|
COPYRIGHT 2006 Harvard Society for Law and Public Policy, Inc.
INTRODUCTION
I. THE NEED FOR REFORM: THE EXPANSION OF LIFE TENURE A. The Expansion of Life Tenure Documented B. Periodization and Related Empirical Issues C. Explaining the Trends in Life Tenure D. Consequences of the Expansion of Life Tenure 1. Reduced Democratic Accountability 2. Increased Politicization of the Confirmation Process 3. A Rise in "Mental Decrepitude" on the Court E. The Rarity of Life Tenure in the World's Constitutional Courts II. TERM LIMITS FOR THE SUPREME COURT A. Imposing Term Limits Through Constitutional Amendment B. The Term Limits Proposal C. Advantages of the Proposal D. Objections to the Proposal (and Responses Thereto) III. ALTERNATIVE MEANS OF SOLVING THE TENURE PROBLEM AND THEIR DRAWBACKS A. Imposing Term Limits by Statute 1. The Calabresi-Lindgren Proposal 2. The Carrington-Cramton Proposal 3. The Constitutionality of the Two Statutory Proposals 4. The Desirability of Imposing Term Limits by Statute B. Imposing Term Limits through Informal Practice 1. Senate-Imposed Limits Through Term Limit Pledges 2. Limits Imposed Through Internal Court Rules 3. Justice-Imposed Limits Through Voluntary Retirement CONCLUSION APPENDIX
INTRODUCTION
In June 2005, at the end of its October 2004 Term, the U.S. Supreme Court's nine members had served together for almost eleven years, longer than any other group of nine Justices in the nation's history. (1) Although the average tenure of a Supreme Court Justice from 1789 through 1970 was 14.9 years, for those Justices who have retired since 1970, the average tenure has jumped to 26.1 years. Moreover, before the death of Chief Justice William Rehnquist in September 2005 and Justice Sandra Day O'Connor's announcement in July 2005 of her retirement that eventually took effect on January 31, 2006, five of the nine Justices had served on the Court for more than seventeen years, and three of those had served for more than twenty-three years. (2) The other four Justices had each already spent between ten and fourteen years on the Court. At the same time, four of these nine Justices were seventy years of age or older, and only one was under sixty-five--once the traditional retirement age in business. (3) Because of the long tenure of these members of the Court, there were no vacancies on the high Court from 1994 to the middle of 2005. (4)
We believe the American constitutional rule granting life tenure to Supreme Court Justices (5) is fundamentally flawed, resulting now in Justices remaining on the Court for longer periods and to a later age than ever before in American history. This trend has led to significantly less frequent vacancies on the Court, which reduces the efficacy of the democratic check that the appointment process provides on the Court's membership. The increase in the longevity of Justices' tenure means that life tenure now guarantees a much longer tenure on the Court than was the case in 1789 or over most of our constitutional history. (6) Moreover, the combination of less frequent vacancies and longer tenures of office means that when vacancies do arise, there is so much at stake that confirmation battles have become much more intense. Finally, as was detailed in a recent article by Professor David Garrow, the advanced age of some Supreme Court Justices has at times led to a problem of "mental decrepitude" on the Court, whereby some Justices have become physically or mentally unable to fulfill their duties during the final stages of their careers. (7) A regime that allows high government officials to exercise great power, totally unchecked, for periods of thirty to forty years, is essentially a relic of pre-democratic times. Although life tenure for Supreme Court Justices may have made sense in the eighteenth-century world of the Framers, it is particularly inappropriate now, given the enormous power that Supreme Court Justices have come to wield. (8)
In this Article, we call for a change to the life tenure rule for Supreme Court Justices. We begin in Part I by analyzing the historical data on the tenure of Supreme Court Justices and responding to a recent critique of this analysis. The causes and consequences of longer judicial tenures are also examined. Part I concludes by describing the approach to judicial tenure that all other major democratic nations and U.S. states have taken to demonstrate that, comparatively, the U.S. Supreme Court's system of life tenure is truly an outlier. (9)
To resolve the problems of life tenure, we propose in Part II that lawmakers pass a constitutional amendment pursuant to Article V of the Constitution instituting a system of staggered, eighteen-year term limits for Supreme Court Justices. (10) The Court's membership would be constitutionally fixed at nine Justices, whose terms would be staggered such that a vacancy would occur on the Court every two years at the end of the term in every odd-numbered calendar year. Every one-term President would thus get to appoint two Justices and every two-term President would get to appoint four. Our proposal would not apply to any of the nine sitting Justices or to any nominee of the President in office when the constitutional amendment is ratified. Supreme Court term limits ought to be phased in, as was done with the two-term limit for Presidents, which did not apply to the incumbent President when it was ratified.
