L. Black, Jr., A New Birth of Freedom: Human Rights, Named & Unnamed 55 (1999) (stating that the decision was “probably the worst holding, in its effect on human rights, ever uttered by the Supreme Court”); Lawrence H. Tribe, American Constitutional Law 1303–11 (3d ed. 2000) (discussing the case and criticism); Steven G. Calabresi & Sarah E. Agudo, Individual Rights Under State Constitutions When the Fourteenth Amendment Was Ratified in 1868: What Rights Are Deeply Rooted in American History and Tradition?, 87 Tex. L. Rev. 7, 73 (2008) (“The [opinion] construed the Fourteenth Amendment in an absurdly narrow way, which has been roundly excoriated by courts and scholars.”).

83 See, e.g., Steven J. Eagle, Property Tests, Due Process Tests and Regulatory Takings Jurisprudence, 2007 BYU L. Rev. 899, 904 (referring to Slaughter-House as “one of the most important decisions of the nineteenth century”).

84For a discussion of the advantages of legal relevance scores for measuring case importance, see Fowler et al., supra note 15; infra Part II.B.

85 See Kosma, supra note 8, at 340 (observing that if the number of such cases was small and not unusually distributed, any distortion of results would be “of minor significance”).

86 See id. at 339 (“Such an opinion, with marginal or questionable influence, may be overrepresented when counting citations.”).

87 See id. at 339–40. Kosma notes that if the pattern of citations continues, the original case may have been “more useful than one might have guessed initially.” Id. at 340.

88 Gerhardt, supra note 32, at 45.

89 Id. at 153.

90 Id. at 45.

91 See infra Table 4.

92Frank B. Cross et al., Warren Court Precedents in the Rehnquist Court, 24 Const. Comment. 3, 4–5 (2007).

93 Id.; see also Frank B. Cross & Stefanie A. Lindquist, Judging the Judges, 58 Duke L.J. 1383, 1391 (2009) (noting the need to consider negative citations).

94Landes et al., supra note 8, at 273. Judge Posner has noted that even negative citations are “motivated by the authority of the previous case.” Posner, supra note 4, at 385.

95In a typical decade, the Court overrules less than 0.002% of its previous opinions. Jeffrey A. Segal et al., The Supreme Court in the American Legal System 316 (2005). Of the 6,363 Supreme Court cases decided between 1946 and 1999, the Court had overruled only 107 of them by 2001. Hansford & Spriggs, supra note 5, at 81.

96 See Choi & Gulati, supra note 8, at 56–57.

97 See sources cited supra note 8.

98A seminal study in this regard is S. Sidney Ulmer, The Use of Power in the Supreme Court: The Opinion Assignments of Earl Warren, 1953-1960, 19 J. Pub. L. 49, 55 (1970) (using citations in the five years following a decision as a measure of importance).

99Kosma, supra note 8, at 333–34.

100Landes & Posner, supra note 8, at 293.

101Fowler et al., supra note 15.

102 Id. at 342.

103E.A. Leicht et al., Large-Scale Structure of Time Evolving Citation Networks, 59 Eur. Physical J. B 75, 83 (2007).

104 Id.

105Merryman, supra note 8, at 615.

106 See Daniel A. Farber, Supreme Court Selection and Measures of Past Judicial Performance, 32 Fla. St. U. L. Rev. 1175, 1178 (2005) (suggesting that more citations to an opinion may be a feature of “just plain luck”).

107 See Posner, supra note 4, at 387 (noting that although quantitative studies of citations inevitably contain much random “noise,” this does not “disable useful statistical analysis”).

108Landes & Posner, supra note 8, at 252.

109Schaefer, supra note 46, at 7.

110Robert H. Jackson, Decisional Law and Stare Decisis, 30 A.B.A. J. 334, 335 (1944).

111 See Fowler et al., supra note 15, for how this list of Supreme Court opinions was created.

112These data are available from Ryan C. Black & James F. Spriggs II, An Empirical Analysis of the Length of U.S. Supreme Court Opinions, 45 Hous. L. Rev. 621 (2008), and Fowler et al., supra note 15.

113Fowler et al., supra note 15, at 328 n.12. By “life,” we mean the period of time from when it was decided through 2005.

114Fowler et al., supra note 15.

115 See Seth J. Chandler, The Network Structure of Supreme Court Jurisprudence, 10 Mathematica J. 501 (2007).

116 See sources cited supra note 8.

117 See Schwartz, supra note 26.

118116 U.S. 616 (1886).

119 The Oxford Guide, supra note 10, at 30.

120 Lucas A. Powe, Jr., The Supreme Court and the American Elite, 1789-2008 (2009).

121Schmerber v. California, 384 U.S. 757, 776 (1966) (Black, J., dissenting). Justice Brandeis said that Boyd was a “case that will be remembered as long as civil liberty lives in the United States.” Olmstead v. United States, 277 U.S. 438, 474 (1928) (Brandeis, J., dissenting).

122384 U.S. 436 (1966).

123310 U.S. 88 (1940).

124Thurgood Marshall, Mr. Justice Murphy and Civil Rights, 48 Mich. L. Rev. 745, 748 (1950).

125 David E. Klein, Making Law in the United States Courts of Appeals 51 (2002).

126 Frank B. Cross, Decision Making in the U.S. Courts of Appeals 2 (2007).

127While district courts offer more opinions than do the circuit courts, their rulings are typically “heavily fact based and jurisdictionally limited in effect, and they do not set the significant legal precedents that make up the law.” Id.

128 Id.

129466 U.S. 668 (1984).

130The third case on the list, Anders v. California, 386 U.S. 738 (1967), deals with habeas corpus standards, as does the fourth case, Jackson v. Virginia, 443 U.S. 307 (1979).

131 See Biskupic & Witt, supra note 11; The Oxford Guide, supra note 10.

132477 U.S. 242 (1986).

133411 U.S. 792 (1973).

134 See Thomas A. Smith, The Web of Law, 44 San Diego L. Rev. 309, 347 (2007) (noting that the summary judgment opinions are “cited so often because the federal courts . . . handle many motions for summary judgment”). Smith argues that it does “not follow” that these opinions would necessarily be the “most important” or “most authoritative” opinions. However, he also notes that it would be hard to deny that they provide “law that is at the core of what federal courts do.” Id.

135This claim may be a bit too strong because a truly unworkable decision on these topics would prove unhelpful to lower courts and presumably result in another Court decision creating a better precedent. However, even a moderately good decision, workable at the lower court level, would not necessarily need overruling or revision.

