Emory Law Journal

Is the Roberts Court Especially Activist? A Study of Invalidating (and Upholding) Federal, State, and Local Laws
Lee Epstein,
Andrew D. Martin Provost Professor of Law and Political Science and the Rader Family Trustee Chair in Law at the University of Southern California (http://epstein.usc.edu/). This Article was prepared for the 2011 Randolph W. Thrower Symposium, Emory University School of Law, February 1, 2011.Andrew D. Martin is Professor of Law and Political Science at Washington University in St. Louis.We thank the National Science Foundation and the Center for Empirical Research in the Law at Washington University for supporting our work on the U.S. Supreme Court. We also thank Stefanie Lindquist and Chad Westerland for supplying data on cases reviewing the constitutionality of legislative action.The dataset we use in this Article is available at http://epstein.usc.edu/research/RobertsActivism.html.

Introduction

Is the Roberts Court especially activist or, depending on your preference, especially lacking in judicial self-restraint? If we define judicial self-restraint as a reluctance to declare legislative action unconstitutional 1Richard A. Posner writes that at least three meanings can be assigned to the term “judicial self-restraint”:(1) [J]udges apply law, they don’t make it (call this “legalism” . . . or, better, “the law made me do it”); (2) judges defer, to a very great extent, to decisions by other officials—appellate judges defer to trial judges and administrative agencies, and all judges to legislative and executive decisions (call this “modesty,” or “institutional competence,” or “process jurisprudence”); (3) judges are highly reluctant to declare legislative or executive action unconstitutional—deference is at its zenith when action is challenged as unconstitutional (call this “constitutional restraint”).Richard A. Posner, The Rise and Fall of Judicial Self-Restraint, 100 Calif. L. Rev. 519, 521 (2012). Judge Posner focuses on (3), as do we. But, unlike Judge Posner, we limit our analysis to legislative action. and confine the analysis to the 1969–2009 Terms, 2This caveat is important because Epstein and Landes, analyzing data for a far longer time span (1937–2009 Terms) but studying federal laws only, reach a somewhat different conclusion. Lee Epstein & William M. Landes, Was There Ever Such a Thing as Judicial Self-Restraint?, 100 Calif. L. Rev. 557 (2012). They find that the majority of Justices appointed before the early 1950s were reluctant to strike down laws regardless of whether the law accorded with their ideological values. Id. at 569–76. But for Justices appointed since 1952, Epstein & Landes’s findings parallel ours: the vast majority were opportunistic restraintists (activists), willing to uphold laws that were consistent with their policy preferences and strike those that were not. Id. at 573–77. the answer is no. The Roberts Justices, just as their immediate predecessors, are neither uniform activists nor committed restraintists. Rather, the Justices’ votes to strike (and uphold) statutes seem to reflect their political preferences toward the policy content of the law, and not an underlying preference for restraint (or activism).

In a nutshell, liberal Justices tend to invalidate conservative laws and conservative Justices, liberal laws. This holds regardless of whether we examine all the Justices’ votes simultaneously 3 See infra Part IV. or each Justice individually. 4 See infra Part V.

I. Simple Approaches to Analyzing Judicial Self-Restraint and Activism

We are not the first to examine judicial self-restraint from an empirical perspective. To the contrary. More than a handful of law professors and social scientists have written ambitious papers using large-N datasets and sophisticated statistical methods. 5Recent examples include Stefanie A. Lindquist & Frank B. Cross, Measuring Judicial Activism (2009); Ruth Colker & Kevin M. Scott, Dissing States?: Invalidation of State Action During the Rehnquist Era, 88 Va. L. Rev. 1301 (2002); Robert M. Howard & Jeffrey A. Segal, A Preference for Deference? The Supreme Court and Judicial Review, 57 Pol. Res. Q. 131 (2004); Stefanie A. Lindquist & Rorie Spill Solberg, Judicial Review by the Burger and Rehnquist Courts: Explaining Justices’ Responses to Constitutional Challenges, 60 Pol. Res. Q. 71 (2007); Rorie Spill Solberg & Stefanie A. Lindquist, Activism, Ideology, and Federalism: Judicial Behavior in Constitutional Challenges Before the Rehnquist Court, 1986–2000, 3 J. Empirical Legal Stud. 237 (2006); David L. Weiden, Judicial Politicization, Ideology, and Activism at the High Courts of the United States, Canada, and Australia, 64 Pol. Res. Q. 335 (2011); and Tom S. Clark & Keith E. Whittington, Ideology, Partisanship, and Judicial Review of Acts of Congress, 1789–2006 (May 22, 2009) (unpublished manuscript) (on file with authors), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1475660. Classic empirical studies of judicial invalidation of legislative action include Jonathan D. Casper, The Supreme Court and National Policy Making, 70 Am. Pol. Sci. Rev. 50 (1976); Robert A. Dahl, Decision-Making in a Democracy: The Supreme Court as a National Policy-Maker, 6 J. Pub. L. 279 (1957); and John B. Gates, Partisan Realignment, Unconstitutional State Policies, and the U.S. Supreme Court, 1837–1964, 31 Am. J. Pol. Sci. 259 (1987). Curiously, though, far simpler analyses tend to dominate contemporary debates 6By this, we mean displays or lists showing the fraction of cases in which the Justices were in the majority when the Court invalidated a (federal, state, or local) law. Examples, in diverse outlets, include Lee Epstein & Thomas G. Walker, Constitutional Law for a Changing America: Institutional Powers and Constraints 37 (7th ed. 2011); Thomas M. Keck, The Most Activist Supreme Court in History: The Road to Modern Judicial Conservatism 202 tbl.6.2 (2004); and Paul Gewirtz & Chad Golder, So Who Are the Activists?, N.Y. Times, July 6, 2005, at A19. —with Figure 1 providing an example.

Figure 1. Voting with the Majority When the Court Invalidates a Federal, State, or Local Law, 1969–2009 Terms 7Data are from the “Justice-Centered Data: Cases Organized by Supreme Court Citation” version of the U.S. Supreme Court Database at http://www.supremecourtdatabase.org/ and include only orally argued cases (decisionType = 1, 6, or 7). We exclude Justice Sotomayor, who participated in only five cases in which the Court invalidated a law.

Figure 1

The idea behind Figure 1 (and similar data displays) is that we can learn about the Justices’ commitment to judicial self-restraint by studying the fraction of cases in which they vote with the majority when the Court invalidates legislative action. 8 See, e.g., Gewirtz & Golder, supra note 6 (“One conclusion our data suggests is that those justices often considered more ‘liberal’—Justices Breyer, Ruth Bader Ginsburg, David Souter and John Paul Stevens—vote least frequently to overturn Congressional statutes, while those often labeled ‘conservative’ vote more frequently to do so. At least by this measure (others are possible, of course), the latter group is the most activist.”). The lower the fraction, the more restrained the Justice (or so the argument goes). Using this strategy, we would conclude that the Justices located at the top of Figure 1 exercise greater restraint than the Justices at the bottom. Some commentators might even draw the inference that liberals (e.g., Brennan and Marshall) are the more aggressive and conservatives (e.g., Alito and Rehnquist) the meeker. 9And, in fact, the correlation between the fraction and the Justices’ relative liberalism is between .33 and .57 (depending on how we measure the Justices’ ideology), suggesting that liberal Justices are more activist. For the reasons we offer in the text, however, we should not make much (if anything) of this correlation.