Our proposal builds on the views of a number of distinguished commentators and judges from broadly varying backgrounds who have opposed life tenure for federal judges, including some of the most venerable figures in American history. Thomas Jefferson, for example, denounced life tenure as wholly inconsistent with our ordered republic. (11) Accordingly, he proposed renewable terms of four or six years for federal judges. (12) Robert Yates, who wrote as Brutus during the ratification period, denounced life tenure for federal judges and the degree to which it separated courts from democratic accountability. (13)
Most relevant to our own proposal are the writings of several modern commentators in support of term limits for Supreme Court Justices. In 1986, Professor Philip Oliver (14) proposed fixed, staggered terms of eighteen years that would, among other benefits, allow for appointments every two years, balance the impact that Presidents can have on the Court's makeup, and eliminate the possibility of Justices' remaining on the Court beyond their vigorous years. (15) Several other commentators have also called for term limits for Supreme Court Justices, or for federal judges generally, but did not propose terms of eighteen years. (16) After an early version of this Article was written and discussed publicly, but before its publication, James DiTullio and John Schochet proposed a system of eighteen-year term limits for Supreme Court Justices in a student Note. (17) Their primary concerns were not that Justices are staying too long on the Court but that the current system allows for strategic timing of retirements, encourages the appointment of young nominees to the Court, and fails to distribute appointments evenly across different presidencies. (18) Finally, Professor L.A. Powe, Jr. recently identified life tenure for members of the Supreme Court as "the Framers' greatest (lasting) mistake," (19) and called for eighteen-year term limits on Supreme Court Justices. (20) Of the leading legal scholars to write about Supreme Court term limits to date, only one figure, Professor Ward Farnsworth of Boston University, has defended life tenure as it currently operates. (21)
Although many commentators have thus called for term limits on Supreme Court Justices, their proposals have received little attention, perhaps for two reasons. First, many Americans mistakenly believe that a system of life tenure is necessary to preserve an independent judiciary. Second, despite these scholars' various proposals, a comprehensive case has yet to be made in the literature for the need to reform life tenure. We seek to make that case by demonstrating that the real-world, practical meaning of life tenure has changed over time and is very different now from what it was in 1789 or even 1939. This significant change provides a strong, nonpartisan justification for reconsidering life tenure.
Our proposal is ultimately a Burkean reform intended to move the Justices back toward an average tenure that is similar to what the average tenure of Justices has been over the totality of American history. Just as the two-term limit on Presidents restored a tradition of Presidents stepping down after eight years in office, our eighteen-year term limit on Supreme Court Justices would push the average tenure of Justices back toward the 14.9-year average tenures that prevailed between 1789 and 1970 and away from the astonishing 26.1 year average tenure enjoyed by Justices who stepped down between 1970 and 2005. (22) Our proposed amendment would thus merely restore the practice that prevailed between 1789 and 1970 and would guarantee that vacancies on the Court would open up on average every two years, with no eleven-year periods without a vacancy as has happened between 1994 and 2005. This then is a fundamentally conservative call for reform, all the more so because we resist the calls of many commentators for a very short tenure for Supreme Court Justices. The eighteen-year nonrenewable term we propose is more than long enough to guarantee judicial independence without producing the pathologies associated with the current system of life tenure.
Our proposal for imposing on Supreme Court Justices a staggered, eighteen-year term limit, with a salary for life and an automatic right to sit on the lower federal courts for life, could theoretically be established in a variety of ways, but the only way we approve of is through passage of a constitutional amendment pursuant to Article V. Accordingly, we outline in Part II below our proposal for a constitutional amendment instituting term limits. (23) We then highlight the advantages of passing such an amendment and address potential counterarguments. Short of amending the Constitution, Professors Paul Carrington and Roger Cramton have recently proposed a system of term limits for Supreme Court Justices instituted by statute. (24) In Part III below, we consider two statutory proposals for instituting Supreme Court term limits, one of our own devising as well as the Carrington-Cramton proposal. We consider the arguments in favor of and against the constitutionality of these two proposed statutes, concluding that statutorily imposed term limits on Supreme Court Justices are unconstitutional. The statutory proposal presents some close constitutional questions, and one grave danger it poses is that it would be manipulable by future Congresses. (25) For these reasons, we believe that term limits ought to be established by a constitutional amendment and that the proposed statute is unconstitutional.
