measuring the influence of supreme...

MEASURING THE INFLUENCE OFSUPREME COURT JUSTICES

MONTGOMERY N. KOSMA*

Abstract

This empirical study measures the influence of 99 retired Supreme Court justices,analyzing over 1.2 million citations to over 24,000 opinions of the Court writtenbetween 1793 and 1991. It models the appointment process as the selection of acapital investment, treating a justice’s output as the precedents generated each termand using citations as a proxy for an opinion’s value. This model is applied to theretired justices and their opinions, and its consistency is tested by independentlyanalyzing citations by subsequent Supreme Court and circuit court opinions. Influ-ence values also demonstrably track the results of a well-known survey of judicialgreatness. The study challenges several common assumptions. Older appointeeshave been no less influential than young appointees, and, on an annual basis, olderappointees have actually been more influential. Private attorneys have made themost influential appointees, and former judges show no special advantages.

Like a financial manager selecting a capital investment, a presidentchoosing a Supreme Court nominee must analyze myriad factors in at-tempting to select the most ‘‘profitable’’ candidate. How profits are defineddepends on the investment goal; for example, a president seeking supportfor a particular policy (for example, paper currency) might focus on predict-able short-term votes, whereas one looking to advance a broader philosophy(for example, originalism) might seek a voice with longevity, authority, andconsistency.

Although a president’s political agenda does not readily lend itself tomore than anecdotal analysis, all presidents do share a common goal intheir Supreme Court nominations: to select candidates who will exert sig-

* Law clerk, the Honorable David B. Sentelle, U.S. Court of Appeals for the D.C. Circuit.I am indebted to Professor William Landes for his substantial assistance in preparing thispaper for publication, as well as to Professor Tyler Shumway and Art Pasternak, who re-viewed early drafts. My thanks also to Chris Bowers, Ross Davies, and Kelly Sutton for theirinspiring discussions during the course of this research and to the John M. Olin Foundationand the Bradley Foundation, which provided financial support.

[Journal of Legal Studies, vol. XXVII (June 1998)] 1998 by The University of Chicago. All rights reserved. 0047-2530/98/2702-0003$01.50

333

334 THE JOURNAL OF LEGAL STUDIES

nificant influence over the subsequent development of legal doctrines. Likea capital investment, which a manager expects to generate positive futurecash flows, a Supreme Court appointee provides a ‘‘return’’ on investmentin the form of periodic generation of influential precedents.

Section I of this study sets forth an economic model for the evaluationof Supreme Court justices. Using citations in subsequent Supreme Courtopinions as a proxy for influence over legal developments, the model de-fines a justice’s Historical Value, the overall level of contribution, and Ap-pointment Value, the return on investment from the perspective of the ap-pointing president. Section II applies the model to the justices and theirmore than 24,000 opinions of the Court, tests its consistency using circuitcourt citation data, and compares the results to a well-known survey of judi-cial greatness. Section III examines the selection process, including the rel-ative influence of each president, the age and prior experience of nominees,and the effects of Senate advice and consent.

I. An Investment Model of Supreme Court Nominations

A. The President as an Investor

The President . . . shall nominate, and by and with the Adviceand Consent of the Senate, shall appoint . . . Judges of the su-preme Court.1

One of the most enduring marks left by a president is undoubtedly hiseffect on the law through his nominations to the Supreme Court.2 Despitethe limits imposed by the Senate confirmation process, the selection of Su-preme Court justices remains an essential part of the presidential preroga-tive.3 Indeed, our constitutional scheme represents a compromise between

1 U.S. Const, art. II, § 2 cl. 2.2 Shortly after President Reagan’s election, President Nixon advised him that ‘‘[your] most

lasting legacy will be your impact on the Supreme Court.’’ William French Smith, Law andJustice in the Reagan Administration: The Memoirs of an Attorney General 58 (1991).

3 The Constitution plainly indicates that the ‘‘Advice and Consent of the Senate’’ appliesto the appointment, but not the nomination, of judges. Guido Calabresi noted that recent pres-idents ‘‘have continued to name people whom they thought would share their views, and thatis their right in the first instance.’’ Hearing of the Senate Judiciary Committee on the Nomi-nation of Clarence Thomas to the Supreme Court (September 17, 1991). Of course, as Tribereminds us, Senate influence over nominations should not be discounted. See Laurence H.Tribe, God Save This Honorable Court: How the Choice of Supreme Court Justices ShapesOur History 89 (1985); see also Federalist 76 (Hamilton), in The Federalist Papers 513–14(Jacob E. Cooke ed. 1961).

INFLUENCE OF SUPREME COURT JUSTICES 335

those who feared the centralization of executive power4 and those who real-ized that a deliberative political body cannot effectively select judges.5

The Framers rightly assessed the importance of Supreme Court appoint-ments—if anything, they underestimated the degree to which national polit-ical battles would be played out before the Court. With an average of 2.7Supreme Court appointments per president, most chief executives havebeen able to indirectly extend their influence over the development of lawand public policy far beyond the expiration of their own term of office.

Presidents throughout history have attempted to maximize this legacy ofinfluence through the strategic selection of Supreme Court nominees. Likea careful financial manager, a president seeks a candidate who will producethe maximum ‘‘return.’’ This return is generally defined in political terms:votes on particular issues, support for a certain jurisprudential philosophy,or satisfaction of some constituency.6 Political favors aside, an appointee’sreturn will be determined in large part by his influence over the develop-ment of the law.

As with any investment, a Supreme Court nominee comes with certainrisks. Some of these risks are realized immediately and depend primarilyon the individual nominee and the political circumstances. For example, acandidate found wanting in experience, or politically unacceptable, may notsurvive Senate confirmation.

Even if confirmed by the Senate, an inexperienced candidate remains a

4 ‘‘Mr. Rutledge [the future justice] was by no means disposed to grant so great a powerto any single person. The people will think we are leaning too much toward Monarchy.’’ 1The Records of the Federal Convention of 1787, at 119 (Max Farrand ed., Yale rev. ed. 1937)(June 5, 1787).

5 Madison, trusting neither the executive nor the legislature alone, offered several propos-als, including executive nomination that ‘‘should become an appointment if not disagreed towithin days by 2/3 of the 2d. branch.’’ Id. at 2:44 (July 26, 1787). Gouverneur Morrisbest summarized the final compromise: ‘‘[A]s the President was to nominate, there would beresponsibility, and as the Senate was to concur, there would be security.’’ Id. at 2:539 (Sep-tember 7, 1787).

6 Henry Abraham borrows the term ‘‘real politics’’ from President Theodore Roosevelt indocumenting the historical concern of nominating presidents: ‘‘Whatever the merits of theother criteria attending Presidential motivations in appointments may be, what must be ofoverriding concern to any nominator is his perception of the candidate’s real politics. Thechief executive’s crucial predictive judgment concerns itself with the nominee’s likely futurevoting pattern on the bench, based on his or her past stance and commitment on matters ofpublic policy insofar as they are reliably discernible. All Presidents have tried, thus, to packthe bench to a greater or lesser extent.’’ Henry J. Abraham, Justices and Presidents: A Politi-cal History of Appointments to the Supreme Court 66 (3d ed. 1992). See also William O.Douglas, The Court Years (1939–1975): The Autobiography of William O. Douglas 242(1980) (‘‘Most Presidents name Justices who, they think, will vote the way they would vote.That is what I would do were I President.’’).


significant risk because of the president’s lack of information. The qualityof a ‘‘stealth’’ nominee’s legal work may not as predictable as that of onewith a more significant track record. Young appointees (to the delight ofthose originally opposing the nomination) have sometimes ‘‘grown in of-fice,’’ exerting influence in other than the anticipated direction.7 Since theyoung appointee will likely have a long career,8 any miscalculation comesat great cost. Attempting to offset this risk may explain the recent rise inthe selection of candidates with prior judicial experience.9

An older candidate, however, has a different profile of risks. While recentyears have shown that there is some risk in selecting an individual with atrack record of controversial legal scholarship,10 such writings only affectthe likelihood of rejection by the Senate. More significant is the risk thatan older appointee will have a shorter tenure on the Court, thereby—atleast, as has been commonly thought—reducing his expected influence rela-tive to a younger appointee.

Using this analogy, I develop in the next section an economic model ofthe nominee as an investment, which will then allow assessment of pastjustices and a critical evaluation of some of these common assumptions.

B. The Investment Model of Judicial Appointments

Consider briefly the net present value (NPV) of a typical capital invest-ment, an asset that generates periodic cash flows:

NPV 5 C0 1t51

Ct

(1 1 r)t, (1)

where r is the discount rate (assumed constant), C0 the initial cash outlay,and Ct the cash flow in time period t. When selecting between mutuallyexclusive investment options, the preferred choice has the highest net pres-ent value.

Similarly, when evaluating a potential nominee, a president seeks a can-didate with a high expected influence over the law. A justice periodicallygenerates output in the form of precedents. This output can be analogizedto the annual cash flow generated by a capital asset. Calculating the ‘‘netpresent value’’ of an appointee requires quantifying the legal influence of

7 See note 34 infra and accompanying text.8 See Section IIIB1 infra.9 See note 52 and Section IIIB2 infra.10 See, for example, S. Scott Gaille, Publishing by United States Court of Appeals Judges:

Before and After the Bork Hearings, 26 J. Legal Stud. 371 (1997); Abraham, supra note 6,at 357–58 (discussing the Bork confirmation).


precedents generated in each year of his career and applying the net presentvalue formulation with an appropriate discount rate.

A president may be thought to take one of two possible approaches.When seeking to use a nomination to effectuate some policy or influencethe direction of particular legal developments, a president focuses on thepolitical returns flowing from his nominee and may deem influence oc-curring far in the future less valuable than any immediate effect generatedby his candidate, both because of its remoteness in time and the difficultyin predicting how the issues before the Court or the nominee’s views onthem may change. In this situation, a candidate should be evaluated byapplying a positive discount rate to his stream of future legal influence.11

This discounted influence shall be called a justice’s Appointment Value(AV) and is defined as NPV using an appropriate positive discount rate r.Other presidents may focus on leaving a historical legacy through an ap-pointment, such that all legal influence exerted by the candidate shouldcount equally, regardless of when it occurs relative to the time of appoint-ment. This measure of influence shall be called a justice’s Historical Value(HV) and is defined as NPV using a discount rate r equal to zero.

Computing these values will require a method for measuring each opin-ion’s influence over future legal developments, an ‘‘inflation’’ adjustmentto allow comparing measurements from different eras, and, for AppointmentValue, an appropriate estimate of a discount rate.12

1. Measuring Influence over Legal Development

With certain adaptations, a scheme developed by William Landes andRichard Posner lends itself well to the quantification of judicial influence,using citations to a judge’s opinions as a proxy for his influence over futurelegal developments. The modern field of legal citation analysis developedout of their 1976 paper describing an economic model treating a body oflegal precedents as depreciable stock of legal capital.13 They postulated that

11 See notes 33–36 infra and accompanying text.12 This analysis assumes that the initial investment cost is zero or, at least, that the cost is

comparable for all nominees, and hence drops out of any comparisons. This cost—the politi-cal capital required to shepherd a nominee through Senate confirmation—is difficult to mea-sure and impossible to describe in terms compatible with future influence over the law. Fur-ther, once a nominee has been confirmed, life tenure makes any initial cost in political termsirrelevant to the evaluation of that individual’s performance as a member of the Court. Ofcourse, such costs remain an important part of the president’s screening and selection pro-cess.