As political scientists have pointed out, however, these conclusions would be premature because Figure 1 is flawed in at least two ways. First, it conveys whether a Justice is in the majority when the Court reviews the constitutionality of a law and invalidates it—not when the Court upholds it. 10 See, e.g., Clark & Whittington, supra note 5, at 13 (“Neither social science analyses nor narrative histories of the Court and judicial review give much attention to the cases in which the Court upholds legislation . . . .”). Besides wasting information (we can learn as much about judicial self-restraint from all the cases and not just invalidations), examining only “strike” cases can lead to misleading inferences. 11 See Stefanie A. Lindquist et al., The Rhetoric of Restraint and the Ideology of Activism, 24 Const. Comment. 103, 108 (2007); Nicholas S. Zeppos, Deference to Political Decisionmakers and the Preferred Scope of Judicial Review, 88 Nw. U. L. Rev. 296, 306 (1993); Clark & Whittington, supra note 5, at 8. Imagine a Justice who almost always votes with the majority when the Court strikes a law (e.g., Justice Marshall in Figure 1). We might conclude that he is an activist. But now suppose the Justice almost always votes with the majority when the Court upholds a law. Would we still deem him an “activist”? Probably not.

We could supply many more examples, but the larger point is that we should consider the fraction of votes to strike in all cases reviewing the constitutionality of laws. This would go some distance toward distinguishing between truly aggressive justices—those willing to strike regardless of whether the Court does—and those who are meeker, voting with the majority regardless of whether the majority strikes or upholds. 12By examining all cases, we are able to develop a clearer picture of judicial self-restraint. But there are other problems that our approach does not solve. E.g., litigants may choose to wait to bring constitutional challenges until there is a “friendly” Supreme Court, and through the certiorari process, Justices may deny difficult cases. (For more on this and other problems, see Epstein & Landes, supra note 2.) How these affect our findings, we are not sure, though we should keep in mind that it takes only one litigant to raise a constitutional challenge, and in a situation with circuit splits, one losing party will almost always find the Supreme Court “friendly.” Still, we can generalize the results of our study only to the set of cases that the Justices agreed to hear and decide.

Analyzing all constitutional review cases would help correct one of Figure 1’s flaws. But not the second, which traces to a failure to consider explanations other than a Justice’s taste for judicial self-restraint. 13 See, e.g., Lindquist et al., supra note 11, at 108; Lindquist & Solberg, supra note 5. That is, even if we observe some Justices regularly voting to strike and others to uphold across all cases, we would not want to make claims about their relative commitment to judicial self-restraint without examining other possible reasons for the pattern.

Of the many possibilities, 14See infra Part III for other factors. the Justices’ political preferences over the substantive policy embedded in the law has received considerable attention in the literature on judicial behavior. The idea is that left-leaning Justices tend to invalidate “conservative” laws. Vice versa for right-leaning Justices.

A number of studies present evidence supporting this hypothesis, 15 See, e.g., Lindquist & Cross, supra note 5; Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited 415–16 (2002); Lori A. Ringhand, Judicial Activism: An Empirical Examination of Voting Behavior on the Rehnquist Natural Court, 24 Const. Comment. 43, 55–58 (2007). and our data too suggest that it has some merit. In Figure 2, in the left-hand column, we show the fraction of votes to strike in all cases in which the Court considered the constitutionality of a conservative law. The right panel shows the same fraction for liberal laws. Following from the literature, 16 See, e.g., Lindquist & Cross, supra note 5; Segal & Spaeth, supra note 15. we measure the ideology of the law on the basis of the ideological direction of the Court’s decision. 17Suppose the Court invalidated a law on the ground that it discriminated against gays. We would code the Court’s decision as liberal and the law as conservative. Fortunately, though, we did not need to make these decisions because the U.S. Supreme Court Database codes the ideological direction of every Supreme Court decision since the 1946 Term. From the “decisionDirection” variable, we constructed the ideological direction of the law. The Database and its definitions of “liberal” and “conservative” are available at http://www.supremecourtdatabase.org/.

Figure 2. Voting to Invalidate a Federal, State, or Local Law in All Cases Reviewing the Constitutionality of a Law, 1969–2009 Terms 18Stefanie Lindquist provided citations to the cases in which the Court reviewed the constitutionality of a law during the Burger and Rehnquist Court years (1969–2004 Terms); we updated the case list following her protocols. See Lindquist & Cross, supra note 5; Lindquist & Solberg, supra note 5; Solberg & Lindquist, supra note 5. The Justices’ votes come from the “Justice-Centered Data: Cases Organized by Supreme Court Citation” version of the U.S. Supreme Court Database, available at http://www.supremecourtdatabase.org/. We exclude Justice Sotomayor (N = 18).

Figure 2

A clear pattern emerges. The conservative Justices (e.g., Alito, Rehnquist, Roberts, and Thomas) appear at the top of the left panel of the figure, meaning that they exercise “judicial restraint” when reviewing conservative laws; liberals (e.g., Black, Ginsburg, Brennan, and Souter), on the other hand, appear at the top of the right panel revealing a similar level of “restraint” over liberal laws.

Statistical analyses confirm this visual inspection. A regression of the fraction to invalidate conservative laws on the Justices’ ideology produces a statistically significant coefficient 19When we use the term “statistically significant,” we mean p ≤ .05. regardless of how we measure ideology. 20For this and all analyses to follow, we estimate the statistical models using three different measures of the Justices’ ideology: (1) their political party affiliation (Republican or Democrat), (2) the mean of their Martin–Quinn score (which is based on the Justices’ votes in non-unanimous cases), and (3) their Segal & Cover score. (1) is available in Lee Epstein et al., The Supreme Court Compendium: Data, Decisions, and Developments (5th ed. 2011). We computed (2) from the Term-by-Term Martin–Quinn scores. See Measures, Martin–Quinn Scores, http://www.mqscores.wustl.edu/measures.php (last visited July 5, 2012). For information on how these scores are computed, see Lee Epstein et al., Ideological Drift Among Supreme Court Justices: Who, When, and How Important?, 101 Nw. U. L. Rev. 1483, 1503 & n.88, 1527 n.156 (2007); and Andrew D. Martin & Kevin M. Quinn, Dynamic Ideal Point Estimation via Markov Chain Monte Carlo for the U.S. Supreme Court, 1953–1999, 10 Pol. Analysis 134 (2002). (3)—the Segal & Cover scores—are available online. See Perceived Qualifications and Ideology of Supreme Court Nominees, 1937–2005, Stony Brook U., http://www.stonybrook.edu/commcms/polisci/professors/qualtable.pdf (last visited July 5, 2012). They run from 0 (most conservative) to 1 (most liberal). Id. For details on how Segal & Cover calculate the scores, see Jeffrey A. Segal & Albert D. Cover, Ideological Values and the Votes of U.S. Supreme Court Justices, 83 Am. Pol. Sci. Rev. 557, 559–61 (1989).For purposes of exposition, we report only the Segal & Cover scores in the text. Using the two other scores does not produce substantively different results. For example, using Segal & Cover’s time-tested ideology scores (derived from an analysis of newspaper editorials written prior to the Senate’s confirmation vote), 21 See Segal & Cover, supra note 20. the likelihood of a very conservative Justice (e.g., a Scalia) voting to invalidate a conservative law is a modest .27; 22The 95% confidence interval is [.15, .38]. for a very liberal Justice (e.g., a Brennan), the predicted probability jumps to .76. 23The 95% confidence interval is [.62, .90]. Put another way, out of every ten conservative laws, the model expects conservative Justices to invalidate three and liberal Justices, seven. When the Court reviews a liberal law, the conservative Justices are more likely to strike (.46) than the liberals (.17). 24The 95% confidence interval for the conservative Justices is [.36, .56]. For the liberals, it is [.05, .29].