Finally, Part III concludes by arguing that a system of term limits could in theory be achieved more informally through a variety of measures. (26) Specifically, we consider the opportunities that the Senate, the Court, and even individual Justices have for informally instituting term limits: the Senate by imposing term-limit pledges on nominees during confirmation hearings, (27) the Court through an adjustment of its internal court rules and seniority system, (28) and individual Justices by establishing an informal tradition of leaving the Court after a term of years, (29) as Presidents did before passage of the Twenty-Second Amendment. (30) Finally, we conclude that the only way to realize a system of Supreme Court term limits is through the passage of a constitutional amendment. We urge lawmakers to consider passing such an amendment before a new wave of resignations occurs. Establishing a system of term limits is an important reform that would correct the problem of a real-world, practical increase in the actual tenure of Supreme Court Justices.
I. THE NEED FOR REFORM: THE EXPANSION OF LIFE TENURE
A. The Expansion of Life Tenure Documented
Life tenure for Supreme Court Justices has been a part of our Constitution since 1789, when the Framers created one Supreme Court and provided that its members "shall hold their Offices during good Behaviour." (31) The Framers followed the eighteenth-century English practice, which developed in the wake of the Glorious Revolution of 1688, of securing judicial independence through life tenure in office for judges. (32) But since 1789, Americans have experienced drastic changes in medicine, technology, politics, and social perceptions of judges and of the law that have changed the practical meaning of life tenure for Justices.
We analyzed this change by calculating the age and tenure in office for each Justice (33) and by examining the number of years between vacancies on the Court. This empirical analysis revealed three critical and significant trends: the real-world, practical meaning of life tenure has expanded over time; Justices have been staying on the Court to more advanced ages than in the past; and, as a result, vacancies have been occurring less frequently than ever before.
Surprisingly, these trends have not been gradual. First, as Chart 1 summarizes, the average tenure of a Supreme Court Justice has increased considerably since the Court's creation in 1789, with the most dramatic increase occurring between 1971 and the present. In the first thirty-two years of the Supreme Court's history, Justices spent an average of just 7.5 years on the Court, perhaps due in large part to the difficult conditions of circuit riding and a series of very short-lived initial appointments, including a short recess appointment for Chief Justice Rutledge. (34) The average tenure of Justices then increased significantly between 1821 and 1850 to 20.8 years before declining over the next four thirty-year periods (spanning the period from 1851 through 1970) to an average tenure of only 12.2 years from 1941 through 1970. Then, from 1971 to 2000, Justices leaving office spent an average of 26.1 years on the Court, an astonishing fourteen-year increase over the prior period, 1941-1970. (35) Justices leaving office between 1971 and 2006 thus spent more than double the amount of time in office, on average, than Justices leaving office between 1941 and 1970.
[ILLUSTRATION OMITTED]
A cumulative average for the period of 1789-1970 puts this dramatic increase, reflected in the tenure of post-1970 retirees, in perspective. Compared to the average of 26.1 years in office for Justices retiring after 1970, the average Justice leaving office between 1789 and 1970 spent only 14.9 years on the bench. Thus, regardless of the basis for comparison--the average of 12.2 years for Justices leaving office during 1941 through 1970 or the average of 14.9 years for Justices leaving office from 1789 through 1970--the increase to an average tenure of 26.1 for Justices leaving office since 1970 is astounding. Indeed, four of the seven longest-serving Justices of all time are among the dozen Justices who have left the bench since 1970: Justices Douglas (36.6 years), Black (34.1 years), and Brennan (33.8 years), and Chief Justice Rehnquist (33.7 years).