13 William M. Landes & Richard A. Posner, Legal Precedent: A Theoretical and EmpiricalAnalysis, 19 J. Law & Econ. 249 (1976). The initial Landes and Posner study, while mostimportant, was not the first such analysis. In the legal world, Merryman performed citationstudies of the California courts from the 1950s through 1970s. See, for example, John HenryMerryman, The Authority of Authority, 6 Stan. L. Rev. 613 (1954). In the scholarly journals


a precedent has a value that can be represented by the number of times itis cited in subsequent opinions. Since most opinions become less relevantover time, as the law and the surrounding culture change and develop, theymodeled the decreasing number of citations over time as depreciation of acapital stock. They then applied their formulation to a sample of around1,000 opinions from the 1960s and 1970s. This initial study provided thefoundation for a fairly rigorous treatment of citation statistics, and sincethen, several other papers and books have been published using citations asa proxy for judicial reputation and for tracking the development and depre-ciation of precedent as legal capital.14

Measuring a justice’s influence on the law falls neatly within the obviousboundaries of citation analysis.15 Authored opinions represent the primarywork product of members of the Court, and the frequency with which aparticular opinion has been cited generally indicates how often it has influ-enced the resolution of subsequent cases.16 Thus, during a given year, a jus-

of economics and other social and physical sciences, several earlier citation studies were per-formed examining the relevance and influence of published articles in the field. See, for ex-ample, Robert K. Merton, The Sociology of Science: Theoretical and Empirical Investiga-tions 508–9, 514–15, 556 (1973). However, as Landes and Posner point out, citations ofprecedent in court opinions differ fundamentally from scholarly citations owing to the roleof authority and doctrines like stare decisis in judicial decision making. See Landes &Posner, supra, at 250–51.

14 Studies using citations to measure influence or reputation include William M. Landes,Lawrence Lessig, & Michael Solimine, Judicial Influence: A Citation Analysis of FederalCourts of Appeals Judges, in this issue, at 271 (ranking the influence of federal appellatejudges); Richard A. Posner, Aging and Old Age 180–201 (Chicago 1995) (using citationanalysis to show that although there is an aging-related decline in judging, it may not set inuntil an unusually advanced age); Richard A. Posner, Cardozo: A Study in Reputation 74–91 (Chicago 1990) (applying citation analysis to compare Cardozo’s reputation to his con-temporaries’). For studies based on a legal capital model, see William M. Landes & RichardA. Posner, Legal Change, Judicial Behavior, and the Diversity Jurisdiction, 9 J. Legal Stud.367 (1980); and Landes & Posner, supra note 13, at 249.

15 Using citation counts to measure influence over the development of law differs subtlyfrom past usage in the study of reputation. Reputation is generally defined as ‘‘a pro-attitudeby other people toward the person ‘whose’ reputation is in issue. . . . [T]he point to be em-phasized is that reputation is conferred by the people doing the reputing rather than producedby the reputed one—and it is conferred for their purposes, not his.’’ Posner, Cardozo, supranote 14, at 59. Reputation studies focus most closely on the citation of nonbinding prece-dents, such as citations of one circuit court of appeals by another. See Landes, Lessig, &Solimine, supra note 14, at 272–73. Here, considering the influence over law exerted by aSupreme Court justice, one need not distinguish between binding and merely persuasive pre-cedents—every citation represents a more or less equal quantum of influence.

16 Following past practice, this study includes citations that distinguish the earlier case be-cause such citations represent an inability to ignore that precedent. See, for example, Landes,Lessig, & Solimine, supra note 14, at 273. In contrast, citations that do not represent a prece-dent’s influence—for example, those that appear in a dissenting opinion, or in an opinionoverruling the precedent—are excluded. See note 37 infra for more details on the citationsincluded in this study.


tice’s influence will be taken to be the sum of the citations to all of hisopinions written during that year.

Other works have addressed most of the common objections to citationanalysis,17 but it is instructive at this point to consider certain potential prob-lems with this type of study. Two hypothetical effects might distort the rela-tionship between number of citations and degree of influence. First, a prece-dent may set such a clear legal standard that subsequent cases settle ratherthan go to trial and appeal, resulting in few citations to the case in lateropinions.18 Such a ‘‘superprecedent’’ (which obviously exerts great influ-ence) would be underrepresented in citation counts. Second, another typeof precedent might leave the law unclear, inducing much further litigationas the courts try to resolve its ambiguity or test the boundaries of the newdoctrine. Such an opinion, with marginal or questionable influence, may beoverrepresented when counting citations.

The first effect probably does not represent a serious problem. As ex-plained by Landes and Posner, such ‘‘superprecedents’’ are unlikely to havea statistically significant effect: ‘‘If a case is highly specific, it will hardlyqualify as a ‘superprecedent’; by definition it will control only those infre-quent cases that present virtually identical facts to those of the case inwhich it was announced. If it is highly general, and therefore more likelyto be an important precedent, it is unlikely to decide—so clearly as to pre-vent disputes or litigation from arising—the specific form of the questionpresented in subsequent cases.’’19 Further, any such case will inevitablyhave some degree of ‘‘open texture’’ where the enunciated legal rule doesnot obviously apply.20 Even were such an impressive settlement-generatingopinion to appear, courts would likely recognize its influence and cite itfrequently (perhaps using similar reasoning in a different context), ratherthan ignore it as this argument implies. Also, as will be demonstrated, evena justice’s most influential opinion constitutes only a small fraction (on av-

17 See, for example, Landes, Lessig, & Solimine, supra note 14, at 272–76.18 Such an opinion has been termed a ‘‘superprecedent’’ and is described by Landes and

Posner. See Landes & Posner, supra note 13, at 251; see also Landes, Lessig, & Solimine,supra note 14, at 274.

19 Landes and Posner, supra note 13, at 251.20 See H. L. A. Hart, The Concept of Law 119–20 (1961). For example, Gideon v. Wain-

right, 372 U.S. 335 (1963), appeared to enunciate a broad per se rule regarding the right tocounsel. Soon thereafter, Gideon was limited to cases of felony prosecutions. See In re Gault,387 U.S. 1 (1967). In this holding it remains unchallenged; however, in the fringe of theGideon rule, there remains room for argument about whether it applies or whether counter-vailing principles limit the right to counsel. See, for example, Argersinger v. Hamlin, 407U.S. 25 (1972) (right to counsel even in cases of ‘‘petty’’ crimes punished by less than 6months imprisonment); Gagnon v. Scarpelli, 411 U.S. 778 (1973) (no right to counsel inhearings to revoke probation). Like any other legal rule, a ‘‘superprecedent’’ will have thisfringe of vagueness and will likely induce further litigation to determine its boundaries.


erage, 4.7 percent) of his total influence.21 Thus, if the number of ‘‘super-precedents’’ is small, and not unusually distributed, we can expect any dis-tortion to be of minor significance.

The seemingly ambiguous precedent is unlikely to be overrepresented incitation counts. One expects that an opinion is cited only as long as it helpsto resolve cases.22 When a more suitable precedent comes along, citations tothe less useful earlier opinion should dwindle. In contrast, if a more helpfulprecedent does not arise, it may be that the first precedent represents anadequate resolution of a tough legal problem and that the initial low esti-mate of its influence, because it seemed vague or unhelpful, was simplywrong. Thus, a truly unhelpful precedent should have a small number ofcitations, but if an apparently unhelpful precedent has a high number of ci-tations, it may be that it was more useful than one might have guessed ini-tially.

Several other factors should be kept in mind, not because of potential todistort citation counts, but as implicit within the definition of influence. Thetraditional responsibility of the chief justice to assign the writing of opin-ions suggests that, to the degree the chief retains control over particularlyimportant cases, his opinions may be more frequently cited than others.23

This is merely another way of saying that the chief justice may tend to be(at least marginally) more influential. The justice in the center chair may bein a position to exert more influence than an associate justice of equal oreven greater innate ability, but this serves to remind us that influence maynot be perfectly correlated with talent.

This methodology may also tend to magnify the influence of the moder-ate ‘‘swing vote’’ justice.24 A justice who is in the middle of a highly di-vided Court will tend to be in the majority more often than the others andwill therefore (if opinions are assigned randomly) author a disproportionateshare of majority opinions, with a presumably higher chance of accumulat-ing citations in the future. One might respond that it is not clear that opin-ions are assigned in such manner or that authorship of opinions in closecases guarantees future citations—an ineffective or unclear opinion is likelyto be superseded. The central point remains the difference between influ-

21 See Table 5 infra and accompanying text.22 Except perhaps for citations like ‘‘Whichever view is taken of Aherns v. Clark, the case

is of little help here.’’ Schlanger v. Seamans, 401 U.S. 487, 489–90 (1971), discussingAherns v. Clark, 335 U.S. 188 (1948). Of course, the Court here did not call Aherns worth-less but simply noted that despite the party’s argument, it did not apply.

23 See text accompanying note 41 infra. Opinions authored by the chief might also bethought to carry more authority than those of other justices.

24 I am indebted to David Currie for noting his skepticism in attributing higher ‘‘value’’to the swing voter and for focusing attention on the distinction inherent in this study betweeninfluence and intellect.


ence and talent: like the chief justice, a swing voter of average ability mayactually exert more influence over the law than a truly great dissenter.

Similarly, a justice’s influence may ebb and flow over time as the compo-sition of the Court changes. Because citations to dissenting opinions are notcounted, the influence of a justice found often in dissent may be under-stated. When a justice disagrees with the rule laid down by the Court, un-less the original opinion is later limited or overruled, the dissent generallyexerts little influence over the future development of legal doctrines. Thisproblem is similar to that which comes from limiting measurement of in-fluence to citations in subsequent cases—other avenues of influence, suchas discussions among the chambers, are completely ignored. In the end, theinfluence measured by this study must be understood as somewhat limitedin its breadth.25

Finally, this analysis does not exclude ‘‘self-citations’’—the citation bya justice of his own prior opinion.26 Such ‘‘judicial narcissism’’ may distorta study of reputation that uses citations to measure how others regard anopinion’s author.27 In contrast, a self-citation should not lower the assess-ment of a precedent’s influence because there is little reason to believe thatjustices cite their own opinions when it is not useful to do so.

2. A ‘‘Price Index’’ for Citations

Citation studies to date have only compared opinions of approximatelythe same vintage.28 To gain a complete picture of the Supreme Court, thisstudy considers its entire history. Because of increasing case volume, aswell as other changes such as longer opinions and increasing reliance onlaw clerks,29 citation counts have undergone inflation—in other words, twoopinions with equivalent influence may have different nominal citationcounts. Figure 1 depicts the fluctuating number of opinions written each

25 It may be that these factors raise more concerns in theory than in fact. For example,considering Table 1 infra, most of the justices known as ‘‘great dissenters’’ score quite wellbased solely on citations to their majority opinions. It does not seem that there were greatdissenters who were not also great justices overall. A more complete resolution of this issueawaits a future study based on data that includes citations to concurring and dissenting opin-ions.

26 See, for example, Landes, Lessig, & Solimine, supra note 14, at 278 n.19 (‘‘Self-citations are more a reflection of how A values A’s work, while non-self-citations reflecthow others value A’s work.’’).