136530 U.S. 466 (2000). See Stephanos Bibas, Apprendi in the States: The Virtues of Federalism as a Structural Limit on Errors, 94 J. Crim. L. & Criminology 1, 11 (2003) (noting the decision’s “disruptive” effect in upsetting settled criminal procedures by potentially requiring the costly “reopening [of] . . . hundreds of thousands of cases”).

137District courts are bound to apply the law of the circuit within which they operate. Hence, their citations may be influenced by the filter of that circuit court.

138Anderson v. Liberty Lobby, Inc., 477 U.S. 242 (1986); Matsushita Electric Indus. Co. v. Zenith Radio Corp., 475 U.S. 574 (1986).

139Celotex Corp. v. Catrett, 477 U.S. 317 (1986).

140411 U.S. 792 (1973).

141Texas Dep’t of Cmty. Affairs v. Burdine, 450 U.S. 248 (1981).

142 Monell v. Dep’t of Soc. Servs., 436 U.S. 658 (1978); Adickes v. S. H. Kress & Co., 398 U.S. 144 (1970).

143Scheuer v. Rhodes, 416 U.S. 232 (1974).

144Network research has been used to study “food webs, electrical power grids, cellular and metabolic networks, the World-Wide Web, the Internet backbone, the neural network of the nematode worm Caenorhabditis elegans, telephone call graphs, . . . and the quintessential ‘old-boy’ network, the overlapping boards of directors of the largest companies in the United States.” Steven H. Strogatz, Exploring Complex Networks, 410 Nature 268, 268 (2001).

145 See Cross et al., supra note 5, at 1237.

146David G. Post & Michael B. Eisen, How Long Is the Coastline of the Law? Thoughts on the Fractal Nature of Legal Systems, 29 J. Legal Stud. 545, 545 (2000).

147Michael J. Gerhardt, The Irrepressibility of Precedent, 86 N.C. L. Rev. 1279, 1289 (2008). Gerhardt suggests that the “clarity of [a precedent’s] significance and meaning . . . depends on the consistency and uniformity with which the Court and other public authorities have cited it.” Id.

148 See generally Jon M. Kleinberg, Authoritative Sources in a Hyperlinked Environment, 46 J. Ass’n Computing Machinery 604 (1999) (proposing and testing a mathematical formula for determining what constitutes an authoritative internet site). See Fowler et al., supra note 15, for a description of the use of this measure for the legal network.

149Fowler et al., supra note 15, at 324–25.

150410 U.S. 113 (1973).

151381 U.S. 479 (1965).

152Fowler et al., supra note 15, at 335.

153 Id.

154492 U.S. 490 (1989).

155476 U.S. 747 (1986); Fowler & Jeon, supra note 12, at 21.

156310 U.S. 296 (1940).

157308 U.S. 147 (1939).

158Fowler et al., supra note 15, at 333.

159Landes & Posner, supra note 8.

160 Id. at 262.

161 See id. at 264.

162 Id. at 263.

163 Id. at 280.

164 Id. at 282–83.

165 Id. at 263. A study of state supreme courts found similar evidence of the depreciation of precedents. Friedman et al., supra note 35, at 807.

166Landes & Posner, supra note 8, at 263 n.19.

167 See Fowler et al., supra note 15.

168372 U.S. 293 (1963).

169Landes & Posner, supra note 8, at 263.

170Unsurprisingly, the choice of citations is closely related to the precedents’ legal relevance to the cases taken by the Court. Spriggs & Hansford, supra note 61, at 143.

171 See Biskupic & Witt, supra note 11, at 322 (noting that for the first 150 years of its history, “the Supreme Court exerted its greatest influence on the states of the Union through its decisions on matters of economic interest”).

172 Hansford & Spriggs, supra note 5, at 62.

173116 U.S. 616 (1886).

174 See Schwartz, supra note 26.

175 Boyd, 116 U.S. at 616.

176Adam M. Samaha, Originalism’s Expiration Date, 30 Cardozo L. Rev. 1295, 1324–25 (2008).

17740 U.S. (15 Pet.) 518 (1841).

178333 U.S. 507 (1948).

179408 U.S. 92 (1972).

180321 U.S. 158 (1944).

181366 U.S. 420 (1961).

182 See 310 U.S. 296 (1940).

183 See 308 U.S. 147 (1939).

184468 U.S. 420 (1984).

185Nancy Staudt et al., On the Role of Ideological Homogeneity in Generating Consequential Constitutional Decisions, 10 U. Pa. J. Const. L. 361, 361–62 (2007).

186 Id. at 370 n.39.

187Schaefer, supra note 46, at 10.

188For documentation, see Harold J. Spaeth et al., The Supreme Court Database (Aug. 26, 2010), http://scdb.wustl.edu/data.php. This source is so accepted in political science that it “would certainly be unusual for a refereed journal to publish a manuscript whose data derived from an alternate source,” and the same is true of law reviews. Lee Epstein et al., The Political (Science) Context of Judging, 47 St. Louis U. L.J. 783, 812 (2003). This source is the “greatest single resource of data on the Court.” Michael Heise, The Past, Present, and Future of Empirical Legal Scholarship: Judicial Decision Making and the New Empiricism, 2002 U. Ill. L. Rev. 819, 848.

189The case types in the United States Supreme Court Judicial Database are categorized as the following: attorneys, civil rights, criminal procedure, due process, economic activity, federalism, First Amendment, judicial power, privacy, federal taxation, interstate relations, unions, and miscellaneous. Spaeth et al., supra note 188. The nature of the categories may not be entirely transparent, and more detailed descriptions of the categorization can be found in the codebook for the database.

190 Lee Epstein et al., The Supreme Court Compendium 80 (3d ed. 2003).

191 Id. at 85.

192 Id.

193 See Spaeth et al., supra note 188.

194 See Stevens, supra note 54, at 2.

195Nat’l Bank v. Whitney, 103 U.S. 99, 102 (1880).

196 Henry M. Hart, Jr. & Albert M. Sacks, The Legal Process: Basic Problems in the Making and Application of Law 587 (1958); see also Lewis F. Powell, Jr., Stare Decisis and Judicial Restraint, 47 Wash. & Lee L. Rev. 281, 286 (1990) (declaring that reliance “is especially important in cases involving property rights and commercial transactions”).

197Payne v. Tennessee, 501 U.S. 808, 828 (1991).