One question that has arisen is whether the trends we document in Chart 1 are merely artifacts of how we defined our periods. (36) In Chart 1, the middle five periods are 30 years in length, whereas the first period is 32 years in length and the last period is 35 years. (37) We must note that the decision to remain on the Court is one that is made continuously, with, for example, Justice Black serving from 1937 to 1971 and Justice Douglas serving from 1939 to 1975. Thus, while the full realization of the trend toward longer tenures does not manifest itself until the retirements that occur after 1970, this post-1970 trend in part reflects behavior and decisions that were made in the 1950s and 1960s by Justices Black and Douglas. Perhaps with Black and Douglas as an example, recent Justices have been remaining on the bench much longer than was common for most of American history. Indeed, all of the last six Justices to leave the Court (starting in 1990) rank among the top quarter in longevity on the bench: all served for 24 years or more. If one adds the 1971-1975 retirees (Black, Harlan, and Douglas) to the 1941-1970 period, the new 1941-1975 mean tenure is 15 years (instead of 12.2 years for the next-to-last period in Chart 1), and the nine post-1975 retirees average 25.1 years on the bench (instead of 26.1 years in the last period in Chart 1). In other words, even if the two longest-serving Justices in the last half-century are lumped with an earlier period, the most recent period still shows a dramatic increase of ten years in the length of judicial tenure over the prior period and over the historical average tenure on the Supreme Court.
For those who would like to see the data on time in office without any period selected by the researcher, we present these same data as a set of overlapping averages to smooth out variations enough to see a trend line, but without assigning Justices to just one period. Chart 2 presents the same data as Chart 1 without any periodization. We chart the mean of the last nine Justices to leave the Court for every retirement or death starting with the ninth Justice to resign, retire, or die in office (in 1804). This graph plots all 95 overlapping sets of nine consecutive retirements, resignations, or deaths that the Court has experienced so far. In other words, if at the end of each judicial tenure, we looked back over the last nine Justices to leave the Court and computed an average tenure for this set of nine Justices, the average of those nine tenures in office (a nine-Justice lagging average) would be the data point represented in Chart 2. (38) For example, the first peak labeled in Chart 2 is a lagging average of 22 years in office corresponding to the year 1851. This is the average of the tenure in office of Justice Levi Woodbury, who died in office in 1851, combined with the tenure of the immediately prior eight Justices who left office by death or resignation between 1829 and 1851. The next data point is the lagging average of the nine Justices who left the Court between 1834 and 1852, when Justice John McKinley died in office.
[ILLUSTRATION OMITTED]
As Chart 2 shows, the five longest of the 95 sets of nine consecutive retirements are the sets ending with the last five retirements (1991, 1993, 1994, 2005, and 2006). In the Nineteenth Century, we see a broad peak that consists of three spikes (three groups of nine Justices, the last of which retired in 1835, 1851, and 1873), then a gradual (but uneven) drop to a bottom in the 1956-1969 period, followed by a rise to the unprecedented levels that we have experienced since 1991. Chart 2 also shows much the same general pattern as Chart 1: a substantial rise, a gradual drop, followed by a rise starting for those Justices who retired after 1970, which by 1991 leads to unprecedented average levels for the last five groups of nine consecutive retirements. Chart 2 shows that our choice of periods in Chart 1 fairly represents the pattern of retirements measured at each individual Justice's departure from office. (39)
Not only are Justices staying on the Court for longer periods, but they are also leaving office at more advanced ages than ever before. As Chart 3 highlights, the average age at which Justices have left office has generally risen over time, but like the average tenure of office, it has dramatically increased for those retiring in the past thirty-five years.
[ILLUSTRATION OMITTED]
In the five thirty-year intervals between 1789-1940, the average age of Justices upon leaving office rose from 58.3 to 72.2 years of age, but then dropped to about 67.6 years of age for the 1941-1970 period. Yet in the last period, 1971-2006, Justices left office at an average age of 78.7 years. Justices who have left office since 1970 have thus been, on average, eleven years older when leaving the Court than Justices who left office in the preceding thirty-year period, 1941-1970, and more than six years older than Justices in the next highest period, 1911-1940, one that famously included the era of the so-called nine old men. In addition, comparing the average retirement age since 1970 with a cumulative average age of all Justices retiring from 1789 through 1970 is equally revealing. The average Justice leaving office after 1970 (age 78.7) is ten years older than the average Justice leaving office prior to 1970 (age 68.3). Thus, the average age at which Justices have retired has increased markedly throughout history, and most sharply in the past thirty-five years.