27 See note 15 supra.28 For example, the initial Landes and Posner study used a random sample of 658 decisions

(roughly one in 10) over a period of 18 months. See Landes & Posner, supra note 13, at 251.The most ambitious study to date included all court of appeals citations to opinions appearingin Federal Reporter between 1955 and 1995 written by 205 sitting court of appeals judges.See Landes, Lessig, & Solimine, supra note 14, at 276–77.

29 See Richard A. Posner, The Federal Courts: Challenge and Reform 148 (1996) (‘‘Lawclerks . . . feel naked unless they are quoting and citing cases and other authorities.’’).

Figu

re1.

—A

nnua

lnu

mbe

rof

opin

ions

wri

tten


year by the Supreme Court (black line) and the circuit courts (gray line).Figure 2 shows the generally increasing number of citations containedwithin the average Supreme Court opinion. Figure 3 shows, for all the Su-preme Court precedents authored each year, the total number of subsequentcitations of those opinions by the Supreme Court and by the circuit courts.Figure 4 shows the average number of subsequent citations by the SupremeCourt and circuit courts.30

To properly compare the influence of opinions from different eras, wemust create a ‘‘price index’’ to adjust for these fluctuating and generallyinflationary trends. The price index is based on an important assumption:that the influence of the average Supreme Court opinion has remained con-stant over time.31 Then, the average number of citations per opinion duringa given year can be used to adjust citation counts to a uniform scale. For-mally, the citation price index CPIt for year t is defined as

CPIt 5Nt

^Nt

j51

c0t, j

⋅ κ , (2)

where Nt is the number of opinions in year t, c0t, j is the raw citation count

for the jth opinion in year t, and κ is a normalization constant chosen suchthat the smallest index value will equal 1.0. The CPI values are smoothedvia a 5-year weighted average to alleviate distortions, not present in a typi-cal large market, where the presence of a single influential opinion causesan inordinate effect on the entire year’s average.32 The inflation-adjustedcount ct, j for the jth opinion in year t is simply

ct, j 5 c0t, j ⋅ CPIt. (3)

30 Figures 3 and 4 show that citations trail off drastically in the 1980s, a truncation re-sulting from the inability to count future citations to these opinions. (A depreciation analysisshows that the ‘‘half-life’’ of a Supreme Court opinion—the time it takes to accumulate one-half of its total expected citations—is about 14 years. See note 36 infra.) The price indexwill adjust for this truncation effect, though as the number of citations decreases, the smallersample size leads to greater uncertainty.

31 This assumption may be open to refinement but should provide a reasonable first-orderapproximation. To test this assumption, Supreme Court and circuit court citations will beseparately adjusted and compared. To the extent that opinions should exhibit a similar degreeof influence in the Supreme Court and in the circuit courts, then finding a strong correlationbetween influence as computed from the two independent sets of price-index-adjusted cita-tion data would support the validity of this assumption. See text accompanying notes 45–46infra for results supporting this hypothesis.

32 This effect is most pronounced in the early years, when the Court wrote relatively fewopinions but occasionally produced a Marbury or a McCulloch. This is illustrated by thedramatic fluctuation in average citations to opinions written between the years 1800 and1825, as shown in Figure 4.

Figu

re2.

—A

vera

genu

mbe

rof

cita

tions

cont

aine

dw

ithin

Supr

eme

Cou

rtop

inio

ns

344

Figu

re3.

—T

otal

subs

eque

ntci

tatio

nsto

Supr

eme

Cou

rtpr

eced

ents

gene

rate

dea

chye

ar

345

Figu

re4.

—A

vera

gesu

bseq

uent

cita

tions

toSu

prem

eC

ourt

prec

eden

tsge

nera

ted

each

year


Figure 5 shows the CPI, separately computed based on subsequent citationsin Supreme Court and circuit court opinions.

For example, each opinion written in the year 1900 was cited (in subse-quent Supreme Court cases) an average of 7.1 times, compared to 10.9 cita-tions per opinion written in 1960. Based on the above formulation, the CPIfor 1900 turns out to be 3.6, compared to a CPI of 2.0 for 1960. A 1900opinion that nominally received 10 citations would have an inflation-adjusted count of 36, as would a 1960 opinion that nominally received 18citations. Thus, a 1900 opinion that has been cited 10 times is deemed tohave influence equivalent to a 1960 opinion that has been cited 18 times.

3. Discount Rate for Appointment Value

A discount rate normally represents the time value of money—a dollartoday can be invested and return more than a dollar tomorrow.33 Similarly,in political terms, an influential Supreme Court opinion in year n maybe worth more to the appointing president than the same opinion in yearn 1 1. Several factors can explain this ‘‘time value of an opinion.’’There is in some sense an opportunity cost—an opinion issued earlierwill have the opportunity to influence more cases. Earlier opinions may bemore valuable to the appointing president because of potential politicalvalue while he remains in office. Finally, early opinions may be more valu-able because they are more likely to be consistent with the president’spolitical desires—the difficulty in predicting how an individual willvote increases with time, as individuals change and the issues before theCourt diverge from those most important at the time of appointment.34 Thus,

33 See Richard A. Brealey & Stewart C. Myers, Principles of Corporate Finance 13–14,149 (4th ed. 1991). Discount rates often include a component for risk, but because all thedata considered here are historical, any uncertainties have already been realized.

34 In the words of Alexander Bickel, ‘‘You shoot an arrow into a far-distant future whenyou appoint a Justice and not the man himself can tell you what he will think about someof the problems that he will face.’’ Time, May 23, 1969, at 24, quoted in Abraham, supranote 6, at 70.

Indeed, such a description has been used of Madison’s appointment of Story (Story ‘‘out-Marshalled Marshall in his nationalism’’), Teddy Roosevelt’s appointment of Holmes (afterHolmes’s vote against the government in Northern Securities v. United States, 193 U.S. 197(1904), TR exclaimed ‘‘I could carve out of a banana a Judge with more backbone thanthat!’’), Wilson’s appointment of McReynolds, Truman’s appointment of Clark and Burton(in Truman’s words, ‘‘Whenever you put a man on the Supreme Court he ceases to be yourfriend’’), Eisenhower’s appointment of Warren and Brennan (Eisenhower said that his ‘‘twobiggest mistakes . . . are both on the Supreme Court’’), Nixon’s appointment of Blackmun,and, to some degree, Powell, and, most recently, Bush’s appointment of Souter. See id. at 7,69–70; Henry Abraham, Can Presidents Really Pack the Supreme Court? in An EssentialSafeguard: Essays on the United States Supreme Court and Its Justices 45–48 (D. Grier Ste-phenson, Jr., ed. 1991). While in many respects even these judges supported the views oftheir appointing president (see Tribe, supra note 3, at 50–76), it is worth noting that this


the president may place the highest value on near-term influential prece-dents.

It is difficult to estimate an appropriate discount rate, but one can be ap-proximated by examining the depreciation of legal capital. Once a prece-dent exists, its value to the legal community declines over time, as issuesbefore the Court change and new precedents accumulate.35 Although the de-preciation rate is not technically the same as the president’s discount rate,it responds to many of the same factors and so provides a reasonable ap-proximation. Data on the distribution of citations by the number of yearselapsed between the citation and the opinion cited, for all 204,186 citationsfound in U.S. Reports, yield a depreciation rate r of 6.5 percent.36 This valuewill be used to discount the value of legal precedent to compute each jus-tice’s Appointment Value.

4. Putting It All Together

Returning to the NPV formulation of equation (1), we now take the finalstep in applying the model to Supreme Court justices. The ‘‘cash flow’’ ina given time period will be the output of a justice, the legal influence gener-ated, in that period: the sum of the inflation-adjusted citation counts for allopinions he wrote in a given term. Based on the number of years since hisappointment, the annual sums are discounted back to ‘‘present value’’ atthe time of appointment and combined to determine the justice’s Appoint-ment Value. In essence,

AV 5t

j

ct, j

(1 1 r)t, (4)

where ct, j is the adjusted citation count for the jth opinion authored by thejustice in year t and r is the discount rate. Historical Value is computed ina similar fashion, but setting r equal to zero to give a justice’s precedentsequal weight, whenever generated.

abbreviated list of philosophically diverging justices includes 10 percent of the justices inthis study.

35 See Landes & Posner, supra note 13, at 267–70.36 The 95 percent confidence range for this coefficient is 60.6 percent. The generalized

nonlinear regression used the Levenberg-Marquardt method, and the overall residual standarddeviation was 245. This fit assumes a single depreciation rate. A much better fit to the datais possible using two depreciation rates: over about the first 28 years, the rate is 6.6 percent,but then it rather abruptly decreases to 3.3 percent. Because an opinion is discounted basedon the number of years since the justice’s appointment, not based on when the citation oc-curs, this change in the discount rate does not enter the analysis—only 13 justices had careerslonger than 28 years, with Justice Douglas, at 36 years, the longest.

Figu

re5.

—C

itatio

npr

ice

inde

x


II. Evaluating the Past Appointees

This section applies the NPV model to the Supreme Court justices andtheir opinions. It computes, using subsequent citations by the SupremeCourt, the Appointment and Historical Values for every opinion of theCourt and for each retired justice. It tests the reliability of these values bycomparing them to results of performing the same analysis using citationsin subsequent circuit court, rather than Supreme Court, opinions (an ‘‘inter-nal consistency’’ check) and to a well-known survey of Supreme Court‘‘greatness’’ (an ‘‘external consistency’’ check).

A. Appointment Value, Historical Value, and Annual Contribution toValue for the Justices of the Supreme Court

This subsection presents the results of applying equations (2), (3), and(4), using for each opinion’s c0

t, j the number of citations to that opinionfound in cases published through volume 503 (1992) of U.S. Reports.37

Tables 1 and 2 show the Appointment Values and Historical Values, re-spectively, for retired38 Supreme Court justices.39 Although there are cer-

37 The underlying data, in the form of a Microsoft Access database, cover 24,491 SupremeCourt opinions: every signed majority opinion in cases from 1793 through 1991. The1,254,985 citations in the database include 204,186 in U.S. Reports (through 1992); 55,630in Federal Reporter 3d (through 1995); 904,604 in Federal Reporter 2d; 159,536 in FederalReporter; and 3,671 in Federal Cases. The 1,156 per curiam opinions are excluded, exceptfor Table 5 infra.

In constructing this database I relied heavily on Shepard’s data, and I thank Sharon Smithof Shepard’s McGraw-Hill for solving my Westlaw bandwidth problems by providing thedatabase on several CD-ROMs. In processing the raw Shepard’s data, I cross-checked andrepaired many errors and inconsistencies. For example (and you can try this yourself onWestlaw), Shepardizing Wickard v. Filburn, 317 U.S. 111 (1942), shows an obviously incor-rect citation listed at 137 U.S. 370. (This occurred about 50 times in the U.S. Reports data.)Much more frequently, Shepard’s incorrectly listed dates and authors of Supreme Court opin-ions, so I reconstructed these using Westlaw and LEXIS.

The data exclude all citations with Shepard’s codes indicating citation in a dissenting opin-ion or one that overrules, reverses, vacates, or supersedes the cited case. See note 16 supraand accompanying text. This resulted in the exclusion of 53,284 citations overall, with 99.7percent of these being citations in dissenting opinions. Shepard’s unfortunately provides noway to distinguish citations found in concurring opinions from those in majority opinions.