198 See Polly J. Price, A Constitutional Significance for Precedent: Originalism, Stare Decisis, and Property Rights, 5 Ave Maria L. Rev. 113 (2007) (reviewing the history of giving strong stare decisis to decisions involving property rights); Lee J. Strang & Bryce G. Poole, The Historical (In)Accuracy of the Brandeis Dichotomy: An Assessment of the Two-Tiered Standard of Stare Decisis for Supreme Court Precedents, 86 N.C. L. Rev. 969 (2008).

199 Payne, 501 U.S. at 852–53 (Marshall, J., dissenting). For example, the reliance interest in abortion decisions has been emphasized. See Michael J. Gerhardt, The Pressure of Precedent: A Critique of the Conservative Approaches to Stare Decisis in Abortion Cases, 10 Const. Comment. 67 (1993).

200This position is generally traced to Justice Brandeis’s dissenting opinion in Burnet v. Coronado Oil & Gas Co., 285 U.S. 393, 406–10 (1932) (Brandeis, J., dissenting) (arguing for separate standards for constitutional and statutory precedents).

201Powell, supra note 196, at 287; see also Payne, 501 U.S. at 828 (noting that precedent has less power in constitutional cases because their “correction through legislative action is practically impossible”).

202 See Lawrence C. Marshall, “Let Congress Do It”: The Case for an Absolute Rule of Stare Decisis, 88 Mich. L. Rev. 177 (1989).

203Amy Coney Barrett, Statutory Stare Decisis in the Courts of Appeals, 73 Geo. Wash. L. Rev. 317, 317 (2005); see also Thomas R. Lee, Stare Decisis in Historical Perspective: From the Founding Era to the Rehnquist Court, 52 Vand. L. Rev. 647, 731 (1999) (discussing the historical development of this perspective).

204James F. Spriggs, II & Thomas G. Hansford, Explaining the Overruling of U.S. Supreme Court Precedent, 63 J. Pol. 1091 (2001).

205 Gerhardt, supra note 32, at 84.

206Harris v. United States, 536 U.S. 545, 557 (2002); see also Agostini v. Felton, 521 U.S. 203, 235–36 (1997) (listing cases respecting this differentiation). Constitutional precedents are subject to only a “weak presumption of correctness.” Rafael Gely, Of Sinking and Escalating: A (Somewhat) New Look at Stare Decisis, 60 U. Pitt. L. Rev. 89, 110 (1998). The distinction reportedly has persisted since the first half of the nineteenth century. Earl M. Maltz, Some Thoughts on the Death of Stare Decisis in Constitutional Law, 1980 Wis. L. Rev. 467, 467.

207The classic investigation of this practice is found in William N. Eskridge, Jr., Overriding Supreme Court Statutory Interpretation Decisions, 101 Yale L.J. 331 (1991). Others have also empirically investigated the practice. See, e.g., Beth Henschen, Statutory Interpretations of the Supreme Court: Congressional Response, 11 Am. Pol. Q. 441 (1983) (examining the practice in the areas of antitrust and labor law); Joseph Ignagni & James Meernik, Explaining Congressional Attempts to Reverse Supreme Court Decisions, 47 Pol. Res. Q. 353 (1994) (considering determinants of when Congress overrides a judicial opinion).

208 See Robert D. Cooter & Tom Ginsburg, Comparative Judicial Discretion: An Empirical Test of Economic Models, 16 Int’l Rev. L. & Econ. 295, 296 (1996) (noting that “prudent judges” would not push their ideology to the point where Congress would need to overturn their interpretation of the law).

209This difficulty may be overstated. There are various ways in which Congress may rewrite statutes so as to accomplish its goal independent of a constitutional decision of the Supreme Court. See Amy L. Padden, Overruling Decisions in the Supreme Court: The Role of a Decision’s Vote, Age, and Subject Matter in the Application of Stare Decisis after Payne v. Tennessee, 82 Geo. L.J. 1689, 1717–18 (1994) (describing some methods). In addition, Congress has a variety of ways of pressuring the Court over constitutional decisions, which may be simpler and more effective than the conventional rewriting of a statute. See Frank B. Cross & Blake J. Nelson, Strategic Institutional Effects on Supreme Court Decisionmaking, 95 Nw. U. L. Rev. 1437, 1459–71 (2001) (reviewing the research on these approaches).

210Strang & Poole, supra note 198, at 974.

211 Id. at 1015–25.

212 See Payne v. Tennessee, 501 U.S. 808, 828 (1991).

213 See Richard A. Posner, How Judges Think 82–87 (2008).

214However, this effect is not clear because legislatures may override both common law and statutory decisions by statute.

215Gely, supra note 206, at 109.

216 See Carolyn Shapiro, Coding Complexity: Bringing Law to the Empirical Analysis of the Supreme Court, 60 Hastings L.J. 477 (2009) (reporting errors in coding for legal issues).

217 See Forrest Maltzman et al., Crafting Law on the Supreme Court: The Collegial Game (2000); Timothy R. Johnson et al., The Influence of Oral Arguments on the U.S. Supreme Court, 100 Am. Pol. Sci. Rev. 99 (2006).

218The reevaluation of the coding found that the database undercounted legal issues overall, especially in issue areas such as judicial power and government structure and operations. See Shapiro, supra note 216, at 519–20. These errors would have the effect of producing statistical noise that interfered with finding a true effect.

219Landes & Posner, supra note 8, at 281.

220 Hansford & Spriggs, supra note 5, at 116–17 figs.7.1 & 7.2 (graphically displaying this effect). Absolute citations remain high for about twenty years but then precipitously decline in frequency. Id.; see also Spriggs & Hansford, supra note 61; Ryan C. Black & James F. Spriggs II, The Depreciation of Precedent on the U.S. Supreme Court 24–25 (June 12, 2009) (unpublished manuscript), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1421413.

221 Hansford & Spriggs, supra note 5, at 24.

222 Id. at 64.

223 See Segal & Spaeth, Attitudinal Model, supra note 76; Segal & Spaeth, Attitudinal Model Revisited, supra note 5. Considerable additional evidence supports the basic findings of these books. See, e.g., Daniel R. Pinello, Linking Party to Judicial Ideology in American Courts: A Meta-Analysis, 20 Just. Sys. J. 219 (1999) (presenting a meta-analysis of the effect of ideology at various court levels and finding it most profound in the Supreme Court).