These regressions consider all the Justices simultaneously. Reestimating them for the Justices individually yields similar results. Of the twenty-one serving between the 1969 and 2009 Terms, eighteen exhibit aggressive (and meek) behavior that aligns with their ideology to a statistically significant degree. 25Again, we exclude Justice Sotomayor because she participated in only eighteen cases in which the Court reviewed the constitutionality of legislative action. Only Justices Harlan, O’Connor, and Kennedy do not fit the pattern—and O’Connor is borderline. 26 p = .10.

II. Modeling Judicial Activism and Self-Restraint

Figure 2 seems to provide a powerful demonstration of the effect of the Justices’ ideology (vis-à-vis the policy content of the law) on their votes to overturn and uphold legislation. Unless we make the heroic assumption that ideology is the only variable that affects votes to invalidate and uphold legislative action, however, it too is insufficient. Other factors likely to play a role center on (1) the Court’s case-selection process, (2) features of the law under review, and (3) the contemporaneous political environment. 27We derive these from previous studies. See, e.g., Epstein & Landes, supra note 2; Howard & Segal, supra note 5; Lindquist et al., supra note 15; Lindquist & Solberg, supra note 5; Clark & Whittington, supra note 5.

To attend to the selection process, we include three variables. 28Ideally, we would also model the selection process. See Lindquist & Solberg, supra note 5, at 76–77 (“[W]e note that the judicial review cases heard by the Supreme Court do not arise on the Court’s docket at random. Instead, the Court’s certiorari process is complex and often involves strategic calculations on the part of the justices. . . . [I]f court intervention is nonrandom, ignoring this selection process raises the likelihood that conclusions about the forces affecting subsequent votes to invalidate legislation will be inaccurate.” (citations omitted)); supra note 12. The first, Direction of the Lower Court’s Decision (coded 1 for liberal; 0 if otherwise), controls for the modern Supreme Court’s propensity to grant petitions to reverse the decision of the court below (i.e., even a very liberal majority will tend to reverse a decision below that upholds a liberal law, and vice versa for a conservative majority). 29Between the 1969 and 2009 Terms, the petitioning party won in 63% of the 4631 cases. This percentage was computed from the U.S. Supreme Court Database at http://www.supremecourtdatabase.org/, using orally argued cases and the “partyWinning” variable. Second, we include dummy variables for the three Chief Justice eras represented in the analysis: Burger (1969–1985 Terms), Rehnquist (1986–2004), and Roberts (2005–2009). 30Burger is the omitted era. These serve as rough proxies for changes in the procedures for processing cases. Finally, we incorporate a variable, Certiorari (1 if the case came on certiorari; 0 if otherwise), to control for the means by which the case arrived at the Court. As Whittington and Clark observe, when the Court has complete discretion over whether to hear a case (as on certiorari), the Justices are free to focus their attention only on those “that raise serious doubts about the constitutional validity of government policies.” 31Clark & Whittington, supra note 5, at 21; accord Colker & Scott, supra note 5, at 1325 (“[T]he Court has considerable control over its docket and therefore tends to accept certiorari on cases in which it desires to reverse the lower court.”). When the Court is more constrained, the challenge to the legislative action may be relatively weak. (Alternatively, to assess the robustness of our results, we include a variable for before and after passage of the Supreme Court Case Selections Act of 1988. 32Pub. L. No. 100-352, § 3, 102 Stat. 662, 662 (codified at 28 U.S.C. § 1257 (2006)) (eliminating appeals as a matter of right from state court decisions and substituting certiorari). )

Turning to (2), the law under review, three relevant variables come to mind: whether the U.S. Congress enacted it, its substantive coverage, and its age. All else being equal, we expect the Court to be less likely to strike federal laws (= 1) relative to legislative action in the states or localities (= 0). In part, this expectation reflects the claim that the Justices should or would be wary of invalidating a law enacted by a majority of representatives from all fifty states unless they are reasonably sure about its unconstitutionality. 33Even Oliver Wendell Holmes observed, “I do not think the United States would come to an end if we lost our power to declare an Act of Congress void. I do think the Union would be imperiled if we could not make that declaration as to the laws of the several States.” Oliver Wendell Holmes, Law and the Court, Speech at a Dinner of the Harvard Law School Association of New York (Feb. 15, 1913), in Collected Legal Papers 291, 295–96 (1920). This expectation also accounts for the fact that the Office of the Solicitor General (SG) usually defends federal laws in the Supreme Court. Relative to the SG—a very effective and successful legal team by almost any measure—state and local attorneys are more of a mixed bag. 34At the least, they are not the “repeat player” that the federal government is. For studies on the SG, see, e.g., Michael A. Bailey et al., Signals from the Tenth Justice: The Political Role of the Solicitor General in Supreme Court Decision Making, 49 Am. J. Pol. Sci. 72 (2005); Chris Nicholson & Paul M. Collins, Jr., The Solicitor General’s Amicus Curiae Strategies in the Supreme Court, 36 Am. Pol. Res. 382 (2008); and Jeffrey A. Segal, Amicus Curiae Briefs by the Solicitor General During the Warren and Burger Courts, 41 W. Pol. Q. 135 (1988). Of course, the SG can file an amicus curiae brief in cases involving the constitutionality of state and local legislative action, and we attend to this with two variables (whether the SG supported the law or supported invalidation).

The type of statute also may be important because, under existing precedent, laws centering on rights and liberties (coded 1) are presumably subjected to more rigorous standards of review than are economic statutes (coded 0). Finally, we control for the age of the law at the time of the Court’s decision. 35We thank Stefanie Lindquist for supplying this information for the 1969–2004 Terms. We updated her data through the 2009 Term. See supra note 18. Although debatable, Dahl’s famous study suggests that older and less important laws are riper for invalidation than newer statutes. 36Dahl, supra note 5; accord Lindquist et al., supra note 11. The mechanism, on Dahl’s account, is Article II’s command: “[The President] shall nominate, and by and with the Advice and Consent of the Senate, shall appoint . . . Judges of the supreme Court.” 37 U.S. Const. art. II, § 2, cl. 2. Assuming the President and Senate have regular opportunities to appoint new Justices, “the policy views dominant on the Court are never for long out of line with the policy views dominant among the lawmaking majorities of the United States.” 38Dahl, supra note 5, at 285.