38 Because the currently sitting justices continue to author opinions, and because citationsto their existing opinions continue to accumulate, they cannot fairly be compared to thosewho have already completed their careers, so they are considered separately. Observant read-ers will also notice the absence of Justice Blair, who is excluded because he never authoredan opinion of the Court. In Blair’s defense, he did contribute an occasional seriatim opinion.See, for example, Chisolm v. Georgia, 2 U.S. (2 Dall) 419, 450–53 (1793). Shepard’s doesnot identify citations to such opinions.

39 Charles Evans Hughes resigned from the court to challenge Woodrow Wilson for thepresidency. See Burnett Anderson, Charles Evans Hughes, in The Supreme Court Justices:Illustrated Biographies, 1789–1993, at 308–9 (Clare Cushman ed. 1993). Because of the re-sulting 14-year gap in his career on the Court, this study separates out Hughes’s tenure aschief justice. The careers of other promoted associate justices are not considered separately.

TABLE 1

Appointment Value for Retired Justices

Name AV Name AV

1. Fuller 10,940 51. Chase, S. P. 2,3802. Waite 10,660 52. Woods 2,3303. Holmes 9,740 53. Clark 2,2504. Harlan I 8,870 54. Harlan II 2,2405. Gray 8,520 55. Clarke 2,2306. Brewer 7,870 56. Davis 2,1607. Brandeis 7,590 57. Lamar, L. 2,1608. White, E. 7,280 58. Lurton 2,0709. Miller 7,230 59. Sanford 2,030

10. Field 7,160 60. Grier 2,02011. Van Devanter 7,080 61. McLean 2,02012. Day 7,070 62. Murphy 1,94013. Brown 7,050 63. Curtis 1,92014. Hughes, C. J. 7,040 64. Hunt 1,91015. Stone 6,990 65. Moody 1,66016. Bradley 6,680 66. Vinson 1,64017. Pitney 6,170 67. Rutledge, W. 1,55018. Marshall, J. 6,160 68. Campbell 1,42019. Brennan 5,800 69. Catron 1,40020. Douglas 5,740 70. Wayne 1,37021. Sutherland 5,700 71. Burton 1,19022. Taft 5,620 72. Baldwin 1,11023. Matthews 5,560 73. Daniel 1,09024. Peckham 5,540 74. Thompson 1,01025. Powell 5,400 75. Jackson, H. 96026. Black 5,200 76. Johnson, W. 85027. Butler 5,100 77. Minton 77028. McReynolds 4,930 78. Goldberg 75029. McKenna 4,840 79. Fortas 61030. White, B. 4,810 80. Whittaker 59031. Blatchford 4,760 81. Woodbury 56032. Swayne 4,560 82. Barbour 52033. Clifford 4,480 83. Washington 38034. Frankfurter 4,480 84. Chase, S. 37035. Hughes, J. 4,460 85. McKinley 35036. Strong 4,370 86. Ellsworth 33037. Stewart 4,030 87. Trimble 28038. Burger 3,880 88. Livingston 22039. Shiras 3,830 89. Byrnes 17040. Story 3,740 90. Todd 13041. Roberts 3,660 91. Paterson 10042. Blackmun 3,560 92. Iredell 7043. Warren 3,430 93. Duvall 6044. Reed 3,410 94. Jay 3045. Taney 3,370 95. Rutledge, J. 3046. Marshall, T. 3,050 96. Cushing 3047. Nelson 2,860 97. Wilson 2048. Cardozo 2,790 98. Johnson, T. 2049. Lamar, J. 2,640 99. Moore 450. Jackson, R. 2,550

TABLE 2

Historical Value for Retired Justices

Name HV Name HV

1. Harlan I 21,650 51. Grier 3,7802. Holmes 19,000 52. Davis 3,6503. Field 18,860 53. Harlan II 3,5304. Miller 17,990 54. Wayne 3,3105. Fuller 17,240 55. Cardozo 3,2706. Marshall, J. 16,530 56. Chase, S. P. 3,2207. Waite 16,420 57. Lamar, J. 2,9908. White, E. 15,330 58. Catron 2,9809. Gray 14,680 59. Clarke 2,730

10. Brennan 13,830 60. Woods 2,72011. Bradley 13,230 61. Murphy 2,46012. Brandeis 13,020 62. Sanford 2,43013. Stone 12,480 63. Lamar, L. 2,42014. Brewer 12,460 64. Curtis 2,34015. White, B. 12,100 65. Lurton 2,33016. Day 11,880 66. Hunt 2,22017. Douglas 11,260 67. Johnson, W. 2,11018. Black 11,130 68. Vinson 2,05019. Van Devanter 10,830 69. Campbell 1,92020. Brown 10,450 70. Daniel 1,91021. McKenna 9,790 71. Moody 1,88022. Hughes, C. J. 9,240 72. Rutledge, W. 1,86023. Clifford 8,650 73. Burton 1,85024. Sutherland 8,590 74. Thompson 1,84025. Story 8,520 75. Baldwin 1,60026. McReynolds 8,450 76. Washington 1,22027. Stewart 8,380 77. Jackson, H. 98028. Powell 8,310 78. Minton 93029. Swayne 8,290 79. Goldberg 84030. Pitney 8,170 80. Woodbury 71031. Peckham 7,810 81. Fortas 71032. Taney 7,660 82. Whittaker 66033. Frankfurter 7,410 83. Barbour 62034. Butler 7,100 84. McKinley 46035. Matthews 7,090 85. Chase, S. 42036. Taft 7,030 86. Ellsworth 38037. Blatchford 6,770 87. Livingston 37038. Blackmun 6,750 88. Trimble 30039. Nelson 6,590 89. Todd 26040. Marshall, T. 6,390 90. Byrnes 17041. Burger 6,380 91. Paterson 14042. Strong 6,260 92. Iredell 9043. Roberts 5,590 93. Duvall 9044. Warren 5,580 94. Cushing 6045. Hughes, J. 5,530 95. Jay 4046. Reed 5,500 96. Rutledge, J. 4047. Shiras 5,230 97. Wilson 3048. McLean 4,400 98. Johnson, T. 2049. Jackson, R. 3,810 99. Moore 450. Clark 3,810


tainly some surprises,40 many of those listed we recognize intuitively asamong the most influential justices. As suspected, the chief justices claima disproportionate share of the top finishers—one-third of the top 15 jus-tices in both Appointment Value and Historical Value.41 Applying a dis-count rate has caused some divergence between AV and HV (for example,John Marshall moves up from rank 18 to rank 6 without discounting),though there remains a strong correlation of 0.9 between the two values.

Another useful benchmark can be defined: the annual contribution tovalue, computed by dividing the Appointment or Historical Value by thenumber of years each individual served on the Court. These figures providea different perspective: which justices were most influential, not over thecourse of an entire career, but in their average year? Tables 3 and 4 pro-vide the annual contribution to Appointment Value (cAV) and HistoricalValue (cHV).42

40 For example, Chief Justice Waite, ranking just after John Marshall in HV, may not bepopularly known as one of the great American judges. Few knew what to expect of Waitewhen he was appointed by President Grant—he had no judicial experience and had nevereven argued a case before the Court. But Waite ‘‘took the reins of leadership . . . proved tobe a good manager . . . won over his colleagues . . . [and] was remarkably industrious, settingan example by taking on the largest share of opinions to write himself and by assuming thework of his colleagues when they were ill.’’ David T. Pride, Morrison R. Waite, in Cushman,ed., supra note 38, at 213. Waite authored an average of 62 opinions every year for 14years—compare that to the 75 opinions produced by the entire Court during the 1995–96term. His extremely high level of productivity apparently translated into substantial influenceover subsequent cases.

41 See text accompanying note 23 supra for further discussion of the influence of the chiefjustice.

42 The current justices are disadvantaged in any comparison with their peers who havecompleted an entire career, especially when considering cumulative values. Further, becausethe available citation data for this study only include citations in opinions published in U.S.Reports through 1992, there is a ‘‘truncation effect’’ on citation counts for more recent opin-ions, for which the CPI adjustment probably only partially compensates; see note 30 supraand accompanying text. The results for the current members of the Court, excluding JusticesSouter, Thomas, Ginsburg, and Breyer, who have not been on the Court long enough to accu-mulate significant citation data, are as follows:

Name AV HV cAV cHV

1. Rehnquist 4,980 8,660 262 4562. Stevens 3,490 5,710 218 3573. O’Connor 3,860 5,380 386 5384. Scalia 1,230 1,450 245 2895. Kennedy 1,210 1,300 404 434

As one might expect, the cumulative AV and HV rank in approximately the order in whichthe justices were appointed: those with longer tenure on the Court have had more opportunityto influence the law. Without further analysis it is difficult to assess the distortion created bythe truncation effect. But even without the additional citations and opinions that will continueto accumulate, Chief Justice Rehnquist has an AV, HV, and cHV high enough to rank him


Table 5 lists the most influential opinion authored by each justice, theopinion’s Historical Value (oHV), and the percentage of the author’s His-torical Value represented by that opinion (%HV). The most influential percuriam opinion is also listed. Most of these opinions are quite familiar, pre-dominantly constitutional cases, but there are also several involving anti-trust, labor, regulated industries, and administrative law.43 An averageweighted by the author’s HV demonstrates that each justice’s ‘‘most influ-ential opinion’’ represents on average only 4.7 percent of that total. Thisindicates that HV is relatively insensitive to the effect of a single influentialopinion, except for justices with extremely low HV, where a single opinionsometimes constitutes a significant portion of that judge’s career-long con-tribution to the law.44

Table 6 presents the 25 most influential Supreme Court opinions, basedon each opinion’s Historical Value (oHV). It also indicates the major sub-ject matter addressed by each case.

B. Internal Consistency Check: Comparing Circuit Court Citations toSupreme Court Citations

To test the internal consistency of the model, this section repeats the Ap-pointment Value and Historical Value computations, using circuit courtrather than Supreme Court citations. Assuming that an opinion will influ-ence Supreme Court and circuit court cases to a similar degree, the sameanalysis applied to the two independent sets of data ought to yield highly

among the most influential of justices. Also noteworthy, Justices Kennedy and O’Connorhave high-ranking cAV and cHV scores.

43 Justice Peckham’s Ex parte Young has been more than three times as influential as hisLochner v. New York—a favorite of law professors, but cited much less frequently by theCourt (Lochner contributed 128 to Peckham’s HV—1.6 percent of his total).

44 The most significant opinions for the current justices listed in note 42 supra are as fol-lows:

Name Opinion oHV % HV

Rehnquist Edelman v Jordan, 415 U.S. 651 (1974) 130 2Stevens Chevron USA v NRDC, 467 U.S. 837 (1984) 270 5O’Connor Teague v Lane, 489 U.S. 288 (1989) 400 7Scalia Hitchcock v Dugger, 481 U.S. 393 (1987) 110 7Kennedy Davis v Michigan, 489 U.S. 803 (1989) 130 10

As one might expect, as tenure increases, the share of total HV claimed by the most influen-tial opinion tends to decline.