224 Mark Tushnet, Red, White, and Blue: A Critical Analysis of Constitutional Law 191–92 (1988).

225Christopher P. Banks, Reversals of Precedent and Judicial Policy-Making: How Judicial Conceptions of Stare Decisis in the U.S. Supreme Court Influence Social Change, 32 Akron L. Rev. 233, 235 (1999); see also Padden, supra note 209, at 1689 (suggesting that stare decisis “is often manipulated by liberals and conservatives alike when precedents are viewed as unappealing”).

226This research was originally published as Segal & Spaeth, supra note 11, and subsequently, in more detail, as Spaeth & Segal, supra note 77.

227 Spaeth & Segal, supra note 77.

228For a summary of these analyses, see Charles A. Johnson, Follow-Up Citations in the U.S. Supreme Court, 39 W. Pol. Q. 538 (1986).

229One study of the early opinions of Chief Justice Roberts found that he viewed prior opinions not as “straightjackets that dictate[d] his decisions” but as “boundaries that shape[d] the nature of his opinions.” Cross, supra note 61, at 1276. Additional empirical foundation comes from research suggesting that a benchmark opinion significantly shaped the subsequent opinions issued by the Court. Herbert M. Kritzer & Mark J. Richards, Jurisprudential Regimes and Supreme Court Decisionmaking: The Lemon Regime and Establishment Clause Cases, 37 Law & Soc’y Rev. 827 (2003); Mark J. Richards & Herbert M. Kritzer, Jurisprudential Regimes in Supreme Court Decision Making, 96 Am. Pol. Sci. Rev. 305 (2002); Kevin M. Scott, Reconsidering the Impact of Jurisprudential Regimes, 87 Soc. Sci. Q. 380 (2006).

230 See, e.g., Lawrence Baum, Judges and Their Audiences 20 (2006) (“[T]he existing evidence does not establish that Justices are motivated solely (or even overwhelmingly) by policy goals”); Timothy Johnson et al., Oral Advocacy Before the United States Supreme Court: Does It Affect the Justices’ Decisions?, 85 Wash. U. L. Rev. 457, 525 (2007) (reporting that “the attitudinal model is inaccurate in its main theoretical claim” that ideology is the only factor driving the Court’s decisions).

231Marshall, supra note 202, at 178.

232 Mitchell S. G. Klein, Law, Courts, and Policy 112 (1984). An example of this effect might be found in Patterson v. McLean Credit Union, 491 U.S. 164 (1989), where the Court “shocked the legal community” by unanimously supporting the reach of civil rights legislation based on precedent. Marshall, supra note 202, at 178.

233Jack Knight & Lee Epstein, The Norm of Stare Decisis, 40 Am. J. Pol. Sci. 1018, 1029 (1996); Spriggs & Hansford, supra note 61 (showing that the Court is more likely to positively interpret precedents that have a higher level of legal vitality).

234Donald R. Songer & Dona Roy, A Critical Test of the Attitudinal Model 17–18 (Apr. 10, 2005) (unpublished manuscript), available at http://www.allacademic.com/meta/p86158_index.html. The study also found no major difference in the lower court ideologies in cases that saw unanimous liberal and unanimous conservative decisions. Id.

235 See Knight & Epstein, supra note 233, at 1027.

236 Lawrence Baum, The Supreme Court 150 (5th ed. 1995).

237Kritzer & Richards, supra note 229, at 315.

238 Hansford & Spriggs, supra note 5, at 10.

239One study found a very limited ideological effect in the Justices’ choice of citations, but it considered only a small number of opinions. Johnson, supra note 228.

240Staudt et al., supra note 185, at 373 (referring to the view that “liberals are more likely to render cutting-edge decisions”).

241 Richard A. Posner, The Federal Courts: Crisis and Reform 217 (1985) [hereinafter Posner, Crisis].

242 See, e.g., John C. Eastman, Stare Decisis: Conservatism’s One–Way Ratchet Problem, in Courts and the Culture Wars 127 (Bradley Watson ed., 2002); Lino A. Graglia, The Myth of a Conservative Supreme Court: The October 2000 Term, 26 Harv. J.L. & Pub. Pol’y 281, 284 (2003) (arguing that this ratchet effect was occurring, as a conservative Court was adhering to liberal precedents); Frank B. Cross, Gay Politics and Precedents, 103 Mich. L. Rev. 1186, 1203 (2005) (reviewing Daniel R. Pinello, Gay Rights and American Law (2003)) (addressing hypothesis that “activist liberal precedents breed activism, while contrary conservative precedents have only a weak countervailing effect, because conservative judges feel bound by stare decisis”).

243 See Pinello, supra note 243 (reporting evidence at different court levels showing that pro-gay rights decisions had a much more significant precedential effect than contrary decisions).

244 See, e.g., Steven G. Calabresi, Text, Precedent, and the Constitution: Some Originalist and Normative Arguments for Overruling Planned Parenthood of Southeastern Pennsylvania v. Casey, 22 Const. Comment. 311 (2005) (exploring the tension between various proposals for understanding the Constitution). Thomas Merrill has thus observed that for originalists, the importance of “judicial precedent—the most important tool of the constitutional lawyer—drops from sight.” Thomas W. Merrill, The Common Law Powers of Federal Courts, 52 U. Chi. L. Rev. 1, 69 (1985).

245 See Gerhardt, supra note 32, at 53.

246347 U.S. 483 (1954).

247388 U.S. 1 (1967).

248 See Lindquist & Cross, supra note 24, at 146–47 (describing the persistence of decisions such as these).

249Dickerson v. United States, 530 U.S. 428, 428 (2000).

250 Frederick P. Lewis, The Context of Judicial Activism: The Endurance of the Warren Court Legacy in a Conservative Age 38 (1999).

251Some studies in law reviews making use of this measure include Andrew D. Martin et al., The Median Justice on the United States Supreme Court, 83 N.C. L. Rev. 1275 (2005); Theodore W. Ruger, Justice Harry Blackmun and the Phenomenon of Judicial Preference Change, 70 Mo. L. Rev. 1209 (2005); and Paul J. Wahlbeck, Strategy and Constraints on Supreme Court Opinion Assignment, 154 U. Pa. L. Rev. 1729 (2006).

252 See Staudt et al., supra note 185, at 363.

253 Id. at 379.

254 Id. at 381–82.

255 See infra Part III.D.

256Cross et al., supra note 92, at 15.

257 Id.

258 Hansford & Spriggs, supra note 5, at 64. The ideological position of the precedent was measured based on the ideological voting pattern of the median member of the majority voting coalition behind the opinion. Id. at 59.