Many scholars agree that the existing regime has a role to play in judicial review, but they take issue with Dahl’s mechanism. To them (and us), the mechanism is not the appointment process but rather the Justices’ institutional interest in avoiding collisions with contemporaneous political actors. 39 E.g., Lee Epstein et al., The Supreme Court as a Strategic National Policymaker, 50 Emory L.J. 583 (2001); Anna Harvey & Barry Friedman, Ducking Trouble: Congressionally Induced Selection Bias in the Supreme Court’s Agenda, 71 J. Pol. 574 (2009). For the most conclusive evidence to date of the effect of Congress on the Court’s interpretation of the Constitution, see Jeffrey A. Segal et al., Congress, the Supreme Court, and Judicial Review: Testing a Constitutional Separation of Powers Model, 55 Am. J. Pol. Sci. 89 (2011). When the Court invalidates laws that Congress, the President, and, presumably, the public desire, it runs the risk of backlash in the form of jurisdiction-stripping laws, budget slashing, or worse. 40For a list of weapons Congress and the President have at their disposal in the constitutional context, see Gerald N. Rosenberg, Judicial Independence and the Reality of Political Power, 54 Rev. Pol. 369, 376–77 (1992); and Epstein et al., supra note 39, at 597–601. To capture this take on Dahl’s account, 41Dahl’s story is about national institutions, while our study includes constitutional review of state and local laws. Nonetheless, we focus on the nation’s mood and federal institutions because, we might speculate, a state/local law well outside the bounds of constitutionality and opposed by the national governing coalition would be unlikely to reach the Court. Moreover, most state/local laws that do reach the Court may well implicate national political interests (or, at the least, have implications beyond the law in question) given Supreme Court Rule 10 (which emphasizes “important” questions). See Sup. Ct. R. 10. we include variables designed to measure the public’s mood (from very liberal to very conservative) 42To capture the public’s mood, we use Stimson’s measure. Stimson computes the measure by analyzing survey responses to a number of questions that tap an underlying liberal–conservative dimension. See James A. Stimson, Public Opinion in America: Moods, Cycles, and Swings (2d ed. 1999); Policy Mood, U. N.C. Chapel Hill, http://www.unc.edu/~cogginse/Policy_Mood.html (last visited July 5, 2012). He provides a dynamic factor analysis model that combines these responses in a principled fashion to yield a quarterly and annual measure of public mood as a single score. See Stimson, supra; Policy Mood, supra. and the partisan composition of the elected branches. 43We added variables to indicate whether the government was unified Republican or unified Democratic. For more sophisticated approaches, see Segal et al., supra note 39.

To summarize, we expect four sets of factors to influence the Justices’ exercise of judicial self-restraint: (1) the Justices’ ideology vis-à-vis the law, (2) the case-selection process, (3) the type of legislative enactment, and (4) the political context. We describe each in Table 1.

Table 1. Description of Basic Variables

Variable and Coding

Mean

(Standard Deviation)

Dependent Variables

Vote to Invalidate (= 1) or Not (= 0) if Legislative Action Is Liberal

0.338

0.473

Vote to Invalidate (= 1) or Not (= 0) if Legislative Action Is Conservative

0.506

0.500

Independent Variables

Justice’s Ideology (most conservative = 0; most liberal = 1) 44In Table 2, we report the Segal & Cover score. We also estimated the models using the Justices’ partisan affiliation and their Martin–Quinn scores. See supra note 20.

0.402 0.332

Direction of the Lower Court’s Decision (liberal = 1; conservative = 0)

0.543

0.498

Case Came on Certiorari (= 1) or Not (= 0)

0.511

0.500

Federal Law (= 1) or Not (= 0)

0.289

0.454

SG Conservative Amicus (= 1) or Not (= 0)

0.100

0.300

SG Liberal Amicus (= 1) or Not (= 0)

0.055

0.228

Civil Liberties Law (= 1) or Not (= 0)

0.772 0.420

Age of Law (at time of the decision)

15.096 19.344

Unified Democratic Government (Congress and the President) (= 1) or Not (= 0)

0.203 0.403

Unified Republican Government (Congress and the President) (= 1) or Not (= 0)

0.076 0.264

Public’s Mood (lower numbers are more conservative)

56.507

3.442

Note: Our dataset contains 7415 votes (5295 over conservative laws and 2040 for liberal laws) cast in 842 cases. The N for the following variables is smaller: (1) Age of Law (7352) because we could not identify the age in seven cases; (2) Direction of the Lower Court Decision (7299) because it is not specified in the Supreme Court Database in thirteen cases; and (3) Public’s Mood (7326) because it is not available for cases argued in 2010.

III. Results: Justices Combined

We begin our assessment of the impact of the variables in Table 1 by modeling the votes (to uphold or invalidate) of all twenty-two Justices (including Sotomayor) in two basic equations: one for liberal laws and the other for conservative laws. 45We estimated them with and without the mood measure because it is unavailable for cases argued in 2010. Also, to check for robustness, we estimated each model using a random-effects logistic regression under the assumption that there may be some unaccounted factors that affect each Justice’s propensity to strike the legislative action. Including a Justice-specific random effect, which allows each Justice to have her own baseline, helps to attend to these unmodeled factors. As the note to Table 2 indicates, it does not matter whether we estimate the models using random-effects regression or logistic regression (clustered on the Justice). Ideology is significant in all the models. (In the next Part, we also estimate the two basic equations but examine the Justices individually.)

Table 2 displays the results. Note, first, that across all four models several variables help account for the Justices’ votes to strike or uphold. For example, when the Court reviews the constitutionality of a federal law, it is more likely to uphold it relative to state or local legislative action. For liberal statutes, eq. 3 predicts that the Court sustains about seven out of every ten laws passed by Congress but only six out of every ten enacted in the states or localities. 46With all other variables set at their mean or mode, the predicted probability of upholding a state/local law is .61 [.54, .68]; for federal laws, the figure is .71 [.61, .81]. To calculate the predicted probabilities, we used eq. 3. For right-leaning statutes (eq. 4), the Justices are more likely to invalidate state and local policies than to uphold them (.53 [.49, .57] predicted probability to strike). Not so of conservative congressional laws: the predicted probability of invalidation is only .32. 47We used eq. 4 with all other variables set at their mean or mode. The 95% confidence interval is [.28, .36].

Table 2. Logistic Regressions of the Justices’ Votes to Invalidate (= 1) or Uphold (= 0) Legislative Action, 1969–2009 Terms

With Public Mood

Without Public Mood

Variable

Liberal Law

(1)

Conservative Law

(2)

Liberal Law

(3)

Conservative Law

(4)

Justice’s Ideology

-1.388*

(0.262)

2.273*

(0.440)

-1.450*

(0.250)

2.229*

(0.442)

Liberal Lower Court

0.623*

(0.106)

-0.270*

(0.049)

0.496*

(0.119)

-0.269*

(0.049)

Certiorari

-0.015

(0.159)

0.016

(0.105)

-0.023

(0.175)

0.019

(0.106)

Federal Law

-0.423*

(0.162)

-0.916*

(0.089)

-0.448*

(0.157)

-0.885*

(0.092)

Dummy for Rehnquist Era

0.677*

(0.235)

0.295*

(0.147)

-0.236

(0.151)

0.264*

(0.124)

Dummy for Roberts Era

0.733

(0.405)

-0.269

(0.289)

-0.233*

(0.279)

-0.394

(0.211)

Conservative SG Amicus

0.434*

(0.165)

-0.832*

(0.121)

0.484*

(0.166)

-0.811*

(0.122)

Liberal SG Amicus

-0.478*

(0.137)

0.645*

(0.225)

-0.426*

(0.138)

0.661*

(0.228)

Civil Liberties Law

-0.023

(0.253)

0.142

(0.132)

-0.053

(0.242)

0.141

(0.137)

Age of Law

-0.000

(0.003)

-0.004*

(0.001)

-0.001

(0.003)

-0.004*

(0.001)

Unified Democratic

-0.251

(0.130)

-0.002

(0.101)

0.217*

(0.100)

0.022

(0.104)

Unified Republican

-0.705*

(0.204)

0.057

(0.167)

-0.449

(0.239)

0.091

(0.188)

Public Mood

-0.162*

(0.040)

-0.009

(0.019)

_____

_____

Constant

8.416*

(2.078)

-0.110

(1.091)

0.395

(0.178)

-0.817*

(0.309)

N of Votes

1977

5170

2013

5223

Log Likelihood

-1159.084

-3129.070

-1205.065

-3167.552

Notes:

(1) p ≤ .05. Variables statistically significant (at p ≤ .05) and correctly signed across all models are in italics. Standard errors are in parentheses. For all four models, we clustered on Justice.