Chart 4 shows the same data without periodization, reflecting a lagging average age for 95 sets of nine consecutive Justices to leave office. This chart reveals peaks of 75 years old in 1881 (reflecting those leaving office in 1864-1881) and 77 years old in 1941 (reflecting those leaving office in 1930-1941). But since 1991, the lagging average age of the last nine retirements has been slightly higher than at those earlier peaks, and much higher than most prior years. Since 1991, the nine-Justice lagging average age at retirement has been an unprecedented 78 to 79 years old (reflecting those leaving office in 1969-1991 through those leaving in 1981-2006).
[ILLUSTRATION OMITTED]
The mean age for both men and women electing to receive Social Security retirement benefits has hovered around 64 since 1970, (40) and the age for receiving full benefits is being increased to age 67 for those born after 1960. Although in 2000, 47% of Americans of ages 60-64 were still in the labor force, that proportion drops by almost half for ages 65-69 (24%). Only 13.5% of Americans were still in the labor force at ages 70-74, and only 5.3% were still working at ages 75 and older. (41) This contrasts with a mean retirement age on the Court since 1971 of 78.7 years. All but three Justices on the current Court, Chief Justice John Roberts and Justices Clarence Thomas and Samuel Alito, are at an age when most Americans have already retired. (42) When one compares the age distribution of the current Supreme Court with the age distribution of all federal judges, the difference is similarly stark. From 1984 to 2002, less than 17% of Article III judges were 66 or older and less than 5% were 71 or older. (43)
Given that Justices have been staying on the Court for longer periods and retiring later in life than ever before, it is not surprising that vacancies on the Court have been opening up much less frequently than in the past. Indeed, as Chart 5 indicates, (44) the average number of years between vacancies has increased sharply in the past thirty years.
[ILLUSTRATION OMITTED]
These figures are affected by the varying size of the Supreme Court over time. (45) During most of the first two periods, the Court had fewer than nine members, which means the figures calculated for those periods are higher than they would have been with a larger Court. With fewer Justices, seats can be expected to open up less frequently. (46) In the third period, from 1851 through 1880, the gaps between vacancies are more difficult to compare because the size of the Court varied from eight to ten members. (47) Yet, looking at the figures for the first two periods, it is probable that, if the Court had been the size it is today, the time between vacancies would tend to be closer to the figures from 1881 through 1970. Indeed, the increase in 1837 from seven to nine members, though primarily a power grab by the Jacksonians, may also have been in part a reaction to the longer tenure, advanced ages, and longer gaps between retirement after 1811 (as suggested by the data in Charts 1 through 5). Since 1869, the Court's membership has been fixed at nine Justices, which makes a comparison to the last four periods the most meaningful to current policy analyses. Chart 5 demonstrates that from 1881 through 1970, the average number of years between commissions stayed consistent at about 1.6 to 1.8; since 1970, it has nearly doubled to 3.1 years.
Moreover, the Court went for nearly eleven years--between 1994 and 2005--without a vacancy, the longest period between vacancies since the Court's membership settled at nine Justices. (48) Eleven years is long enough in theory to deprive a successful, two-term President of the chance to appoint even a single Justice.
The cumulative average from 1789 through 1970 further highlights the remarkable increase in time between vacancies that has occurred since 1970: on average, vacancies occurred on the Court every 1.9 years from 1789 to 1970 and then began occurring only every 3.1 years since 1971. After the two 1971 appointments, 3.4 years elapsed between vacancies. Thus, in the past few decades, vacancies have opened up every 3.1 to 3.4 years, which is about double those in the most comparable years--from 1881 through 1970--and more than one year longer than the cumulative average from 1789 through 1970.