259 Id. at 65.

260Black & Spriggs, supra note 220, at 23 (discussing a case’s ideological “depreciation”).

261 Walter Murphy, Elements of Judicial Strategy 98 (1964); see also Jeffrey R. Lax & Charles M. Cameron, Bargaining and Opinion Assignment on the US Supreme Court, 23 J.L. Econ. & Org. 276, 277 (2007) (“[T]he policy impact of a legal opinion depends partly on its persuasiveness, clarity, and craftsmanship—its legal quality . . . .”).

262Robert J. Hume, The Impact of Judicial Opinion Language on the Transmission of Federal Circuit Court Precedents, 43 Law & Soc’y Rev. 127 (2009).

263 Id. at 138–41.

264Mark Alan Thurmon, When the Court Divides: Reconsidering the Precedential Value of Supreme Court Plurality Decisions, 42 Duke L.J. 419, 449 (1992).

265 Hansford & Spriggs, supra note 5, at 41.

266 Richard A. Posner, The Federal Courts: Challenge and Reform 236 (1999) [hereinafter Posner, Challenge].

267 Payne v. Tennessee, 501 U.S. 808, 828–29 (1991).

268 See, e.g., Bob Woodward & Scott Armstrong, The Brethren: Inside the Supreme Court 345 (1979) (discussing Justice Brennan’s view that unanimity was required for the impact of United States v. Nixon); Gabriel J. Chin & Anjali Abraham, Beyond the Supermajority: Post-Adoption Ratification of the Equality Amendments, 50 Ariz. L. Rev. 25, 26 (2008) (“The impact and legitimacy of Brown v. Board of Education flowed in part from the Court’s unanimous rejection of segregation . . . .” (footnote omitted)); Frank B. Cross & Stefanie Lindquist, The Decisional Significance of the Chief Justice, 154 U. Pa. L. Rev. 1665, 1677–78 (2006) (“The ability to produce unanimous decisions in controversial cases, such as Brown and United States v. Nixon, [was] often regarded as vital . . . .” (footnotes omitted)); Micah Schwartzman, Judicial Sincerity, 94 Va. L. Rev. 987, 1023 (2008) (citing Brown as an example of a case “in which unanimity was thought to be essential to the Court’s legitimacy”).

269Lee Epstein et al., On the Capacity of the Roberts Court to Generate Consequential Precedent, 86 N.C. L. Rev. 1299, 1306 (2008); see also Bradley C. Canon & Charles A. Johnson, Judicial Policies: Implementation and Impact 168–69 (2d ed. 1999) (contending that unanimous decisions create “final, clear and persuasive policy”).

270 Walter F. Murphy, Elements of Judicial Strategy 66 (1964).

271Evan H. Caminker, Sincere and Strategic Voting Norms on Multimember Courts, 97 Mich. L. Rev. 2297, 2321 n.73 (1999).

272 Payne, 501 U.S. at 828–29. This position has been criticized as substantially undermining the power of stare decisis. See Padden, supra note 209, at 1713–14.

273Frederick Schauer, Statutory Construction and the Coordinating Function of Plain Meaning, 1990 Sup. Ct. Rev. 231, 247.

274Epstein et al., supra note 269, at 1306; see also Staudt et al., supra note 185, at 372 (“[A]s each additional Justice agrees to sign on, each presumably with his or her preferences, the decision becomes more and more diluted and thus produces less of an impact than could be achieved by five simpatico Justices.”).

275Staudt et al., supra note 185, at 380.

276These data are from Spaeth et al., supra note 188.

277Hume, supra note 262, at 132.

278John Henry Merryman, Toward a Theory of Citations: An Empirical Study of the Citation Practice of the California Supreme Court in 1950, 1960, and 1970, 50 S. Cal. L. Rev. 381, 421 (1977).

279 Id. at 422. Merryman’s study of the California Supreme Court supported this as the justices who cited the most overall authority also cited the most authoritative sources. Id.

280Robert A. Leflar, Honest Judicial Opinions, 74 Nw. U. L. Rev. 721, 724 (1979).

281 See Johnson et al., supra note 228.

282 See Cross, supra note 61, at 1276 (suggesting that Chief Justice Roberts views citations as a means for shifting the law, “creating a new path of stare decisis that will direct the course of future rulings”).

283Parents Involved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701, 746–48 (2007) (plurality opinion).

284These data come from Fowler et al., supra note 15, at 336–37.

285 See Friedman et al., supra note 35, at 775 (comparing similar opinions of different national courts and finding the French opinions shorter than those of American courts, which were in turn much shorter than opinions of English courts). There is also some systematic variation in opinion length of U.S. state supreme courts. Id. at 781–83.

286Peter H. Schuck & E. Donald Elliott, To the Chevron Station: An Empirical Study of Federal Administrative Law, 1990 Duke L.J. 984, 1003–04.

287Friedman et al., supra note 35, at 775–76.

288 Id. at 778.

289 See Scott Phillips & Ryken Grattet, Judicial Rhetoric, Meaning-Making, and the Institutionalization of Hate Crime Law, 34 Law & Soc’y Rev. 567, 587 (2000).

290Schuck & Elliott, supra note 286, at 1004.

291Cross, supra note 126, at 66.

292 William Domnarski, Federal Judges Revealed 193 (2009) (quoting Abner J. Mikva, Honorable Abner J. Mikva Oral History, in Historical Society of the District of Columbia Circuit Oral History Program 210 (1996)).

293 Id. (quoting Wilbur Frank Pell, Jr., The Oral History of Judge Wilbur F. Pell, Jr., in Circuit Library, U.S. Court of Appeals for the Seventh Circuit 64 (1998)).

294 Id. (quoting Abner J. Mikva, Honorable Abner J. Mikva Oral History, in Historical Society of the District of Columbia Circuit Oral History Program 210 (1996)).

295 See Posner, Challenge, supra note 266, at 147 (suggesting that longer opinions “reduce the opinion’s usefulness as a guide”).

296 See supra note 65.

297 Cross, supra note 126, at 65.

298 Id. at 225–26. As expected, reversals also produced more citations. Id. at 214–15.

299 Posner, Crisis, supra note 241, at 236 (reporting that longer opinions were more likely to be cited).

300Black & Spriggs, supra note 112.

301 Id. at 661.

302 Id. at 676–79.

303We obtained these data from Black & Spriggs. See id.