(2) Includes all twenty-two Justices serving between the 1969 and 2009 Terms.

(3) We present models with (1 & 2) and without (3 & 4) Public Mood because mood data are available only for cases argued through 2009.

(4) We conducted two types of robustness checks: (1) we substituted two other measures for the Segal & Cover scores; and (2) we estimated the models using random-effects logistic regression, with the “panel” set to Justice. 48 See supra note 20. Neither produced substantively different results.

We do not know whether the preference for federal over state and local laws reflects extreme reluctance to invalidate action taken by the U.S. Congress or the expertise of the SG. We can report, though, that the presence of the SG as an amicus curiae exerts a statistically significant effect on the Court’s willingness to strike or uphold state/local policies. Suppose the Justices are reviewing a liberal state (or local) law, and the SG files in support. The odds of the Court upholding the law are relatively high (.71 [.64, .77]). 49We used eq. 3 with all other variables set at their mean or mode. But should the SG argue for invalidation, the odds of upholding fall to about 50–50. 50.49 [.39, .59]. We see a similar, though even more dramatic, effect for conservative state/local laws. The probability flips from .67 to uphold when the SG files in support 51We used eq. 4 with all other variables set at their mean or mode. The 95% confidence interval is [.54, .79]. to .69 to invalidate when the SG files in opposition. 52The 95% confidence interval is [.56, .81].

No doubt, then, a host of factors affect the Justices’ votes. And yet, the basic ideological effect unearthed in Figure 2 remains. Regardless of how we specify the models in Table 2, we cannot wipe out the central finding that conservative Justices are significantly more likely to strike liberal statutes and liberals, conservative laws. 53 See supra notes accompanying Table 2. The statistical effect is that persistent and that consistent. It is also quite strong, as Figure 3 shows.

Figure 3. Predicted Probability of Invalidating Legislative Action Across the Range of Ideology

Figure 3

Notes:

  • The panel on the left is based on eq. 3 in Table 2. The panel on the right is from eq. 4. All other variables are set at their mean or mode.

  • The solid line shows the predicted probability; the dashed lines show the upper and lower bounds of the 95% confidence interval.

Starting with the left panel, the probability of invalidating a liberal law decreases from .52 to an improbable .20 as we move from the most conservative to most liberal Justices, all else being equal. As for conservative laws, the probability declines even more dramatically, from a .81 invalidation prediction for the most liberal Justices to .31 for the extreme conservatives.

Does this suggest that the liberal Justices are more “strike happy” (i.e., more activist) than the conservatives? Perhaps. 54To reiterate, for liberal laws, extreme conservatives invalidate about 52% of the time, compared to 81% for extreme liberals. But the comparison assumes an equal number of opportunities to invalidate conservative and liberal laws—an assumption that the data do not meet. As the totals in Table 2 indicate, between the 1969 and 2009 Terms, the Court reviewed the constitutionality of conservative laws nearly three times as often as liberal laws. 55Although we can only speculate on why this is the case, the answer must come from either the supply side or the demand side—the supply of legal provisions to review or the demand to review by the Justices. On the supply side, much of the existing law was conservative, especially in the domains of civil rights and civil liberties. On the demand side, a Court with a liberal majority, at least until the presidency of George H.W. Bush, would typically choose to review conservative laws, all else being equal. On this logic, we would expect the Roberts Court to consider the constitutionality of a greater number of liberal legal provisions than conservative ones.

Finally, and of special relevance to our claim about the Roberts Justices, the effect of ideology is relatively consistent across all three Chief Justice eras. Variations on all the regressions in Table 2 yield significant coefficients on the Justice’s Ideology variable when we estimate them separately by Chief Justice era. 56 See infra notes accompanying Figure 4. Figure 4 displays the predictions.

Figure 4. Predicted Probability of Invalidating Legislative Action Across the Range of Ideology, by Chief Justice Era

4a. Liberal Law

Figure 4a

4b. Conservative Law

Figure 4b

Notes:

Looking across the panels, a similar pattern emerges. Conservatives are far more likely to strike liberal laws, while the liberals are far more likely to uphold them. The same holds in reverse for conservative laws.

And it generally holds for the Burger, Rehnquist, and Roberts Justices, though some potentially intriguing trends emerge. Turning first to the left wing of the Court, a comparison of the two panels in Figure 4 shows that the liberals have grown less willing to strike liberal laws, from a predicted probability of .35 during the Burger years, to .23 during the Rehnquist Court, to .06 for the Roberts years. But they are also less likely to strike laws with conservative policy content (the predicted probability declines from .78 in the Rehnquist era to .62 in the Roberts era). The right wing of the Court, on the other hand, has grown somewhat more opportunistically restraintist (activist). The conservative Justices are less willing to invalidate conservative laws, from a predicted probability of .36 during the Rehnquist years to .15 for the Roberts era. At the same time, as Figure 4b shows, their willingness to strike laws has increased slightly, from .55 (Rehnquist Court) to .58 (Roberts Court).

IV. Results: Individual Justices

Whether these are trends or blips due to data sparseness for the Roberts Court, we cannot say. Moreover, even if they are trends, we do not know if they reflect a growing tendency toward judicial self-restraint on the part of the liberals (and opportunistic restraint on the part of conservatives) or unmodeled features of the case-selection process.

What we can do is explore how many of the individual Justices conform to the basic patterns unearthed in Table 2 and Figure 4. We do so by estimating separate regression models for each of the nineteen Justices with fifty or more case participations. Table 3 displays the results. 57This eliminates Justices Black (N = 37), Harlan (N = 44), and Sotomayor (N = 18).

Table 3. Predicted Probability of Invalidating Legislative Action for Nineteen Justices with More Than Fifty Case Participations, 1969–2009 Terms

Justice (Order from Most Conservative to Most Liberal 58We used the mean of the Martin–Quinn scores. ) (N)

Coefficient on “Law” (1 = liberal; 0 = otherwise)

Predicted Probability of Striking a Conservative Law

Predicted Probability of Striking a Liberal Law

Thomas (247)

1.847*

(0.329)

0.20

[0.13, 0.28]

0.62

[0.47, 0.76]

Rehnquist (682)

1.649*

(0.241)

0.17

[0.13, 0.22]

0.51

[0.40, 0.63]

Scalia (362)

1.416*

(0.271)

0.27

[0.21, 0.35]

0.61

[0.47, 0.72]

Burger (458)

0.674*

(0.289)

0.36

[0.29, 0.44]

0.53

[0.38, 0.67]

Alito (59)

5.277*

(1.769)

0.02

[0.00-0.09]

0.54

[0.10-0.94]

Roberts (65)

1.509*

(0.699)

0.17

[0.06-0.37]

0.46

[0.16-0.79]

Powell (415)

-0.761*

(0.287)

0.50

[0.42, 0.59]

0.32

[0.20, 0.45]

O’Connor (415)

0.382

(0.240)

____

____

Kennedy (323)

0.252

(0.263)

____

____

Stewart (336)

-0.499

(0.355)

____

____

White (606)

-0.838*

(0.255)

0.46

[0.39, 0.53]

0.27

[0.17, 0.39]

Blackmun (600)

-1.478*

(0.263)

0.62

[0.55, 0.69]

0.28

[0.19, 0.39]

Souter (241)

-2.237*

[0.355]

0.71

[0.60, 0.80]

0.22

[0.11, 0.35]