Strikingly, since the Court was fixed at nine members in 1869, three of the five longest times between vacancies occurred in the last thirty years: between November 12, 1975 and July 3, 1981; between July 3, 1981 and September 26, 1986; and between August 3, 1994 and September 4, 2005. Jimmy Carter was the only President in American history to serve at least one complete term and never make an appointment to the Supreme Court. If George W. Bush had lost his bid for re-election in 2004, he would have been the second. As it is, he is only the third person elected twice to the Presidency who has had to wait until his second full term to make his first Supreme Court appointment (the others being Franklin Roosevelt and James Monroe). Of the thirty-four presidential four-year terms since the number of Justices was finally fixed at nine in 1869, only four expired without an appointment to the Supreme Court. Among the first twenty-seven terms from 1869 through 1973, only once did a four-year presidential term pass without an appointment (FDR, 1933-1937). By contrast, among the last seven completed terms, three--almost half--were devoid of Supreme Court appointments: Jimmy Carter's term, Bill Clinton's second term, and George W. Bush's first term. There can be no doubt that Supreme Court vacancies are opening up much less often in the post-Warren Court era.
Chart 6 shows a lagging average of the last nine appointments to the Court when each Justice leaves the bench. The mean period between openings was 3.8 years when Chief Justice Rehnquist died, the longest lagging average in Chart 6, and was 3.4 years when Justice O'Connor was replaced, the third-longest nine-Justice average in history. Note that since 1869 when the Court was fixed at nine Justices, the average number of years for the last nine vacancies has gone above 2.1 years only twice: in 1937 when the "nine old men" held sway, and continuously since 1986, the entire period of the Rehnquist Court.
[ILLUSTRATION OMITTED]
These historical trends represent a grand change in the practical meaning of the Constitution's grant of life tenure to Supreme Court Justices. The Founding Fathers were famously known for their disdain for "unaccountable autocrats out of touch with the typical citizen's concerns; who cling to power long after they have sufficient health to perform their duties; who cannot be removed from office by democratic agency." (49) The Framers gave Supreme Court Justices life tenure in an era when the average American could expect to live only thirty-five years. (50) Now, Justices are appointed at roughly the same average age as in the early years of our history, (51) but they benefit from an average life expectancy of seventy-seven years. (52) Of course, this statement alone significantly overstates the relevant difference because of higher rates of infant mortality two hundred years ago. Thus, a more relevant comparison might be that in 1850, white men who reached the age of forty could expect to live another 27.9 years, compared to such men in 2001, who could expect to live another 37.3 years. Largely as a result of this 9.4-year increase in life expectancy, today the average Justice who is appointed to the Court in his early fifties can expect to sit on the Court for nearly three decades, whereas the average Justice appointed to the Court in his early fifties in 1789 might have expected to sit on the Court for only two decades. Today's Justices enjoy a potential tenure that is fifty percent longer than that of their typical eighteenth- and nineteenth-century predecessors.
B. Periodization and Related Empirical Issues
In a somewhat odd article in the Washington University Law Quarterly, Professor David R. Stras and Ryan W. Scott first argue against some of the empirical claims of a trend toward longer tenures on the U.S. Supreme Court, but then suggest increasing retirement benefits as a solution to a problem whose existence they mostly reject or minimize. (53) As we expressed in public discussions at the Duke Law School conference on term limits for Justices in the spring of 2005, we too favor offering greater financial benefits as an inducement to earlier retirement from the Court. Unless these golden parachutes were extremely generous, however, we suspect that they would be effective only occasionally in the first few decades after implementation. But over time, expanded benefits could help introduce a norm of retirement by offering a focal point for norm creation. Thus, our disagreements with Stras and Scott are related to their empirical claims, not their policy proposals, which we believe offer considerable merit.
Disagreeing with our claim that Supreme Court tenure has increased dramatically for Justices retiring since 1971, Stras and Scott argue that our "claim is empirical, and a closer look at the data suggests that it depends more on the chosen period lengths than a bona fide trend." (54) Thus, Stras and Scott's main empirical argument is that the trend that we document in longer judicial tenures for the twelve Justices retiring since 1971 is a function of the periods we chose. They claim that splitting our data into other groupings makes the pattern in our data disappear:
A year here or there, on one side or the other of a cutoff, and the average for the period might rise or fall considerably. Yet a rendering of the data that flattens those periods, taking judges in groups of five, likewise reveals no dramatic recent trend or unprecedented length of service. (55)
Note that Stras and Scott assert as if it were a fact that "[a] year here or there, on one side or the other of a cutoff, and the average for the period might rise or fall considerably." As a quick check of our data revealed to us, this claim is simply false. We tried starting each of our seven Chart 1 periods a year earlier or a year later, or ending a year earlier or a year later. Examining all 28 different one-year changes in cutoff dates, we discovered that in every case, there were trivial differences at most. For 13 of the 28 changes in cutoff dates, there was no change at all in mean tenure on the Court with a year added or subtracted from a period; in 12 of the remaining cases the differences in mean tenure on the Court were less than one year. In only 3 cases did the differences in mean tenures for a period range as high as 1.0 years, with the largest difference being 1.3 years. In that case, if the 1941-1970 period had started in 1942 instead of 1941, the mean tenure for Justices leaving the Court in that slightly shorter period would have been 13.5 years, rather than 12.2 years, a trivial difference when contrasted with the jump to a mean tenure of 26.1 years in the 1971-2006 period. It appears that Stras and Scott never bothered to check their factual claim that a year here or there would make a considerable difference.