304We obtained these data from Black & Spriggs. See id.

305Ray Forrester, Supreme Court Opinions—Style and Substance: An Appeal for Reform, 47 Hastings L.J. 167, 186 (1995).

306550 U.S. 437, 459 (2007) (Alito, J., concurring).

307United States v. Carolene Prods. Co., 304 U.S. 144, 153 n.4 (1938).

308J.M. Balkin, The Footnote, 83 Nw. U. L. Rev. 275, 281–82 (1989).

309 See, e.g., Dirks v. SEC, 463 U.S. 646, 655 n.14 (1983) (noting that certain outsiders, including lawyers and auditors, could be considered insiders for insider-trading liability); Ernst & Ernst v. Hochfelder, 425 U.S. 185, 193 n.12 (1976) (providing the basis for a finding that recklessness could satisfy the scienter requirement of section 10(b) of the Securities Exchange Act of 1934).

310Adam Freedman, Footnotes: The ‘Insidious Plague’ That Helps Lawyers Explain ‘Fractured’ Reality, N.Y. L.J., Dec. 8, 2008, at S6.

311 Id. A defender of footnoting suggests that the “footnote will often be an appropriate place for the opinion-writer to set forth his or her doubts about the state of the law or the legal precept being announced.” Edward R. Becker, In Praise of Footnotes, 167 F.R.D. 283, 287 (1996).

312 See Posner, Challenge, supra note 266, at 236 (observing that at the circuit court level “the more footnotes an opinion has (holding the length of the opinion constant), the less likely it is to be cited, perhaps because footnotes make it more difficult for readers to extract a clear holding from an opinion”).

313Balkin, supra note 308, at 276.

314 Breeden v. Sprague Nat’l Bank (In re Bennett Funding Group, Inc.), 44 Collier Bankr. Cas. 2d (MB) 151, 157 n.7 (B.A.P. 2d Cir. 2000).

315 See, e.g., Arthur J. Goldberg, The Rise and Fall (We Hope) of Footnotes, 69 A.B.A. J. 255 (1983) (“Footnotes . . . cause more problems than they solve.”); Abner J. Mikva, Goodbye to Footnotes, 56 U. Colo. L. Rev. 647, 647 (1985) (declaring the footnote to be an “abomination”); In Justice Breyer’s Opinion, A Footnote Has No Place, N.Y. Times, July 28, 1995, at B18 (summarizing Justice Breyer’s antipathy for footnotes).

316Memorandum of Briefing Notes from Supreme Court Justice Lewis F. Powell to Law Clerks 20 (Sept. 10, 1984) (on file with the Washington and Lee University Law School).

317 Domnarski, supra note 292, at 195 (quoting Abner J. Mikva, Honorable Abner J. Mikva Oral History, in Historical Society of the District of Columbia Circuit Oral History Program 211 (1996)).

318 See, e.g., Becker, supra note 311, at 283 (suggesting that “well-conceived and well-crafted footnotes are valuable tools” for judges). Judge Becker contends that the use of footnotes enables a judge to write a more readable opinion for different audiences and provide a fuller understanding of the nuances of the case. Id. at 285–86.

319We obtained these data from Black & Spriggs, supra note 112.

320We obtained these data from Hansford & Spriggs, supra note 5, at 62.

321Paul M. Collins, Jr., Amici Curiae and Dissensus on the U.S. Supreme Court, 5 J. Empirical Legal Stud. 143 (2008).

322 See, e.g., Maltzman et al., supra note 217, at 45–46 (using this measure for political salience of a case); Virginia A. Hettinger et al., Comparing Attitudinal and Strategic Accounts of Dissenting Behavior on the U.S. Courts of Appeals, 48 Am. J. Pol. Sci. 123, 126 (2004) (same).

323 See Fowler et al., supra note 15 (using data collected from Shepard’s Citations Service).

324We determined whether a case struck down a federal statute as unconstitutional using Spaeth et al., supra note 188.

325Cass Sunstein thus notes that “it is best to measure judicial activism by seeing how often a court strikes down the actions of other parts of government, especially those of Congress.” Cass R. Sunstein, Radicals in Robes 42–43 (2005). Political scientists likewise conclude that the “most dramatic instances of a lack of judicial restraint—or, conversely, the manifestation of judicial activism—are decisions that declare acts of Congress . . . unconstitutional.” Segal & Spaeth, Attitudinal Model Revisited, supra note 5, at 413.

326We identified per curiam opinions using Spaeth et al., supra note 188. We identified per curiam opinions using Spaeth’s variable labeled “mow,” which lists the author of each majority opinion.

327 Hansford & Spriggs, supra note 5, at 64; see also Frank B. Cross et al., Citations in the U.S. Supreme Court: An Empirical Study of Their Use and Significance, 2010 U. Ill. L. Rev. 489, 539.

328 See Laura Krugman Ray, The Road to Bush v. Gore: The History of the Supreme Court’s Use of the Per Curiam Opinion, 79 Neb. L. Rev. 517, 521–30 (2000).

329For example, Brandenburg v. Ohio set a major First Amendment precedent via per curiam opinion. 395 U.S. 444 (1969) (overturning the Ohio Criminal Syndicalism statute, overruling Whitney v. California, 274 U.S. 357 (1927), and articulating a new test—the “imminent lawless action” test—for judging so-called seditious speech under the First Amendment).

330Whereas in Part II we listed the most significant cases in terms of citations to a case from all subsequent opinions (majority, concurring, and dissenting), the dependent variables in the following models are counts of citations to a case in majority opinions only.

331The “*” represents statistical significance at the .05 level (two-tailed test). The number of observations (i.e., cases) is 6,661.

332The legal relevance score is based first, on the quantity of citations, meaning the total number of citations a case has received. Each of those citations is then weighted by the “quality” of the citing case, which is assessed based on the number of citations in that precedent, where each of those citations is weighted by its quality. See Fowler et al., supra note 15, at 330–32.

333 See Posner, Challenge, supra note 266, at 146–47; cf. Stephen J. Choi et al., Judicial Evaluations and Information Forcing: Ranking State High Courts and Their Judges, 58 Duke L.J. 1313, 1322 (2009) (“[J]udicial decisionmaking is better when it is conservative and minimalist.”).

334This effect is not certain because we do not have a record of citations to the footnotes themselves. However, it is a highly plausible inference, absent a persuasive reason why more language in footnotes would somehow demean the power of the language in the body of the opinion.