Breyer (200)

-1.838*

[0.382]

0.53

[0.41, 0.65]

0.16

[0.07, 0.29]

Ginsburg (206)

-2.309*

[0.368]

0.67

[0.54, 0.78]

0.17

[0.08, 0.30]

Stevens (636)

-1.882*

[0.229]

0.72

[0.66, 0.77]

0.28

[0.20, 0.38]

Brennan (550)

-3.530*

(0.360)

0.89 [0.84, 0.92]

0.20

[0.11, 0.32]

Marshall (561)

-3.343*

(0.324)

0.89

[0.85, 0.93]

0.23[0.14, 0.36]

Douglas (158)

-3.573*

(0.772)

0.90 [0.80, 0.96]

0.23

[0.05, 0.52]

Notes:

Even after controlling for other factors that might influence votes to invalidate or sustain legislation, ideology exerts a significant effect for sixteen of the nineteen Justices (84%). The five most conservative Justices in our dataset (Thomas, Rehnquist, Scalia, Burger, and Alito) are all highly unlikely to invalidate conservative laws. The range in the predicted probabilities of invalidating is narrow, from .36 for Burger to .02 for Alito. The five most liberal Justices (Douglas, Marshall, Brennan, Stevens, and Ginsburg), on the other hand, are highly unlikely to sustain conservative laws. The model predicts that Ginsburg, the most likely of the five to uphold, votes to strike in seven out of every ten cases; Douglas, the least likely, strikes in nine out of ten. The story is much the same for left-leaning laws, only in the reverse. Moving down the third column (from the most conservative to most liberal Justice), the predicted probability of invalidating a liberal law declines in a near monotonic fashion.

What of the Roberts Justices? Though it may be too soon to say much about Alito and Roberts (their Ns are small and the confidence intervals around the predictions, large), they conform to the basic ideological pattern. Both are significantly more likely to uphold conservative laws and invalidate left-leaning policy. The ideological effect is even starker for the more extreme conservatives, Scalia and Thomas. As for the four left-leaning Justices in our dataset (Souter, Breyer, Ginsburg, and Stevens), they too evince a different approach to judicial review, based on the underlying policy content of the law.

The Court’s “super median,” 60Lee Epstein & Tonja Jacobi, Super Medians, 61 Stan. L. Rev. 37 (2008). Justice Kennedy, is more of a puzzle. Judged on the basis of whether he was in the majority when the Court struck down legislative action, Kennedy is the most aggressive of the Roberts Justices. 61And, indeed, Kennedy is the most aggressive Justice in our entire dataset. See Figure 1. When the Court invalidated a federal, state, or local law between the 2005–2009 Terms, Kennedy was in the majority 94% of the time. Only Roberts (at 72%) came close. But, unlike the other Roberts Justices, no underlying ideological pattern seems to exist to Kennedy’s votes; as a statistical matter, he is equally as likely to sustain (or invalidate) conservative as liberal laws.

If, however, we consider only laws reviewed by the Court since the start of the Roberts Court years, even Kennedy’s ideology shows in his votes—he too is now substantially more likely to strike liberal laws than conservative laws. 62He is significantly (p = .03) more likely to invalidate liberal state/local policies, though not federal laws. Modeling all laws in one equation (with a variable indicating whether the law was federal or state/local) yields a p-value of .07 on the Ideology of the Law variable. The p-value for all Terms prior to 2004 is not close to statistically significant (.84).

Along these lines, Figure 5 is illuminating. There we compare the predicted probabilities of invalidating laws for Justices Ginsburg, Kennedy, and Alito since the 2005 Term. All else being equal, Kennedy is not as extreme as either Roberts or Breyer, but the difference in his willingness to invalidate liberal versus conservative law is substantial: 38% versus 16%.

Figure 5. Predicted Probability of Invalidating a Law for Justices Ginsburg, Kennedy, and Alito, 2005–2009 Terms

Figure 5

Note: We generated the predictions using the same models in Table 3, except we limited them to the Roberts Court years.

Conclusion

Ours will certainly not be the last word on judicial review during the Roberts years. The data we have amassed cover only the first five Terms of the current era. As Justices come and go, the Court may move further to the right, turn sharply to the left, or perhaps steer a more moderate course. Whether the underlying patterns we have uncovered will dissipate remains another story. At least since 1969, 63And maybe well before. See Epstein & Landes, supra note 2; Clark & Whittington, supra note 5. there is a clear ideological structure to the Court’s review of the constitutionality of federal, state, and local laws—the Roberts Justices not excepted.

Footnotes

Provost Professor of Law and Political Science and the Rader Family Trustee Chair in Law at the University of Southern California (http://epstein.usc.edu/). This Article was prepared for the 2011 Randolph W. Thrower Symposium, Emory University School of Law, February 1, 2011.Andrew D. Martin is Professor of Law and Political Science at Washington University in St. Louis.We thank the National Science Foundation and the Center for Empirical Research in the Law at Washington University for supporting our work on the U.S. Supreme Court. We also thank Stefanie Lindquist and Chad Westerland for supplying data on cases reviewing the constitutionality of legislative action.The dataset we use in this Article is available at http://epstein.usc.edu/research/RobertsActivism.html.

1Richard A. Posner writes that at least three meanings can be assigned to the term “judicial self-restraint”:(1) [J]udges apply law, they don’t make it (call this “legalism” . . . or, better, “the law made me do it”); (2) judges defer, to a very great extent, to decisions by other officials—appellate judges defer to trial judges and administrative agencies, and all judges to legislative and executive decisions (call this “modesty,” or “institutional competence,” or “process jurisprudence”); (3) judges are highly reluctant to declare legislative or executive action unconstitutional—deference is at its zenith when action is challenged as unconstitutional (call this “constitutional restraint”).Richard A. Posner, The Rise and Fall of Judicial Self-Restraint, 100 Calif. L. Rev. 519, 521 (2012). Judge Posner focuses on (3), as do we. But, unlike Judge Posner, we limit our analysis to legislative action.

2This caveat is important because Epstein and Landes, analyzing data for a far longer time span (1937–2009 Terms) but studying federal laws only, reach a somewhat different conclusion. Lee Epstein & William M. Landes, Was There Ever Such a Thing as Judicial Self-Restraint?, 100 Calif. L. Rev. 557 (2012). They find that the majority of Justices appointed before the early 1950s were reluctant to strike down laws regardless of whether the law accorded with their ideological values. Id. at 569–76. But for Justices appointed since 1952, Epstein & Landes’s findings parallel ours: the vast majority were opportunistic restraintists (activists), willing to uphold laws that were consistent with their policy preferences and strike those that were not. Id. at 573–77.