Then Stras and Scott argue that "a rendering of the data that flattens those periods, taking judges in groups of five, likewise reveals no dramatic recent trend or unprecedented length of service." (56) They never explain why one would want to "flatten" effects. Such an approach is a recipe for committing Type II error (false acceptance of the null hypothesis). (57) To reject the null hypothesis that the data are random, one should test the possibility that there might be a trend in the data. If one wants to reduce data into groups to facilitate easier data presentation, one must to be careful not to eliminate any trends or effects that might be in the data. Instead of testing any apparent trends in the data to see whether they are statistically significant, Stras and Scott appear to be straining to find ways of presenting the data that lump Justices together in a way that will make the patterns in the data disappear. Doing this tends to produce Type II error, which is precisely what happens with Stras and Scott's analysis.
We chose the periods in Chart 1 because they made the presentation of the data clearer and did not suppress the most important time trends in the data. For those worried about our periodization, Chart 2 presents the data without periodization, (58) charting the mean of the last nine Justices to retire for every retirement starting with the ninth Justice to resign, retire, or die in office (in 1804). Chart 2 nicely supports the periods we chose to use in Chart 1. (59)
Instead of our Charts 1 or 2, Stras and Scott offer two simple charts, one that divides the Justices by the decade they retired (60) and another that divides the Justices into groups of five Justices based on when they were appointed to the Court. (61) They do not report the cell counts for their chart dividing the Justices by decade, (62) perhaps for a reason. In discussing this chart, they mention the average tenures for Justices retiring in four decades: the 1830s, the 1970s, the 1980s, and the 1990s. (63) Stras and Scott, however, fail to mention that the sample sizes for their estimates for length of tenure during these decades are only 3 cases for each of these first three decades and 4 cases for the 1990s. By contrast, in our Chart 1, by using periods of at least 30 years, the smallest sample size for any of our groups is 11 cases (for the 1821-1850 period). For Stras and Scott to report a mean for a cell count smaller than 5 cases is a questionable choice, a problem that can cause them to see patterns where there are none or to miss patterns that are present. The reason for presenting the data in grouped averages is to reduce the mass of individual data points to recognizable patterns or trends. If too few data points are combined, then one can mistake meaningful differences for random variation, as Stras and Scott appear to have done. Further, even in their highly unreliable chart with the very small sample sizes for each decade, three of the four decades with highest mean tenure are three of the last four decades (the 1970s, 1990s and 2000s), a result that tends to support, rather than reject, our hypothesis that the period since 1971 is indeed substantially different from the typical pre-1970 period. More importantly, Stras and Scott do not offer theoretical or empirical explanations for selecting time periods with small sample sizes and for encouraging Type II error that partially suppresses the pattern we hypothesize.
The problems with Stras and Scott's Chart 2 are even more serious than with their Chart 1. In their Chart 2, (64) Stras and Scott present data on the average tenure of Justices after they have combined the Justices into groups of five based on when the Justices were appointed. Given that we are writing about a problem of delayed retirement, not appointment, it is unclear why they would seek to test our hypothesis or our groupings by using the date of appointment rather than the date of retirement. They disclose the switch in a footnote, but do not offer a theoretical reason for making it. The most likely effect of this grouping would seem to be to facilitate Type II error by suppressing any patterns in the data.