335 See Hansford & Spriggs, supra note 5, at 84–91.

336Michael Tomz et al., CLARIFY: Software for Interpreting and Presenting Statistical Results, Harvard (June 1, 2001), http://gking.harvard.edu/clarify; see also Gary King et al., Making the Most of Statistical Analyses: Improving Interpretation and Presentation, 44 Am. J. Pol. Sci. 347 (2000) (discussing software for statistical analysis, including the authors’ own CLARIFY software).

337 See generally Posner, supra note 36 (discussing the effects of various judicial writing styles).

338Lax & Cameron, supra note 261, at 282.

339 E.g., Bush v. Gore, 531 U.S. 98, 109 (2000) (per curiam) (“Our consideration is limited to the present circumstances . . . .”).

340 See Gerhardt, supra note 32, at 62 (suggesting that the “building of coalitions” means that the opinion’s “content is a function of the majority’s preferences”).

341429 U.S. 190 (1976).

342 Epstein & Knight, supra note 52, at 5–10.

343 See Maltzman et al., supra note 217, at 94.

344Chief Justice Rehnquist observed that decision making “inevitably has a large individual component,” but that it is “filtered through the deliberative process of the court as a body.” William H. Rehnquist, Remarks on the Process of Judging, 49 Wash. & Lee L. Rev. 263, 270 (1992).

345 See Pablo T. Spiller, The Choices Justices Make, By Lee Epstein & Jack Knight, 94 Am. Pol. Sci. Rev. 943, 943 (2000) (book review) (“Once the median policy is proposed, no other proposal will beat it, and it becomes the outcome.”). The general theory provides that “Supreme Court opinion authors make strategic calculations about the need to craft opinions that are acceptable to their colleagues on the bench.” Paul J. Wahlbeck et al., Marshalling the Court: Bargaining and Accommodation on the United States Supreme Court, 42 Am. J. Pol. Sci. 294, 294 (1998).

346Lax & Cameron, supra note 261, at 276–77 (“If the Median Voter Theorem applies, the content of every Supreme Court opinion must devolve to the wishes of the median justice; the identity and preferences of the opinion’s author . . . cannot matter.” (citation omitted)).

347 See, e.g., James F. Spriggs II et al., Bargaining on the U.S. Supreme Court: Justices’ Responses to Majority Opinion Drafts, 61 J. Pol. 485 (1999). Sometimes, Justices explicitly refuse to join a draft opinion. Id. at 487–88. The vast majority of Justices in the original majority coalition simply join the opinion without challenge, however. Id. at 498 (noting that this is the case for over 80% of the Justices). When accommodations are made to other Justices, they reflect numerous concerns, rather than simply conforming to preferences of the median Justice. See Wahlbeck et al., supra note 345.

348 See Epstein & Knight, supra note 52, at 58–79 (discussing bargaining over the opinion).

349Black & Spriggs, supra note 112, at 662.

350 See, e.g., Virginia A. Hettinger et al., Separate Opinion Writing on the United States Courts of Appeals, 31 Am. Pol. Res. 215 (2003) (discussing costs of separate opinion writing); Lax & Cameron, supra note 261 (modeling the opinion content based on the cost of writing separately).

351See Lax & Cameron, supra note 261, at 277 (noting that if writing a Supreme Court opinion takes “costly time and effort,” this fact would create a “wedge” that the assigned opinion author could use “to move an opinion away from the median justice’s most preferred policy”).

352Cliff Carrubba et al., Does the Median Justice Control the Content of Supreme Court Opinions? (Sept. 29, 2010) (unpublished manuscript), available at http://www.allacademic.com/meta/p360479_index.html.

353Chris W. Bonneau et al., Agenda Control, the Median Justice, and the Majority Opinion on the U.S. Supreme Court, 51 Am. J. Pol. Sci. 890 (2007).

354Given the large number of control variables, the remaining errors are likely to be random. See Cross & Lindquist, supra note 93, at 1392 (“[E]rrors are likely to be randomly distributed throughout the judicial population and may therefore be considered statistical ‘noise.’ The presence of this feature tends to make it more difficult to find true statistical significance and consequently may add further confidence to results that find such significance.” (footnote omitted)); see also Staudt et al., supra note 185, at 376 (noting that measurement error “is absorbed into the disturbance term” so that the error causes an underestimate of true effects); Richard A. Posner, The Theory and Practice of Citations Analysis, with Special Reference to Law and Economics 12–13 (Univ. of Chi. Law Sch., John M. Olin Law & Econ. Working Paper No. 83, 1999), available at http://www.law.uchicago.edu/files/files/83.RAP_.Citations-40532.pdf (“Critics of citations analysis often fail to note that if errors in data are randomly distributed with respect to the variable of interest . . . they are unlikely to invalidate the conclusions of the study, provided that the data sample is large.”).

355Cook, supra note 1, at 1128. Thus, an opinion author may be able to turn a “little fish” of a case into a “choice morsel.” Id.

356Abe Fortas, Chief Justice Warren: The Enigma of Leadership, 84 Yale L.J. 405, 405 (1975).

357Forrest Maltzman & Paul J. Wahlbeck, Opinion Assignment on the Rehnquist Court, 89 Judicature 121, 122 (2005).

358 Richard A. Posner, Economic Analysis of Law 585 (7th ed. 2007).

359 See Cass R. Sunstein, One Case at a Time: Judicial Minimalism on the Supreme Court (1999) (setting out the theory of judicial minimalism).

360The minimalist Justice strives to “resolve the problem at hand without also resolving a series of other problems that might have relevant differences.” Sunstein, supra note 325, at 29.

361 Sunstein, supra note 359, at 3.

362 See J. Woodford Howard, Jr., Courts of Appeals in the Federal Judicial System: A Study of the Second, Fifth, and District of Columbia Circuits 160–62 (1981).

363 See, e.g., Cynthia L. Cates & Wayne V. McIntosh, Retail Jurisprudence: The Judge as Entrepreneur in the Marketplace of Ideas, 11 J.L. & Pol. 709 (1995) (discussing the notion of judicial entrepreneurship).

364 Sunstein, supra note 359, at 9. Sunstein focuses on the contemporaneous Court, but historic Justices might be similarly categorized. See, e.g., Sheldon Gelman, The Hedgehog, the Fox, and the Minimalist, 89 Geo. L.J. 2297, 2303 n.38 (2001) (reviewing Sunstein, supra note 359) (describing Justice White as minimalist in orientation).