TABLE 3

Annual Contribution to Appointment Value

Name cAV Name cAV

1. Waite 761 51. McReynolds 1832. Hughes, J. 743 52. Marshall, J. 1813. Matthews 694 53. McKenna 1794. Hughes, C. J. 640 54. Reed 1795. Taft 624 55. Campbell 1776. Pitney 617 56. Stewart 1757. Lamar, J. 528 57. Brennan 1718. Lurton 518 58. Byrnes 1659. Fuller 497 59. Douglas 160

10. Jackson, H. 479 60. White, B. 15511. Brown 470 61. Fortas 15312. Cardozo 465 62. Black 15313. Strong 437 63. Blackmun 14814. Blatchford 432 64. Davis 14415. Lamar, L. 431 65. Harlan II 14016. Peckham 427 66. Trimble 13917. Gray 426 67. Marshall, T. 12718. Moody 416 68. Clark 12519. Brewer 393 69. Taney 12020. Woods 389 70. Whittaker 11821. Day 372 71. Story 11322. Clarke 371 72. Minton 10923. Powell 360 73. Nelson 10624. Sutherland 356 74. Barbour 10525. Shiras 348 75. Woodbury 9326. Stone 333 76. Burton 9227. Brandeis 330 77. Grier 8428. Holmes 325 78. Ellsworth 8229. Curtis 320 79. Baldwin 7930. Butler 319 80. McLean 6531. Bradley 304 81. Daniel 6132. Sanford 290 82. Thompson 5133. Van Devanter 272 83. Catron 5034. White, E. 270 84. Wayne 4335. Chase, S. P. 265 85. Johnson, W. 2836. Harlan I 261 86. McKinley 2537. Rutledge, W. 259 87. Chase, S. 2438. Miller 258 88. Johnson, T. 1639. Goldberg 250 89. Livingston 1440. Roberts 244 90. Washington 1341. Swayne 240 91. Iredell 842. Vinson 234 92. Paterson 843. Burger 228 93. Todd 744. Murphy 215 94. Rutledge, J. 645. Warren 214 95. Jay 546. Hunt 212 96. Wilson 247. Field 210 97. Duvall 248. Jackson, R. 196 98. Cushing 149. Clifford 195 99. Moore 150. Frankfurter 195

TABLE 4

Annual Contribution to Historical Value

Name cHV Name cHV

1. Waite 1,173 51. Douglas 3132. Hughes, J. 921 52. McReynolds 3133. Matthews 887 53. Rutledge, W. 3094. Hughes, C. J. 840 54. Jackson, R. 2935. Pitney 817 55. Vinson 2926. Fuller 784 56. Reed 2907. Taft 781 57. Blackmun 2818. Gray 734 58. Goldberg 2799. Brown 697 59. Murphy 274

10. Miller 642 60. Taney 27311. Harlan I 637 61. Marshall, T. 26612. Holmes 633 62. Story 25813. Strong 626 63. Hunt 24714. Day 625 64. Nelson 24415. Brewer 623 65. Davis 24316. Blatchford 616 66. Campbell 24017. Bradley 601 67. Harlan II 22018. Peckham 601 68. Clark 21119. Lamar, J. 599 69. Fortas 17820. Stone 594 70. Byrnes 17421. Lurton 582 71. Grier 15722. White, E. 568 72. Trimble 15123. Brandeis 566 73. Burton 14224. Field 555 74. McLean 14225. Powell 554 75. Minton 13326. Cardozo 545 76. Whittaker 13227. Sutherland 537 77. Barbour 12328. Jackson, H. 489 78. Woodbury 11829. Marshall, J. 486 79. Baldwin 11430. Lamar, L. 484 80. Daniel 10631. Shiras 475 81. Catron 10632. Moody 471 82. Wayne 10333. Clarke 455 83. Ellsworth 9534. Woods 453 84. Thompson 9235. Butler 444 85. Johnson, W. 7036. Swayne 437 86. Washington 4137. Van Devanter 417 87. McKinley 3338. Brennan 407 88. Chase, S. 2839. White, B. 390 89. Livingston 2340. Curtis 389 90. Johnson, T. 1841. Clifford 376 91. Todd 1342. Burger 375 92. Paterson 1143. Roberts 373 93. Iredell 1044. Stewart 364 94. Rutledge, J. 845. McKenna 363 95. Jay 746. Chase, S. P. 358 96. Duvall 447. Warren 349 97. Wilson 448. Sanford 347 98. Cushing 349. Black 327 99. Moore 150. Frankfurter 322

356

TABLE 5

Most Influential Opinion of Each Justice

Name Opinion oHV %HV

Baldwin United States v. Arredondo, 31 U.S. 691 (1832) 260 16Barbour Wilcox v. Jackson, 38 U.S. 498 (1839) 210 34Black Johnson v. Zerbst, 304 U.S. 458 (1938) 280 2Blackmun Mills v. Maryland, 486 U.S. 367 (1988) 150 2Blatchford Counselman v. Hitchcock, 142 U.S. 547 (1892) 280 4Bradley Boyd v. United States, 116 U.S. 616 (1886) 680 5Brandeis Erie R.R. Co. v. Tompkins, 304 U.S. 64 (1938) 320 2Brennan Roth v. United States, 354 U.S. 476 (1957) 280 2Brewer Reagan v. Farmers’ Loan, 154 U.S. 362 (1894) 250 2Brown Union Refrig. Tr. Co. v. Kentucky, 199 U.S. 194 (1905) 370 4Burger Lockett v. Ohio, 438 U.S. 586 (1978) 170 3Burton Joint Anti-Fascist R. Comm. v. McGrath, 341 U.S. 123 (1951) 110 6Butler Agnello v. United States, 269 U.S. 20 (1925) 220 3Byrnes Ward v. Texas, 316 U.S. 547 (1942) 30 15Campbell Taylor v. Carryl, 61 U.S. 583 (1858) 190 10Cardozo Palko v. Connecticut, 302 U.S. 319 (1937) 220 7Catron D’Arcy v. Ketchum, 52 U.S. 165 (1851) 120 4Chase, S. Calder v. Bull, 3 U.S. 386 (1799) 250 60Chase, S. P. Veazie Bank v. Fenno, 75 U.S. 533 (1869) 150 5Clark Mapp v. Ohio, 367 U.S. 643 (1961) 220 6Clarke Gouled v. United States, 255 U.S. 298 (1921) 200 7Clifford Suydam v. Williamson, 61 U.S. 427 (1858) 180 2Curtis Cooley v. Board of Wardens, 53 U.S. 299 (1852) 480 21Cushing Marine Insurance Co. v. Young, 9 U.S. 187 (1809) 30 46Daniel Cary v. Curtis, 44 U.S. 236 (1845) 132 7Davis Leavenworth R. Co. v. United States, 92 U.S. 733 (1876) 190 5Day Weeks v. United States, 232 U.S. 383 (1914) 430 4Douglas Monroe v. Pape, 365 U.S. 167 (1961) 220 2Duvall Walton v. United States, 22 U.S. 651 (1824) 30 35Ellsworth Turner v. Bank of North America, 4 U.S. 8 (1800) 120 30Field Pennoyer v. Neff, 95 U.S. 714 (1878) 430 2Fortas Tinker v. Des Moines Sch. Dist., 393 U.S. 503 (1969) 90 12Frankfurter San Diego Bldg. Tr. Coun. v. Garmon, 359 U.S. 236 (1959) 170 2Fuller Pollock v. Farmers’ Loan & Tr., 157 U.S. 429 (1895) 230 1Goldberg Cox v. Louisiana, 379 U.S. 536 (1965) 120 14Gray Shively v. Bowlby, 152 U.S. 1 (1894) 310 2Grier Peck v. Jenness, 48 U.S. 612 (1849) 140 4Harlan I Mugler v. Kansas, 123 U.S. 623 (1887) 370 2Harlan II NAACP v. Alabama, 357 U.S. 449 (1958) 300 8Holmes Missouri v. Illinois, 200 U.S. 496 (1906) 500 3Hughes, C. J. Stromberg v. California, 283 U.S. 359 (1931) 300 3Hughes, J. Minnesota Rate Cases, 230 U.S. 352 (1913) 450 8Hunt United States v. Railroad Co., 84 U.S. 322 (1873) 120 5Iredell Chisholm v. Georgia, 2 U.S. 419 (1793) 80 88Jackson, H. Mexican Cent. R. Co. v. Pinkney, 149 U.S. 194 (1893) 80 8Jackson, R. Johnson v. United States, 333 U.S. 10 (1948) 190 5Jay Glass v. The Sloop Betsey, 3 U.S. 6 (1794) 30 67Johnson, T. Georgia v. Braislford, 2 U.S. 402 (1793) 20 100Johnson, W. United States v. Hudson & Goodwin, 11 U.S. 32 (1812) 190 9Lamar, J. Gompers v. Buck’s Stove & Range, 221 U.S. 418 (1911) 180 6

357

TABLE 5 (Continued )

Name Opinion oHV %HV

Lamar, L. Kidd v. Pearson, 128 U.S. 1 (1888) 200 8Livingston Hopkins v. Lee, 19 U.S. 109 (1821) 60 17Lurton United States v. Chandler-Dunbar, 229 U.S. 53 (1913) 190 8Marshall, J. McCulloch v. Maryland, 17 U.S. 316 (1819) 1,090 7Marshall, T. Dunn v. Blumstein, 405 U.S. 330 (1972) 110 2Matthews Yick Wo v. Hopkins, 118 U.S. 356 (1886) 480 7McKenna Southern Pac. Term. Co. v. ICC, 219 U.S. 498 (1911) 200 2McKinley Pollard v. Hagan, 44 U.S. 212 (1845) 250 53McLean Smith v. Turner, 48 U.S. 283 (1849) 190 4McReynolds Pierce v. Society of Sisters, 268 U.S. 510 (1925) 320 4Miller Slaughter-House Cases, 83 U.S. 36 (1873) 370 2Minton United States v. Rabinowitz, 339 U.S. 56 (1950) 110 11Moody Twining v. New Jersey, 211 U.S. 78 (1908) 280 15Moore Bas v. Tingy, 4 U.S. 37 (1801) 4 100Murphy Thornhill v. Alabama, 310 U.S. 88 (1940) 220 9Nelson Freeman v. Howe, 65 U.S. 450 (1861) 240 4Paterson Stuart v. Laird, 5 U.S. 299 (1803) 50 35Peckham Ex parte Young, 209 U.S. 123 (1908) 450 6Per Curiam Mooney v. Holohan, 294 U.S. 103 (1935) 200 17Pitney Frank v. Mangum, 237 U.S. 309 (1915) 200 2Powell Eddings v. Oklahoma, 455 U.S. 104 (1982) 180 2Reed Brown v. Allen, 344 U.S. 443 (1953) 170 3Roberts Cantwell v. Connecticut, 310 U.S. 296 (1940) 350 6Rutledge, J. United States v. Peters, 3 U.S. 121 (1795) 40 100Rutledge, W. Thomas v. Collins, 323 U.S. 516 (1945) 170 9Sanford Whitney v. California, 274 U.S. 357 (1927) 270 11Shiras Eustis v. Bolles, 150 U.S. 361 (1893) 160 3Stewart Katz v. United States, 389 U.S. 347 (1967) 160 2Stone Virginian R. Co. v. System Fed., 300 U.S. 515 (1937) 220 2Story Martin v. Hunter’s Lessee, 14 U.S. 304 (1816) 580 7Strong Ex parte Virginia, 100 U.S. 339 (1880) 330 5Sutherland Powell v. Alabama, 287 U.S. 45 (1932) 320 4Swayne Gilman v. Philadelphia, 70 U.S. 713 (1866) 220 3Taft Carroll v. United States, 267 U.S. 132 (1925) 340 5Taney The Propeller Genessee Chief, 53 U.S. 443 (1852) 270 4Thompson Kendall v. United States ex rel Stokes, 37 U.S. 524 (1838) 170 9Todd Robinson v. Campbell, 16 U.S. 212 (1818) 120 48Trimble Edwards’ Lessee v. Darby, 25 U.S. 206 (1827) 60 21Van Devanter Second Employers’ Liability Cases, 223 U.S. 1 (1912) 350 3Vinson United States v. UMW, 330 U.S. 258 (1947) 130 6Waite Munn v. Illinois, 94 U.S. 113 (1877) 320 2Warren Brown v. Board of Education, 347 U.S. 483 (1954) 300 5Washington Ogden v. Saunders, 25 U.S. 213 (1827) 150 13Wayne Jefferson Branch Bank v. Skelly, 66 U.S. 436 (1862) 230 7White, B. Will v. Michigan State Police, 491 U.S. 58 (1989) 180 2White, E. Standard Oil Co. v. United States, 221 U.S. 1 (1911) 380 3Whittaker Staub v. City Of Baxley, 355 U.S. 313 (1958) 90 14Wilson United States v. Hamilton, 3 U.S. 17 (1795) 30 75Woodbury Reeside v. Walker, 52 U.S. 272 (1851) 60 8Woods United States v. Harris, 106 U.S. 629 (1883) 90 3

Note.—oHV 5 the opinion’s Historical Value; %HV 5 the percentage of the author’s HistoricalValue represented by the opinion.