Stras and Scott report that dividing the data into twenty-one groups of five Justices based on date of appointment "reveals no dramatic recent trend or unprecedented length of service." (65) We, of course, contend that those Justices retiring since 1971 serve longer on the court than those before. But by examining Justices in groups of five Justices by year of appointment, Stras and Scott's groupings do not contain even one complete set of five Justices who retired after 1970. Their last full group of five appointees includes Abe Fortas, who was appointed in 1965 and served only 3.6 years on the Court, and Arthur Goldberg, who was appointed in 1962 and served only 2.8 years. Because Fortas and Goldberg did not resign after 1970, they should not be lumped with the post-1970 retirees in testing our hypothesis about post-1970 retirees.
After this last full set of five appointees (including Goldberg and Fortas), Stras and Scott's last group of appointees (and last data point) has only three members--Justices Powell, Rehnquist, and O'Connor (66)--again raising concerns about very small sample sizes. But indeed, most other ways of making groupings of five consecutive Justices would still not support Stras and Scott's claim that the trend since 1970 "reveals no dramatic recent trend or unprecedented length of service." (67) We tested every one of the 99 sets of five consecutive Justices based on date of retirement or death, and the longest average tenure for any set of five consecutive Justices leaving office occurred with the death of Justice Rehnquist in 2005.
In order to suppress (or, in their words, "flatten") the unprecedented length of tenure of retirees from the Court since 1971 (as revealed in our Charts 1 and 2), Stras and Scott went to extraordinary lengths. As we have shown, in their first chart based on tenure by decade, they routinely base conclusions on unreliable cell counts of three cases. Even given this questionable choice, three of the four decades with the longest average tenure are since 1970, which supports our conclusion. Their second chart, based on groupings of five Justices, has even more problems:
(1) Stras and Scott switch from date of retirement to date of appointment with no justification, though our hypothesis concerns length of tenure at retirement, not appointment;
(2) By using year of appointment, they are able to lump the short and highly unusual 1960s tenure and resignations of Abe Fortas and Arthur Goldberg with the post-1970 retirees;
(3) By using year of appointment, they have no complete set of five justices retiring after 1970 to test our hypothesis that post-1970 retirees are different.
The pattern of a recent unprecedented length of tenure on the Court is not an artifact of how the data are grouped; it is a real pattern in the data, which our grouping nicely reflects. We found one or more recent strings of Justices as the longest serving in history when we examined all 99 strings of five consecutive Justices leaving the Court; all 98 strings of six consecutive Justices leaving the Court; all 97 strings of seven consecutive Justices; all 96 strings of eight consecutive Justices, all 95 strings of nine consecutive Justices (shown in Chart 2); all 94 strings of ten consecutive Justices, all 93 strings of eleven consecutive Justices; and all 92 strings of twelve consecutive Justices. These robust results do not depend on any particular periodization, just on having a string of at least five Justices leaving office. (68)
Stras and Scott also try to make our arguments and our periodization look arbitrary by creating a false impression of the contrasts that we (and those who report our data) find most important. After asserting that our empirical claim "depends more on the chosen period lengths than a bona fide trend," Stras and Scott assert,
Calabresi and Lindgren document an increase in term length from an average of 12.2 years during 1941-1970 to an average of 26.1 years from 1971-2000. Related statistics have since popped up in a number of editorials in the popular press [Stras and Scott's footnote here points to two popular newspaper columns]. On its face, the figure reveals an "astonishing" and "dramatic" increase. (69)
Stras and Scott then proceed to point out that the Justices leaving the bench in the 1941-1970 period had uncharacteristically short tenures, which our Chart 1 above nicely summarizes.
Stras and Scott here make it appear that the primary argument on changes in the length of tenure made by us and by the reporters and columnists they cite is a contrast of the 1941-1970 period, which is perhaps unusual, with the 1971-2000 period. This implication is false. Neither of the two articles they cite actually contrasts these two periods; both instead compare the post-1970 period to the 182-year period from 1789 to 1970. (70) The most prominent other news articles to discuss our data--in the New York Times (71) and the Boston Globe (72) (both cited in an earlier draft of Stras and Scott) (73)--also compare the post-1970 period to the 182-year period from 1789 to 1970, with no mention of the 1941-1970 period.
In the public version of our 100-page manuscript, (74) other than in Chart...
Read the full article for free courtesy of your local library.
|
Bookmark this article
Print this article
Link to this article
Email this article
Digg It!
Add to del.icio.us
RSS