365 Sunstein, supra note 325, at 26.

366 See Landes & Posner, supra note 8, at 268 (describing why more specific precedents will depreciate faster than more general ones).

367 Sunstein, supra note 359, at 3.

368 Id. at 19. Thus, maximalists “may be quite surprised by the conduct of subsequent courts, which characterize prior language as ‘dicta’ . . . . [that] turns prior decisions into minimalist ones.” Id. at 20.

369 See Choi et al., supra note 333, at 1322 (contending that if minimalist decisions indeed produce better law, “minimalist opinions will be cited more, not the creative and expansive ones”).

370 See, e.g., Robert Anderson IV, Measuring Meta-Doctrine: An Empirical Assessment of Judicial Minimalism in the Supreme Court, 32 Harv. J.L. & Pub. Pol’y 1045 (2009).

371 Id. at 1063–68.

372Kosma, supra note 8, at 340.

373The self-assignment effect has been studied, with early research finding that Chief Justices tend to assign themselves important cases as well as those decided unanimously. Elliot E. Slotnick, The Chief Justices and Self-Assignment of Majority Opinions: A Research Note, 31 W. Pol. Q. 219, 225 (1978). These findings were confirmed in Saul Brenner, Strategic Choice and Opinion Assignment on the U.S. Supreme Court: A Reexamination, 35 W. Pol. Q. 204 (1982). The significance of this effect is modified, though, by the Chief Justice’s need to balance other goals, such as Court harmony. See Forrest Maltzman & Paul J. Wahlbeck, A Conditional Model of Opinion Assignment on the Supreme Court, 57 Pol. Res. Q. 551 (2004). Opinion assignments may also be influenced by a desire to punish or reward members of the Court. Cross & Lindquist, supra note 268, at 1673. External perception of the opinion may also influence opinion assignment. See David W. Rohde, Policy Goals, Strategic Choice and Majority Opinion Assignments in the U.S. Supreme Court, 16 Midwest J. Pol. Sci. 652, 677–78 (1972). Moreover, the Chief may need to assign important cases to the most moderate member of the majority coalition in divided decisions in order to hold the majority. See Theodore S. Arrington & Saul Brenner, Testing Murphy’s Strategic Model: Assigning the Majority Opinion to the Marginal Justice in the Conference Coalition on the U.S. Supreme Court, 36 Am. Pol. Res. 416 (2008). The effect will also vary by Chief Justice; one study found that Chief Justice Rehnquist did not self-assign especially important decisions of his Court. Forrest Maltzman & Paul J. Wahlbeck, May It Please the Chief? Opinion Assignments in the Rehnquist Court, 40 Am. J. Pol. Sci. 421, 421 (1996).

374The second Justice Harlan, for example, has been characterized as a “great dissenter.” Tinsley E. Yarbrough, John Marshall Harlan: Great Dissenter of the Warren Court (1992). This might be ascribed to the fact that he was ideologically out of step with the liberal Warren Court majority. This honorific was also given to Justice Holmes. See William J. Brennan, Jr., In Defense of Dissents, 37 Hastings L.J. 427, 429 (1986); M. Todd Henderson, From Seriatim to Consensus and Back Again: A Theory of Dissent, 2007 Sup. Ct. Rev. 283, 284 n.7.

375For a quantitative measure of the Justices least likely to join majority opinions each year since 1956, see Peter A. Hook, The Aggregate Harmony Metric and a Statistical and Visual Contextualization of the Rehnquist Court: 50 Years of Data, 24 Const. Comment. 221, 244–46 (2007).

376Our study controls for this bias somewhat, with the variables for minimum-winning coalition and unanimity.

377Bernard Schwartz, Supreme Court Superstars: The Ten Greatest Justices, 31 Tulsa L.J. 93, 148 (1995).

378Kosma, supra note 8.

379 Id. at 351.

380 Id. at 353 n.42.

381The classic of this genre is Herbert Wechsler, Toward Neutral Principles of Constitutional Law, 73 Harv. L. Rev. 1 (1959) (concluding that the opinion in Brown could not be justified as principled decision making). Some considered the opinion to be “a travesty of judicial craft.” Lani Guinier, Demosprudence Through Dissent, 122 Harv. L. Rev. 4, 134 (2008). The opinion’s reliance on children’s reactions to different racial dolls was both legally questionable and grounded in relatively bad social science. See Richard Kluger, Simple Justice: The History of Brown v. Board of Education and Black America’s Struggle for Equality 355–56 (1976) (addressing criticisms of the study relied upon); Michael Heise, Judicial Decision-Making, Social Science Evidence, and Equal Educational Opportunity: Uneasy Relations and Uncertain Futures, 31 Seattle U. L. Rev. 863, 867–68 (2008).

382 See Cross & Lindquist, supra note 93, at 1421. Even after criticizing the opinion in Brown, Wechsler wrote that it had “the best chance of making an enduring contribution to the quality of our society of any [opinion] that I know in recent years.” Wechsler, supra note 381, at 27.

383 See, e.g., Choi & Gulati, supra note 8.

384Schaefer, supra note 109, at 11.

385Farber, supra note 106, at 1179. Some dispute this hypothesis, suggesting that citations do not fully capture the true judicial virtues. See Cross & Lindquist, supra note 93, at 1391–93. Those virtues are explored in Lawrence B. Solum, The Aretaic Turn in Constitutional Theory, 70 Brook. L. Rev. 475 (2005). The citation measure is surely imperfect, but it has some validity and would appear to capture some of the key judicial virtues, while preserving objectivity. See Cross & Lindquist, supra note 93, at 1393–95.

386Knight, supra note 6, at 1553.

387Kosma, supra note 8, at 360–62.

388 See, e.g., Frederick Schauer, Precedent, 39 Stan. L. Rev. 571, 599 (1987) (discussing this efficiency rationale for precedent).

389Gely, supra note 215, at 107.

39060 U.S. (19 How.) 393 (1857) (holding that slaves did not enjoy constitutional protections of citizenship).

391323 U.S. 214 (1944) (sanctioning the curtailment of civil rights for one targeted racial group).

392 See Black & Spriggs, supra note 220.

393Research shows that for lower court usage, Justice Scalia’s opinions also have a disproportionately high rate of negative citations, though these remain a small fraction of his total citations. See Frank B. Cross, Determinants of Citations to Supreme Court Opinions (and the Remarkable Influence of Justice Scalia), 18 Sup. Ct. Econ. Rev. 177 (2010).