358

TABLE 6

The Twenty-Five Most Influential Opinions

Opinion oHV Subject Matter

1. McCulloch v. Maryland, 17 U.S. 316 (1819)(J. Marshall) 1,090 Necessary and proper clause

2. Gibbons v. Ogden, 22 U.S. 1 (1824)(J. Marshall) 720 Federal commerce regulation

3. Boyd v. United States, 116 U.S. 616 (1886)(Bradley) 680 Search and seizure

4. Martin v. Hunter’s Lessee, 14 U.S. 304(1816) (Story) 580 Judicial review

5. Osborn v. Bank of the United States, 22 U.S.738 (1924) (J. Marshall) 530 Federal question jurisdiction

6. Missouri v. Illinois, 200 U.S. 496 (1906)(Holmes) 500 Original jurisdiction

7. Cooley v. Board of Wardens, 53 U.S. 299(1852) (Curtis) 480 State commerce regulation

8. Brown v. Maryland, 25 U.S. 419 (1827)(J. Marshall) 480 State commerce regulation

9. Yick Wo v. Hopkins, 118 U.S. 356 (1886)(Matthews) 480 Fourteenth Amendment

10. Ex parte Young, 209 U.S. 123 (1908)(Peckham) 450 Sovereign immunity

11. Minnesota Rate Cases, 230 U.S. 352 (1913)(Hughes) 450 Federal commerce regulation

12. Pennoyer v. Neff, 95 U.S. 714 (1878) (Field) 430 Personal jurisdiction13. Weeks v. United States, 232 U.S. 383 (1914)

(Day) 430 Search and seizure14. Marbury v. Madison, 5 U.S. 137 (1803)

(J. Marshall) 390 Judicial review15. Standard Oil Co. v. United States, 221 U.S.

1 (1911) (E. White) 380 Sherman Act16. Union Refrig. Trans. Co. v. Kentucky, 199

U.S. 194 (1905) (Brown) 370 State commerce regulation17. Slaughter-House Cases, 83 U.S. 36 (1873)

(Miller) 370 Fourteenth Amendment18. Cohens v. Virginia, 19 U.S. 264 (1821)

(J. Marshall) 370 Judicial review19. Mugler v. Kansas, 123 U.S. 623 (1887)

(Harlan I) 470 Fourteenth Amendment20. Swift & Co. v. United States, 196 U.S. 375

(1905) (Holmes) 360 Federal commerce regulation21. Fletcher v. Peck, 10 U.S. 87 (1810)

(J. Marshall) 350 Contracts clause22. Cantwell v. Connecticut, 310 U.S. 296

(1940) (Roberts) 350 Free exercise23. Second Employers’ Liability Cases, 223 U.S.

1 (1912) (Van Devanter) 340 Federal commerce regulation24. Carroll v. United States, 267 U.S. 132 (1925)

(Taft) 340 Search and seizure25. Paul v. Virginia, 75 U.S. 168 (1869)

(Field) 340 State commerce regulation

Note.—oHV 5 the opinion’s Historical Value.359


correlated results only if the methodology is truly measuring influence andthe citation data are properly inflation adjusted.45

Each justice’s AV and HV are recomputed by setting the raw number ofcitations c0

t, j in equations (2) and (3) to the number of times his jth opinionof year t was cited in circuit court opinions published in the various editionsof the Federal Reporter. Comparing these results to the AV and HV pre-sented earlier demonstrates a very strong relationship. A linear regressionof the two computations shows that the AV from circuit court citation datais 7.1 times the AV from Supreme Court citation data, with an R2 of 0.93;similarly, the circuit court HV is 7.0 times the Supreme Court HV, with anR2 of 0.95.

Problems in defining the CPI would likely show up as a poor correlationbetween these two measurements. For example, because variations in cita-tion rates in the courts have not always corresponded closely,46 one wouldexpect a low R2 if there were a problem in adjusting for changing citationpatterns. Instead, the strongly correlated results from similar analyses on twoindependent sets of citation data indicate consistency in the methodology.

C. External Consistency Check: Comparing Historical Value to OtherMeasurements of Influence

In 1978, Albert P. Blaustein and Roy M. Mersky published a study of 65law school deans and professors of law, history, and political science, whowere commissioned to rank the first 96 justices into categories of ‘‘great,’’‘‘near great,’’ ‘‘average,’’ ‘‘below average,’’ and ‘‘failure.’’47 Compared tothe current study, Blaustein and Mersky’s survey used far more subjectivecriteria for evaluating justices and defined Greatness more broadly than ajustice’s influence over the development of law.48 Thus, many justices who

45 See note 31 supra and accompanying text.46 See Figures 3 and 4.47 See Albert P. Blaustein and Roy M. Mersky, The First One Hundred Justices: Statistical

Studies on the Supreme Court of the United States 32–51 (1978). This survey rated 12 jus-tices ‘‘great’’ (J. Marshall, Story, Taney, Harlan I, Holmes, Hughes, Brandeis, Stone, Car-dozo, Black, Frankfurter, and Warren); 15 ‘‘near great’’ (W. Johnson, Curtis, Miller, Field,Bradley, Waite, E. White, Taft, Sutherland, Douglas, R. Jackson, W. Rutledge, Harlan II,Brennan, and Fortas); 55 ‘‘average’’; 6 ‘‘below average’’ (T. Johnson, Moore, Trimble, Bar-bour, Woods, and H. Jackson); and 8 ‘‘failure’’ (Van Devanter, McReynolds, Butler, Byrnes,Burton, Vinson, Minton, and Whittaker).

48 At least one author has suggested that the most proper way for rating and comparingjustices is by assessing their relative influence over legal developments. Criticizing a morerecent but less thorough study than Blaustein and Mersky, Schwartz noted that, to him, ‘‘Su-preme Court greatness is virtually synonymous with influence on the law.’’ BernardSchwartz, Supreme Court Superstars: The Ten Greatest Justices, 31 Tulsa L. J. 93, 157(1995) (discussing Robert C. Bradley, Who Are the Great Justices and What Criteria Did


exerted little or no influence—including, for example, Justice Duvall, whoDavid Currie hailed as the ‘‘least significant Justice’’49—are ranked in the‘‘average’’ category.

Nevertheless, comparing the results of these two studies demonstrates astrong positive relationship between Historical Value and Greatness. Table7 shows the results of a multiple regression with dummy variables for eachof the Greatness categories, leaving the ‘‘average’’ category out of the re-gression as a baseline. The coefficients show that HV declines with theGreatness category, from ‘‘great’’ through ‘‘below average,’’ with strongstatistical significance. Thus, Historical Value as defined and calculated bythis study is indeed related to a justice’s perceived reputation.

Interestingly, the ‘‘failures’’ are not statistically distinguishable fromthose rated ‘‘average.’’ The survey ranked the relatively unknown justicesas ‘‘average,’’ with ‘‘failure’’ apparently reserved for justices deemed af-firmatively bad. We should not therefore be surprised to see justices in the‘‘failure’’ category scoring higher than those considered ‘‘below average.’’For example, chief among the ‘‘failures’’ is Justice McReynolds, whosenegative reputation derives far more from his blatant misogyny and anti-

They Meet? in Great Justices of the U.S. Supreme Court 1 (William D. Pederson and NormanW. Provizer eds. 1993)).

Although Bradley’s study includes a very useful bibliography of surveys of judicial great-ness (id. at 24–31), he includes much less information than Blaustein and Mersky, and hisresults appear questionable in several respects. In the spirit of David Letterman, Bradley re-quested only that each participant identify the top 10 justices, not that they rate each justice.His samples were drawn from scholars (in political science, but apparently not law), attorneys(but only those listed in the telephone directory), Illinois state and municipal judges (notfederal judges), and students (from two undergraduate courses and one graduate seminar inpolitical science). Id. at 11–13. Blaustein and Mersky, in contrast, supported the credibilityof their sample population by providing the reader with a list of the 65 recognized scholarswho participated in their study. See Blaustein and Mersky, supra note 46, at 117–18.

Bradley’s survey results were listed by participant group and as totals, apparently withoutany weighting. Including the category ‘‘great justices according to students,’’ for example,seems dubious: these students rated Justice Blackmun higher than true greats like Black,Brandeis, Cardozo, Frankfurter, Story, or Taney. Bradley, id. at 14–17. If nothing else, thissurvey proved that students (at least those surveyed) know little about the Court besides thenames of some of its recent members. Id. at 17–18.

49 See David P. Currie, The Most Insignificant Justice, 50 U. Chi. L. Rev. 460 (1983).Justice Duvall also scores poorly here—though I would hesitate to call him the most insig-nificant. His rank is 93 out of 99 in both Tables 1 and 2. Even if we take into account Du-vall’s long and unproductive tenure on the Court (demoting him to rank 97 and 96 in Tables3 and 4, respectively), Duvall was nearly twice as influential as Cushing, whom I might con-sider the most insignificant based on his poor HV after his 20 years on the Court. Especiallygiven the high expectations based on his promising contributions at the Philadelphia conven-tion, I would also consider Justice Wilson a stronger candidate for the most insignificant.Frank Easterbrook might have been too hasty in awarding Justice Todd his insignificanceaward: in the career total values of Tables 1 and 2, Todd does not penetrate the bottom 10percent. See Frank H. Easterbrook, The Most Insignificant Justice: Further Evidence, 50 U.Chi. L. Rev. 481, 494–96 (1983).


TABLE 7

Comparison of Historical Value to Blaustein andMersky’s Greatness Survey

HV

Greatness Category Coefficient t-Ratio

Constant 4,380 6.83

Great 6,470 4.45Near great 4,750 3.46Below average 23,600 21.78Failure 2370 2.21

Note.—R2 5 .28. The constant term reflects justices rated ‘‘av-erage.’’ See Albert P. Blaustein & Ray M. Mersky, The First OneHundred Justices: Statistical Studies on the Supreme Court of theUnited States (1978), and text for category descriptions.