3 See infra Part IV.

4 See infra Part V.

5Recent examples include Stefanie A. Lindquist & Frank B. Cross, Measuring Judicial Activism (2009); Ruth Colker & Kevin M. Scott, Dissing States?: Invalidation of State Action During the Rehnquist Era, 88 Va. L. Rev. 1301 (2002); Robert M. Howard & Jeffrey A. Segal, A Preference for Deference? The Supreme Court and Judicial Review, 57 Pol. Res. Q. 131 (2004); Stefanie A. Lindquist & Rorie Spill Solberg, Judicial Review by the Burger and Rehnquist Courts: Explaining Justices’ Responses to Constitutional Challenges, 60 Pol. Res. Q. 71 (2007); Rorie Spill Solberg & Stefanie A. Lindquist, Activism, Ideology, and Federalism: Judicial Behavior in Constitutional Challenges Before the Rehnquist Court, 1986–2000, 3 J. Empirical Legal Stud. 237 (2006); David L. Weiden, Judicial Politicization, Ideology, and Activism at the High Courts of the United States, Canada, and Australia, 64 Pol. Res. Q. 335 (2011); and Tom S. Clark & Keith E. Whittington, Ideology, Partisanship, and Judicial Review of Acts of Congress, 1789–2006 (May 22, 2009) (unpublished manuscript) (on file with authors), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1475660. Classic empirical studies of judicial invalidation of legislative action include Jonathan D. Casper, The Supreme Court and National Policy Making, 70 Am. Pol. Sci. Rev. 50 (1976); Robert A. Dahl, Decision-Making in a Democracy: The Supreme Court as a National Policy-Maker, 6 J. Pub. L. 279 (1957); and John B. Gates, Partisan Realignment, Unconstitutional State Policies, and the U.S. Supreme Court, 1837–1964, 31 Am. J. Pol. Sci. 259 (1987).

6By this, we mean displays or lists showing the fraction of cases in which the Justices were in the majority when the Court invalidated a (federal, state, or local) law. Examples, in diverse outlets, include Lee Epstein & Thomas G. Walker, Constitutional Law for a Changing America: Institutional Powers and Constraints 37 (7th ed. 2011); Thomas M. Keck, The Most Activist Supreme Court in History: The Road to Modern Judicial Conservatism 202 tbl.6.2 (2004); and Paul Gewirtz & Chad Golder, So Who Are the Activists?, N.Y. Times, July 6, 2005, at A19.

7Data are from the “Justice-Centered Data: Cases Organized by Supreme Court Citation” version of the U.S. Supreme Court Database at http://www.supremecourtdatabase.org/ and include only orally argued cases (decisionType = 1, 6, or 7). We exclude Justice Sotomayor, who participated in only five cases in which the Court invalidated a law.

8 See, e.g., Gewirtz & Golder, supra note 6 (“One conclusion our data suggests is that those justices often considered more ‘liberal’—Justices Breyer, Ruth Bader Ginsburg, David Souter and John Paul Stevens—vote least frequently to overturn Congressional statutes, while those often labeled ‘conservative’ vote more frequently to do so. At least by this measure (others are possible, of course), the latter group is the most activist.”).

9And, in fact, the correlation between the fraction and the Justices’ relative liberalism is between .33 and .57 (depending on how we measure the Justices’ ideology), suggesting that liberal Justices are more activist. For the reasons we offer in the text, however, we should not make much (if anything) of this correlation.

10 See, e.g., Clark & Whittington, supra note 5, at 13 (“Neither social science analyses nor narrative histories of the Court and judicial review give much attention to the cases in which the Court upholds legislation . . . .”).

11 See Stefanie A. Lindquist et al., The Rhetoric of Restraint and the Ideology of Activism, 24 Const. Comment. 103, 108 (2007); Nicholas S. Zeppos, Deference to Political Decisionmakers and the Preferred Scope of Judicial Review, 88 Nw. U. L. Rev. 296, 306 (1993); Clark & Whittington, supra note 5, at 8.

12By examining all cases, we are able to develop a clearer picture of judicial self-restraint. But there are other problems that our approach does not solve. E.g., litigants may choose to wait to bring constitutional challenges until there is a “friendly” Supreme Court, and through the certiorari process, Justices may deny difficult cases. (For more on this and other problems, see Epstein & Landes, supra note 2.) How these affect our findings, we are not sure, though we should keep in mind that it takes only one litigant to raise a constitutional challenge, and in a situation with circuit splits, one losing party will almost always find the Supreme Court “friendly.” Still, we can generalize the results of our study only to the set of cases that the Justices agreed to hear and decide.

13 See, e.g., Lindquist et al., supra note 11, at 108; Lindquist & Solberg, supra note 5.

14See infra Part III for other factors.

15 See, e.g., Lindquist & Cross, supra note 5; Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited 415–16 (2002); Lori A. Ringhand, Judicial Activism: An Empirical Examination of Voting Behavior on the Rehnquist Natural Court, 24 Const. Comment. 43, 55–58 (2007).

16 See, e.g., Lindquist & Cross, supra note 5; Segal & Spaeth, supra note 15.

17Suppose the Court invalidated a law on the ground that it discriminated against gays. We would code the Court’s decision as liberal and the law as conservative. Fortunately, though, we did not need to make these decisions because the U.S. Supreme Court Database codes the ideological direction of every Supreme Court decision since the 1946 Term. From the “decisionDirection” variable, we constructed the ideological direction of the law. The Database and its definitions of “liberal” and “conservative” are available at http://www.supremecourtdatabase.org/.

18Stefanie Lindquist provided citations to the cases in which the Court reviewed the constitutionality of a law during the Burger and Rehnquist Court years (1969–2004 Terms); we updated the case list following her protocols. See Lindquist & Cross, supra note 5; Lindquist & Solberg, supra note 5; Solberg & Lindquist, supra note 5. The Justices’ votes come from the “Justice-Centered Data: Cases Organized by Supreme Court Citation” version of the U.S. Supreme Court Database, available at http://www.supremecourtdatabase.org/. We exclude Justice Sotomayor (N = 18).

19When we use the term “statistically significant,” we mean p ≤ .05.

20For this and all analyses to follow, we estimate the statistical models using three different measures of the Justices’ ideology: (1) their political party affiliation (Republican or Democrat), (2) the mean of their Martin–Quinn score (which is based on the Justices’ votes in non-unanimous cases), and (3) their Segal & Cover score. (1) is available in Lee Epstein et al., The Supreme Court Compendium: Data, Decisions, and Developments (5th ed. 2011). We computed (2) from the Term-by-Term Martin–Quinn scores. See Measures, Martin–Quinn Scores, http://www.mqscores.wustl.edu/measures.php (last visited July 5, 2012). For information on how these scores are computed, see Lee Epstein et al., Ideological Drift Among Supreme Court Justices: Who, When, and How Important?, 101 Nw. U. L. Rev. 1483, 1503 & n.88, 1527 n.156 (2007); and Andrew D. Martin & Kevin M. Quinn, Dynamic Ideal Point Estimation via Markov Chain Monte Carlo for the U.S. Supreme Court, 1953–1999, 10 Pol. Analysis 134 (2002). (3)—the Segal & Cover scores—are available online. See Perceived Qualifications and Ideology of Supreme Court Nominees, 1937–2005, Stony Brook U., http://www.stonybrook.edu/commcms/polisci/professors/qualtable.pdf (last visited July 5, 2012). They run from 0 (most conservative) to 1 (most liberal). Id. For details on how Segal & Cover calculate the scores, see Jeffrey A. Segal & Albert D. Cover, Ideological Values and the Votes of U.S. Supreme Court Justices, 83 Am. Pol. Sci. Rev. 557, 559–61 (1989).For purposes of exposition, we report only the Segal & Cover scores in the text. Using the two other scores does not produce substantively different results.

21 See Segal & Cover, supra note 20.

22The 95% confidence interval is [.15, .38].

23The 95% confidence interval is [.62, .90].

24The 95% confidence interval for the conservative Justices is [.36, .56]. For the liberals, it is [.05, .29].

25Again, we exclude Justice Sotomayor because she participated in only eighteen cases in which the Court reviewed the constitutionality of legislative action.

26 p = .10.

27We derive these from previous studies. See, e.g., Epstein & Landes, supra note 2; Howard & Segal, supra note 5; Lindquist et al., supra note 15; Lindquist & Solberg, supra note 5; Clark & Whittington, supra note 5.