Semitism than from incompetence or, more to the point, failure to generateinfluential precedents.50 Indeed, were justices like McReynolds less influen-tial, they might not have attracted enough notice to be deemed a ‘‘failure’’by the set of twentieth-century scholars surveyed by Blaustein and Mersky.

III. Court Packing: The Presidents and TheirSelection Processes

Throughout history, presidents have to varying degrees sought to packthe Court. Like the Blaustein and Mersky survey, most evaluation of poten-tial nominees has been an intuitive, subjective process. Presidents considera wide range of factors, which include, among other things, the candidate’s‘‘real politics,’’51 perceived legal ability, experience, age, and health. Theirchoices have sometimes been constrained by political circumstances, oftenin the form of a promised confirmation battle over a candidate unacceptableto a powerful Senate faction.

A candidate’s ‘‘real politics’’ is thought to affect the direction of his in-fluence—in what way will he shape the law? Intellectual ability, legal expe-rience, and age are perceived to affect the expected magnitude of his influ-ence. Nominees with prior judicial experience are thought to have less of a‘‘learning curve’’ and to hold more sway with their colleagues.52 Age and

50 See Abraham, supra note 6, at 177–80. Easterbrook, though writing for an altogetherdifferent purpose, appears to concur with this understanding of Blaustein and Mersky’s clas-sifications. See Easterbrook, supra note 49, at 484.

51 See note 6 supra and accompanying text.52 Over the last 50 years, exactly half of the Supreme Court appointees were federal

judges, and just over 60 percent were judges in either state or federal courts. In the last 20years, every appointee has been a judge, with only two from state courts. See also Abraham,supra note 6, at 56–59.


health, as proxies for anticipated longevity, relate to the number of termsduring which a candidate will vote and generate opinions and thereby in-fluence the course of the law. Many have assumed that younger appointeeswill have longer tenure on the Court and, therefore, will tend to have agreater effect on the law.53

The political cost of waging a confirmation fight may sometimes lead apresident to select a ‘‘safe’’ candidate—one with obvious merit, perhapseven a member of the opposing political party—or a ‘‘stealth’’ candidate—one who has taken few public positions on controversial legal or politicalissues. The presence of a strong Senate limits the president’s freedom injudicial selection, perhaps sometimes leading him to settle on someone heviews as a less-than-ideal nominee.

Presidents have relied on these assumptions without substantial informa-tion regarding their merit. This section explores these issues, using asbenchmarks the retired justices’ AV and HV, as developed in Section II.

A. The Presidents

The relative effectiveness of each president’s Supreme Court nomina-tions can be compared by combining the scores of their appointees.54 Table8 ranks the presidents by the total combined Historical Value of their ap-pointees, showing that many of the great presidents also exerted a signifi-cant effect on the law through their Supreme Court appointments.55 Table9 compares the appointees’ average Historical Value and AppointmentValue by president. Not surprisingly, many of the presidents listed here as

53 For example, despite his renowned influence over President Harding’s judicial nomina-tions, Chief Justice Taft could not persuade him to nominate the 64-year-old scholar-judge,Charles M. Hough. Murphy writes that ‘‘the Chief Justice’s highest praise was reserved forJudge Charles Hough of the U.S. Circuit Court of Appeals. Despite several personal pleasby Taft and his use of intermediaries, Harding refused to appoint Hough because he thoughtthe judge too old.’’ Walter F. Murphy, Elements of Judicial Strategy 75 (1964); see alsoAbraham, supra note 6, at 192. More recently, in a strategic effort to leave an enduring judi-cial legacy, Presidents Reagan and Bush sought young candidates, like Justice ClarenceThomas, who could remain on the Court for 30–40 years. See Neil A. Lewis, The 1992 Cam-paign: Selection of Conservative Judges Insures a President’s Legacy, N.Y. Times A13 (July1, 1992).

54 Presidents Bush and Clinton are excluded from these charts as too recent to fairly com-pare their influence. Also excluded are the four presidents who made no Court appointments:Carter, Andrew Johnson, William Henry Harrison, and Taylor. Of these, only Carter serveda full 4-year term.

55 Among the top finishers here, Presidents Lincoln, FDR, Cleveland, T. Roosevelt, Wil-son, and Eisenhower consistently rank high on surveys assessing the importance of presi-dents. These surveys also consistently rank Nixon, Grant, and Harding as the worst threepresidents, though each showed a great deal of success with their Supreme Court appoint-ments. See Abraham, supra note 6, at 415–17 (tabulating results from eight different presi-dential surveys).


TABLE 8

Presidents Ranked by Total HistoricalValue of Justices Appointed

President N Net HV Net AV

1. Lincoln 5 52,000 23,5002. Roosevelt, F. 9 45,400 26,2003. Cleveland 4 42,800 25,9004. Grant 4 38,100 23,6005. Roosevelt, T. 3 32,800 18,5006. Eisenhower 4 32,000 16,1007. Wilson 5 31,200 20,4008. Nixon 5 30,100 17,8009. Taft 4 29,800 22,400

10. Harrison, B. 4 29,100 19,70011. Hayes 2 24,400 11,20012. Arthur 2 21,500 13,30013. Hoover 3 18,100 13,50014. Harding 3 18,100 12,80015. Adams, J. 5 17,800 6,50016. Jackson 3 17,600 8,40017. Kennedy 2 12,900 5,60018. Coolidge 1 12,500 7,00019. McKinley 1 9,800 4,80020. Buchanan 1 8,600 4,50021. Madison 2 8,600 3,80022. Reagan 3 8,100 6,30023. Garfield 1 7,100 5,60024. Johnson, L. 2 7,100 3,70025. Truman 3 6,800 4,70026. Tyler 1 6,600 2,90027. Ford 1 5,700 3,50028. Van Buren 3 5,400 2,80029. Polk 2 4,500 2,60030. Jefferson 3 2,700 1,20031. Fillmore 1 2,300 1,90032. Pierce 1 1,900 1,40033. Monroe 1 1,800 1,00034. Washington 10 1,200 1,00035. Adams, J. Q. 1 300 300

Note.—N 5 number of appointees. AV 5 Appointment Value,and HV 5 Historical Value.

most influential were known for their focus on the Court. Despite FDR’smany nominations during his lengthy term of office, Lincoln was actuallya more effective Court packer.56 Although Washington had the most ap-

56 Although ‘‘Court-packing and the name of President F.D. Roosevelt have become syn-onymous’’ (Abraham, supra note 6, at 67), Lincoln actually pursued a more successful(though less audacious) Court-packing plan, lobbying Congress in 1863 to create a tenth seaton the Court and filling it by nominating Stephen Field. Id. at 120–21.


TABLE 9

Presidents Ranked by Average HistoricalValue of Appointees

President Average HV Average AV

1. Coolidge 12,500 7,0002. Hayes 12,200 5,6003. Roosevelt, T. 10,900 6,2004. Cleveland 10,700 6,5005. Arthur 10,700 6,6006. Lincoln 10,400 4,7007. McKinley 9,800 4,8008. Grant 9,500 5,9009. Buchanan 8,600 4,500

10. Wilson 7,800 5,10011. Nixon 7,500 4,50012. Harrison, B. 7,300 4,90013. Garfield 7,100 5,60014. Tyler 6,600 2,90015. Kennedy 6,500 2,80016. Eisenhower 6,400 3,20017. Taft 6,000 4,50018. Hoover 6,000 4,50019. Taft 6,000 4,30020. Ford 5,900 2,20021. Adams, J. 5,700 3,50022. Roosevelt, F. 5,000 2,90023. Madison 4,300 1,90024. Johnson, L. 3,500 1,70025. Jackson 3,500 1,80026. Reagan 2,700 2,10027. Polk 2,300 1,60028. Truman 2,300 1,90029. Fillmore 2,200 1,30030. Pierce 1,900 1,40031. Van Buren 1,800 90032. Monroe 1,800 1,00033. Jefferson 900 40034. Adams, J. Q. 300 30035. Washington 100 100

Note.—HV 5 Historical Value, and AV 5 AppointmentValue.

pointees of any president, even in combination his were among the leastinfluential of justices. This might be due in part to the low caseload of theCourt in its early years; however, the CPI should largely adjust for suchvariations. More likely, the cases decided by the pre-Marshall Court havesimply had relatively little ongoing effect on the law.

Table 10 compares the justices’ age at appointment and various influencemeasurements based on the political party of the appointing president. In


TABLE 10

Average Age, Duration, and Influence Values by President’s Political Party

Party N Age AV HV cAV cHV

Republican 43 55.8 4,890 8,230 361 536Democrat 35 52.5 2,870 5,060 182 282

Note.—AV 5 Appointment Value, HV 5 Historical Value, cAV 5 the annual contribution to theAppointment Value, and cHV 5 the annual contribution to the Historical Value.

every category, the appointees of Republican presidents have been signifi-cantly more influential than those of Democrats.57 Especially interesting inlight of their stronger influence scores, the 15 Republican presidents gener-ally selected older candidates (on average, 56 years for Republicans vs. 53years for Democrats). In terms of overall effect on law, the greater numberof Republican presidents combines with their greater effectiveness in candi-date selection to yield a total HV of around 354,000 for their nominees,nearly double the total of 177,000 for nominees of Democratic presidents.

B. Age, Prior Experience, and Political Factors Affecting Selection

Some of the most significant assumptions made by presidents involve therelationships between a potential nominee’s age, experience, and likely in-fluence. This section focuses on these important issues of judicial selection.

The 99 Supreme Court justices from Section II are classified by their ageat appointment into five categories: ‘‘45 and under’’ (16 justices); ‘‘46–50’’ (16 justices); ‘‘51–55’’ (27 justices); ‘‘56–60’’ (26 justices); and‘‘over 60’’ (14 justices). The multiple regression uses dummy variables forthese categories, leaving out the ‘‘45 and under’’ category as a baseline.The justices are also organized by predominant area of prior experiencewhen nominated58 into four categories: ‘‘executive and legislative offi-cers’’59 (27 justices); ‘‘judges’’60 (41 justices); ‘‘law professors’’ (4 jus-

57 A selection bias might explain this disparity between the two principal parties. To someextent, a justice may be more likely to cite precedents authored by someone of similar legaland political views. Because more justices have been appointed by presidents of one party,to the extent that such views are correlated with political party, the opinions authored by thatparty’s appointees may be cited disproportionately often.

58 The predominant area for each justice was determined by excluding any positions heldfor less than 4 years and focusing on both the significance of contributions in particular fieldsand the most recent positions held prior to nomination.

59 This category includes federal executive and legislative branch officials as well as stategovernors. These groups have been combined because initial analysis demonstrated no sig-nificant differences between them.

60 Again, because an initial analysis demonstrated no significant differences between stateand federal judges, the groups have been combined.


TABLE 11

Multiple Regression: Appointment Value and Historical Value

AV HV

Independent Variable Coefficient t-Ratio Coefficient t-Ratio

Constant 800 .91 2,260 1.30

46–50 690 .72 710 .3851–55 1,470 1.69 920 .5456–60 310 .36 21,500 2.88.60 1,360 1.27 148 .07

Judge 540 .77 1,410 1.02Law professor 1,790 1.23 3,000 1.05Private attorney 1,920 2.45 4,210 2.73

Senate different party 920 1.18 920 .60First-term president 920 1.38 1,660 1.26After failed nomination 250 .34 840 .58Appointee different party 270 .32 1,090 .65R2 .18 .16

tices); and ‘‘private attorneys’’ (27 justices). Similarly, dummy variablesare used for these categories, leaving out executive and legislative officers.Several political factors are also included: whether the Senate was con-trolled by a different party than the president; whether the appointment oc-curred during the president’s first term of office; whether it occurred aftera previous unsuccessful nomination; and whether the appointee was from adifferent political party than the president.

Table 11 displays the multiple regression results for Appointment Valueand Historical Value; Table 12 displays the results for annual contributionto Appointment Value and to Historical Value.

1. Age

As most have suspected, a younger appointee is more likely to have alonger tenure on the Court. A linear regression shows that the average ca-reer duration d for past appointees is given by

d 5 43 2 0.50 ⋅ a, (5)(6.16) (23.86)

where a is the age at appointment, with an R2 of 0.13 and the t-ratios onthe coefficients as indicated. Thus, the average 60-year-old appointee willremain on the Court for 13 years. This result supports the common wisdomthat selecting a younger judge will increase the candidate’s expected dura-


TABLE 12

Multiple Regression: Annual Contribution to AppointmentValue and Historical Value

cAV cHV

Independent Variable Coefficient t-Ratio Coefficient t-Ratio

Constant 22 .41 54 .71

46–50 100 1.67 133 1.5951–55 145 2.69 172 2.2956–60 132 2.46 126 1.67.60 250 3.78 277 2.99

Judge 8 .20 36 .59Law professor 76 .85 141 1.12Private attorney 83 1.72 172 2.53

Senate different party 85 1.76 106 1.57First-term president 56 1.37 91 1.58After failed nomination 232 2.71 26 2.09Appointee different party 35 .67 40 .54R2 .29 .26

tion—though only by about 6 months per year younger, with the R2 show-ing that there remains much unexplained variation in length of tenure.

Many presidents have nominated young candidates in the expectationthat they would exert greater influence by virtue of their life tenure. How-ever, the extremely low t-ratios on the age coefficients in Table 11 indicatethat age actually has no predictive value for the candidate’s net career in-fluence, even though younger candidates, having longer careers, produceopinions over a longer period of time.

This presents a puzzle: if younger appointees have longer careers, whydo they not exhibit an overall greater degree of influence? Table 12 sug-gests an explanation. The age coefficients in the multiple regression forcAV and cHV are all significant, most at the 0.05 level, with all coefficientsexhibiting positive effects relative to the ‘‘45 and under’’ baseline that gen-erally increase with age. With respect to both cAV and cHV, the appointeesolder than 60 are as a category the most influential. Older appointees areclear winners in average annual influence.

There are several possible explanations for this unexpected result. First,an older appointee brings more experience—the ‘‘wisdom of years’’—tothe judicial role, resulting not only in a shorter learning curve, but also inbetter opinions.61 In the judicial profession, age may bring added respect,

61 See Posner, Aging and Old Age, supra note 14, at 192–99, discussing possible explana-tions for the enhanced productivity of older judges.


more reputation with colleagues and lower courts, and the assignment ofmore important opinions. Some have suggested that there is a limit on thenumber of years for which an individual can remain interested and able tocontribute effectively; a younger appointee might have a higher probabilityof bumping up against such natural limitations.62 Finally, older candidateshave a more substantial track record of past performance, enabling a presi-dent to make a more informed judgment about that person’s likely futureperformance on the Court.

As an odd corollary of these results, a 60-year-old appointee appearsmore influential than a 60-year-old justice who has already been sitting onthe Court for 10 years. Imagine that the two 60-year-olds perform at anidentical level and that the annual performance of the sitting justice hasrisen, not declined, during the last 10 years. Assuming an equivalent levelof performance into the future, the annual contribution values for the newappointee will be higher than that of the sitting justice because the sittingjustice’s 10 years of ‘‘learning the ropes’’ will always tend to depress hislifetime average.63 Also, because there is not a one-to-one relationship be-tween expected career duration and age at appointment,64 the younger can-didate, spending his early years gaining needed experience, may not end upserving long enough to make the best use of that experience. Thus, thehigher average annual influence of older candidates suggests the importanceof acquiring a critical mass of experience before joining the Court.65

2. Prior Experience

Unusually risk averse, Presidents Ford, Reagan, Bush, and Clinton haveunanimously appointed individuals with prior judicial experience. Of thecurrent members of the Court, only Chief Justice Rehnquist had not served

62 For example, former Harvard Law School dean Erwin N. Griswold reflected that JusticeDouglas ‘‘had great energy, great imagination, and [for] his first 10–15 years he was a greatjustice. After that, he, in my judgment, went steadily down hill. He got bored with the Court’swork. He dashed it off and the final 10 years at least of his membership on the Court wasnot, in my view, distinguished, and I’ve heard the same reaction expressed by other people.’’Hearing of the Senate Judiciary Committee on the Nomination of Clarence Thomas to theSupreme Court (September 17, 1991).

63 Of course, this assumes that one with 10 years of experience on the Supreme Courtwould perform no better or worse than one with 10 years of some other type of experience,an assumption that certainly bears further inquiry.

64 See equation (5) supra and accompanying text.65 This conclusion is supported by Posner, who notes that it is important ‘‘for an appellate

judge to bring to his job a background of knowledge about the behaviors out of which thecases he will be judging arise: to be, in short, ‘experienced.’ ’’ Posner, Aging and Old Age,supra note 14, at 193. Posner notes the particular relevance of this point to the optimum ageof appointment. Id.


in the judiciary when nominated by President Nixon. Throughout history,nearly half of those appointed to the Court came after significant judicialcareers. Some have suggested that this development represents an unhealthytrend toward professionalization of the Supreme Court, which as a vital po-litical body would be better served by a more diverse collection of individ-uals.66

The multiple regression results demonstrate that the average influence ofjustices coming from private practice is significantly greater than thosefrom other categories. In terms of both net influence and average annualinfluence, Tables 11 and 12 show that private attorneys have very high andstatistically significant coefficients (with all except cAV significant at the0.05 level). The coefficients on the other categories suggest that law profes-sors on average score close to private attorneys and that judges score lower,though better than the baseline executive and legislative officials. However,with the low t-ratios on these terms, little significance can be attributed tothose results.

Former judges as a class appear not to have been the most consistentlyinfluential of justices.67 Especially given the similar backgrounds of most ofthe current members of the Supreme Court, these results argue in favor ofbroader consideration of private attorneys (and perhaps law professors) forupcoming appointments.

3. Political Factors

For the most part, the political factors analyzed seem to have little sig-nificant effect on the various measures of influence. Surprisingly, the coef-ficients on most of the factors listed in Tables 11 and 12 are consistentlypositive.

When the Senate is controlled by his political opponents, a president maymake ‘‘safer’’ selections in attempting to fill a vacancy on the Court, andone might think that this constraint on the president’s selection could hinderhis ability to select the most influential candidate. In terms of cAV andcHV, Table 12 shows a marginally significant positive effect on an ap-pointee’s expected annual influence when the Senate is controlled by an op-

66 See, for example, Lee C. Bollinger, The Mind in the Major American Law School, 91Mich. L. Rev. 2167, 2176 (1993) (‘‘I sense that, especially with the tendency exhibited overthe past decades to give a high priority to prior judicial experience in making appointmentsto the bench, that we are heading toward a professionalized judiciary . . . . That does notbode well for society.’’).

67 Abraham similarly concluded that ‘‘[t]o raise judicial experience to the level of eitheran express or implied requirement would render a distinct disservice to the Supreme Court.’’Abraham, supra note 6, at 58. He notes that the justices rated ‘‘great’’ by Blaustein andMersky’s survey had surprisingly few years of judicial experience. Id. at 58 n.5.


position party. It might be that the political obstacle of Senate confirmationencourages discipline in the selection process, perhaps forcing it to turn onthe nominee’s experience and quality rather than political factors. In anyevent, a ‘‘hostile’’ Senate has tended to yield justices who are somewhatmore influential than otherwise.68

Over two-thirds of the justices were selected by presidents serving theirfirst term of office. First-term nominees have large positive influence coef-ficients, though only the cHV t-ratio approaches statistical significance.

One would also expect selection of a safe, ‘‘confirmable’’ nominee aftera prior candidate was rejected by the Senate. The regressions show that lit-tle predictive value can be attributed to this factor. It can at least be said thatthe rejection of a president’s nominee does not consistently lead to subparsubsequent nominations as some might have expected.69

IV. Conclusion: Can Judicial Selection Be Improved?

This study has developed an economic model of Supreme Court appoint-ments, a tool for assessing and comparing the influence of judges from thestandpoint of an appointing president. Based on analysis of the careers ofpast justices, it appears that certain common presidential assumptions maybe open to question.

The prospective Court packer would do well to place less emphasis onseeking the youngest confirmable nominee. Given two candidates of differ-ent age, a president should not select the younger simply because he expectsthat candidate to have a longer career and hence greater influence. Instead,this study suggests that age should be a positive factor in the selection ofSupreme Court justices, if it is considered at all. A nominating presidentshould also realize that an older candidate is more likely to provide an im-mediate effect on the course of the law. The selection of older candidatesshould also reduce the variability in quality of justices, not to mention thelikelihood of justices diverging philosophically from the views they held(or were perceived to hold) when nominated.

Although state and federal judges may be ‘‘safe’’ Supreme Court nomi-

68 An analysis of the fraction of the Senate voting to confirm showed no relationship be-tween the fraction voting for confirmation and the eventual influence of the appointee. Thisis not unexpected: most of the disciplinary effect of the advice and consent process occursbefore the confirmation is ever brought to a vote. A president facing a potentially hostileSenate knows well in advance to tailor his selection to overcome objections that could derailconfirmation.

69 A perfect example lies in President James Madison’s simultaneous nomination of twojustices after the Senate rejected Alexander Wolcott and John Quincy Adams refused toserve. Madison nominated Joseph Story, one of the truly great justices, and Gabriel Duvall,one of the most insignificant. See note 49 supra.


nations, private attorneys have made for significantly more influential jus-tices. Given the homogeneity of the present Court, this study recommendsheightened attention to diversity of background in future nominations andurges the consideration of a private attorney or perhaps a law professor.70

Of course, selection of Supreme Court justices will always entail individ-ual scrutiny, the gauging of character and of the political climate. Any no-tion that a statistical study can provide a president with ‘‘the answer’’ as towhich candidate will make the most influential justice is simply ludicrous.The value of a quantitative study such as this is not the replacement of po-litical judgment but its education, so that the nomination process can pro-ceed absent the baggage of uninformed (and potentially misleading) intu-itions. This study, questioning and testing certain basic assumptions,provides a first step in that education.

70 This recommendation is suggested by the results showing that private attorneys havetended to be the most influential of justices, although this study attempts no analysis of theeffects of homogeneity. A promising area for further research would consider the entire Courtas a portfolio of investments. One could hypothesize that, as with financial investments, di-versity minimizes risk and then consider which combinations of factors (for example, age,prior experience, political affiliation, and so forth) have produced the most effective Courts.

measuring the influence of supreme...

Documents