28Ideally, we would also model the selection process. See Lindquist & Solberg, supra note 5, at 76–77 (“[W]e note that the judicial review cases heard by the Supreme Court do not arise on the Court’s docket at random. Instead, the Court’s certiorari process is complex and often involves strategic calculations on the part of the justices. . . . [I]f court intervention is nonrandom, ignoring this selection process raises the likelihood that conclusions about the forces affecting subsequent votes to invalidate legislation will be inaccurate.” (citations omitted)); supra note 12.

29Between the 1969 and 2009 Terms, the petitioning party won in 63% of the 4631 cases. This percentage was computed from the U.S. Supreme Court Database at http://www.supremecourtdatabase.org/, using orally argued cases and the “partyWinning” variable.

30Burger is the omitted era.

31Clark & Whittington, supra note 5, at 21; accord Colker & Scott, supra note 5, at 1325 (“[T]he Court has considerable control over its docket and therefore tends to accept certiorari on cases in which it desires to reverse the lower court.”).

32Pub. L. No. 100-352, § 3, 102 Stat. 662, 662 (codified at 28 U.S.C. § 1257 (2006)) (eliminating appeals as a matter of right from state court decisions and substituting certiorari).

33Even Oliver Wendell Holmes observed, “I do not think the United States would come to an end if we lost our power to declare an Act of Congress void. I do think the Union would be imperiled if we could not make that declaration as to the laws of the several States.” Oliver Wendell Holmes, Law and the Court, Speech at a Dinner of the Harvard Law School Association of New York (Feb. 15, 1913), in Collected Legal Papers 291, 295–96 (1920).

34At the least, they are not the “repeat player” that the federal government is. For studies on the SG, see, e.g., Michael A. Bailey et al., Signals from the Tenth Justice: The Political Role of the Solicitor General in Supreme Court Decision Making, 49 Am. J. Pol. Sci. 72 (2005); Chris Nicholson & Paul M. Collins, Jr., The Solicitor General’s Amicus Curiae Strategies in the Supreme Court, 36 Am. Pol. Res. 382 (2008); and Jeffrey A. Segal, Amicus Curiae Briefs by the Solicitor General During the Warren and Burger Courts, 41 W. Pol. Q. 135 (1988).

35We thank Stefanie Lindquist for supplying this information for the 1969–2004 Terms. We updated her data through the 2009 Term. See supra note 18.

36Dahl, supra note 5; accord Lindquist et al., supra note 11.

37 U.S. Const. art. II, § 2, cl. 2.

38Dahl, supra note 5, at 285.

39 E.g., Lee Epstein et al., The Supreme Court as a Strategic National Policymaker, 50 Emory L.J. 583 (2001); Anna Harvey & Barry Friedman, Ducking Trouble: Congressionally Induced Selection Bias in the Supreme Court’s Agenda, 71 J. Pol. 574 (2009). For the most conclusive evidence to date of the effect of Congress on the Court’s interpretation of the Constitution, see Jeffrey A. Segal et al., Congress, the Supreme Court, and Judicial Review: Testing a Constitutional Separation of Powers Model, 55 Am. J. Pol. Sci. 89 (2011).

40For a list of weapons Congress and the President have at their disposal in the constitutional context, see Gerald N. Rosenberg, Judicial Independence and the Reality of Political Power, 54 Rev. Pol. 369, 376–77 (1992); and Epstein et al., supra note 39, at 597–601.

41Dahl’s story is about national institutions, while our study includes constitutional review of state and local laws. Nonetheless, we focus on the nation’s mood and federal institutions because, we might speculate, a state/local law well outside the bounds of constitutionality and opposed by the national governing coalition would be unlikely to reach the Court. Moreover, most state/local laws that do reach the Court may well implicate national political interests (or, at the least, have implications beyond the law in question) given Supreme Court Rule 10 (which emphasizes “important” questions). See Sup. Ct. R. 10.

42To capture the public’s mood, we use Stimson’s measure. Stimson computes the measure by analyzing survey responses to a number of questions that tap an underlying liberal–conservative dimension. See James A. Stimson, Public Opinion in America: Moods, Cycles, and Swings (2d ed. 1999); Policy Mood, U. N.C. Chapel Hill, http://www.unc.edu/~cogginse/Policy_Mood.html (last visited July 5, 2012). He provides a dynamic factor analysis model that combines these responses in a principled fashion to yield a quarterly and annual measure of public mood as a single score. See Stimson, supra; Policy Mood, supra.

43We added variables to indicate whether the government was unified Republican or unified Democratic. For more sophisticated approaches, see Segal et al., supra note 39.

44In Table 2, we report the Segal & Cover score. We also estimated the models using the Justices’ partisan affiliation and their Martin–Quinn scores. See supra note 20.

45We estimated them with and without the mood measure because it is unavailable for cases argued in 2010. Also, to check for robustness, we estimated each model using a random-effects logistic regression under the assumption that there may be some unaccounted factors that affect each Justice’s propensity to strike the legislative action. Including a Justice-specific random effect, which allows each Justice to have her own baseline, helps to attend to these unmodeled factors. As the note to Table 2 indicates, it does not matter whether we estimate the models using random-effects regression or logistic regression (clustered on the Justice). Ideology is significant in all the models.

46With all other variables set at their mean or mode, the predicted probability of upholding a state/local law is .61 [.54, .68]; for federal laws, the figure is .71 [.61, .81]. To calculate the predicted probabilities, we used eq. 3.

47We used eq. 4 with all other variables set at their mean or mode. The 95% confidence interval is [.28, .36].

48 See supra note 20.

49We used eq. 3 with all other variables set at their mean or mode.

50.49 [.39, .59].

51We used eq. 4 with all other variables set at their mean or mode. The 95% confidence interval is [.54, .79].

52The 95% confidence interval is [.56, .81].

53 See supra notes accompanying Table 2.

54To reiterate, for liberal laws, extreme conservatives invalidate about 52% of the time, compared to 81% for extreme liberals.

55Although we can only speculate on why this is the case, the answer must come from either the supply side or the demand side—the supply of legal provisions to review or the demand to review by the Justices. On the supply side, much of the existing law was conservative, especially in the domains of civil rights and civil liberties. On the demand side, a Court with a liberal majority, at least until the presidency of George H.W. Bush, would typically choose to review conservative laws, all else being equal. On this logic, we would expect the Roberts Court to consider the constitutionality of a greater number of liberal legal provisions than conservative ones.

56 See infra notes accompanying Figure 4.

57This eliminates Justices Black (N = 37), Harlan (N = 44), and Sotomayor (N = 18).

58We used the mean of the Martin–Quinn scores.

59We included these variables, and not others in Table 2, to maximize the number of cases and to minimize collinearity problems in some of the models (and to depict the coefficients for the same model for each Justice). We assessed robustness by adding as many variables as we could to each model. The basic findings remain unchanged.

60Lee Epstein & Tonja Jacobi, Super Medians, 61 Stan. L. Rev. 37 (2008).

61And, indeed, Kennedy is the most aggressive Justice in our entire dataset. See Figure 1.

62He is significantly (p = .03) more likely to invalidate liberal state/local policies, though not federal laws. Modeling all laws in one equation (with a variable indicating whether the law was federal or state/local) yields a p-value of .07 on the Ideology of the Law variable. The p-value for all Terms prior to 2004 is not close to statistically significant (.84).

63And maybe well before. See Epstein & Landes, supra note 2; Clark & Whittington, supra note 5.