The Long and the Short of It: The Influence of Briefs on Outcomes in the Roberts Court.
On April 27, 2016, the Supreme Court heard oral arguments in McDonnell v. United States, a corruption case concerning the appropriate standard for prosecutions. (1) Immediately after a deputy solicitor general began to lay out the government's defense, Chief Justice Roberts raised questions about an "extraordinary document"--an amicus curiae brief by former government attorneys filed in opposition to the government's position. (2) As some commentators predicted, (3) Justice Roberts's questions, in combination with inquiries from other Justices at oral argument, portended doom for the government's argument. (4) This case is just one of many examples in which information found in briefs appears to have been important to Justices. (5)
As amicus participation at the Supreme Court has grown over the years, so has the interest in such filings. (6) Both have continued to grow during the Roberts Court, with commentators and scholars considering the role of amicus briefs in the decisions the Justices make. (7) Scholars theorize that briefs influence the Justices by conveying information relevant for deciding cases. (8) This information can help Justices discern the range of alternatives available in a case, how such options map onto their preferences, and the practical consequences of different legal rulings. (9)
Although the conceptual focus of much of the literature is on the information provided by briefs, empirical analyses do not typically examine the informational content of briefs. (10) Rather, many previous studies focus on the relationship between the number of briefs on a side and the likelihood of that side winning a case. (11) These studies generally find a modest relationship between the likelihood of winning and the side with the most briefs. Without suggesting that the volume of briefs is irrelevant, we argue research needs to consider the informational content of briefs directly. In this Article, rather than simply analyzing the raw number of briefs submitted, we consider the influence of the relative amount and breadth of information contained in the party and amici briefs for each side of a case on the Justices' votes. This pooled approach to exploring the role of all briefs is ideal based on evidence of sophisticated coordination among parties and amici. (12)
We first discuss the prevailing theoretical approaches to understanding the influence of briefs. This discussion leads to a series of hypotheses about the role of information in influencing the Justices' votes on the merits.
We then describe our dataset, which consists of 9,912 litigant and amici briefs filed in all orally argued cases decided by the Court between the 2005 to 2015 terms. From these briefs, we create several measures of the amount and breadth of information provided for each side. Our empirical tests determine to what degree the Justices' votes depend on the relative amount and breadth of information provided by each side, as well as the difference in the number of briefs filed for each side, while controlling for other relevant variables.
Our results indicate the Justices are more likely to vote for the litigant whose supporting briefs provide a greater amount of information to the Court. Importantly, after including variables for the content of the information in briefs, there is no evidence the Justices are more likely to vote for the side with more briefs. This result suggests the mechanism underlying the influence of briefs is informational content rather than merely the raw number of supporting briefs.
I. THEORIES REGARDING THE INFLUENCE OF BRIEFS
There are multiple theories regarding the importance of briefs on the decisions of the Court. First, according to accounts that focus on the role of information in briefs, briefs are valuable because they contain facts and arguments that help the Justices map preferences onto policies (13) or discover "legally relevant information." (14) Under such theories, additional information in favor of a side is likely to increase support for that side, as it reduces uncertainty and helps the Justices craft their decisions. (15) One can consider the amount of information presented in terms of, first, the total amount of information presented, where the side with the greater amount of information should be advantaged. In addition to the relative volume of information across the two sides of a case, they can differ in terms of the breadth of the information they provide. Some briefs might discuss a wider array of issues and ideas, and the side with a wider array of legal theories and approaches may be more likely to win. Studies of the role of both ideology and cognitive processes indicate, however, that the influence of information may be filtered through the preferences of the Justices. (16)
Second, affected groups or interest group theory indicates the number of groups filing briefs on a side may be important, regardless of amount of overall information provided, because it indicates general support for the position (17) or specific support from powerful groups. (18) Finally, some scholars theorize briefs are more persuasive when they contain repetitive information. (19) Specifically, persuasion theory indicates the number of voices repeating an argument is more important than the total amount of information presented by a side.
From these various accounts of the influence of briefs on Supreme Court Justices, we propose three hypotheses regarding the role of amicus briefs on the Roberts Court:
Number of Briefs Hypothesis: An increase in the relative number of briefs filed arguing for reversal, over those arguing for affirmance, will increase the probability a Justice votes for the petitioner. Total Information Hypothesis: An increase in the total amount of information in briefs filed arguing for reversal, over those arguing for affirmance, will increase the probability a Justice votes for the petitioner. Breadth of Information Hypothesis: An increase in the range of different information in briefs filled arguing for reversal, over those arguing for affirmance, will increase the probability a Justice votes for the petitioner.
II. DATA, MEASURES, AND METHODS
To evaluate the effects of information provided to the Roberts Court, we have amassed all briefs available from Westlaw for orally argued cases from 2005 to 2015. Because our focus is on the merits stage, we exclude all briefs regarding whether a petition for certiorari should be granted as well as any briefs regarding nondispositive matters. We have also excluded the small number of supplemental briefs filed by the parties. The end result is a dataset composed of 9,912 briefs. (20)
We explore the cumulative effect of all the briefs filed by both litigants and amici on each side of a case. The unit of analysis is the case-vote; and the outcome variable equals one if a Justice voted in favor of the petitioner (and zero otherwise). Information on whether each Justice voted in favor of the petitioner was obtained from the Supreme Court Database. (21) In order to contrast our findings with those from the literature we estimate two models. The first model includes all of our relative informational variables in order to test the three hypotheses. For comparison, we estimate a second model that excludes all relative information variables except for the difference in the number of briefs on each side. Since the outcome variable can only take on two possible values (zero or one), we utilize probit modes which are a type of statistical model that is specifically designed to model such binary outcomes. Both models also include fixed effects for term and Justice (although they are not shown).
Our primary explanatory variables are a series of measures of the relative content presented to the Court. For each of these measures, the difference is calculated by subtracting the relevant quantity for briefs filed on the respondent's side from the relevant quantity for the briefs filed on the petitioner's side. For example, the variable Difference in Number of Briefs is the number of briefs filed urging reversal minus the number of briefs filed urging affirmance. Each side includes all relevant briefs filed by both parties and amici. Since the outcome variable is whether a Justice voted for the petitioner, positive coefficients on our difference measures indicate that more briefs is better, the pattern anticipated by all of our hypotheses.
We go beyond simply counting the number of briefs on each side to measure both relative aggregate information and the relative breadth of the information presented on each side. For our first measure of total information we begin with the standard approach of using a simple word count as a rough approximation of the information in a document. (22) However, the impact of additional information will tend to decrease as the total word count grows. The difference between 2,000 and 1,000 words should be much more substantial than the difference between 12,000 words and 11,000 words. Consequently, we take the standard precaution of using the natural log of the word count of the briefs on each side before calculating the difference.
Our second measure of the total amount of information on each side is the total number of citations (in hundreds) to Supreme Court precedent. Legal citations are an important source of a particular type of information. (23) Attorneys naturally incorporate reference to case law in their legal arguments to the Supreme Court. Citations to precedent from the Supreme Court are the most theoretically relevant sources because they are most likely to influence the Court. References to such cases take on a finite number of forms and, therefore, can be reliably extracted from text. (24)
We also use words and citations to calculate the relative breadth of information presented by briefs. To do so, we count the number of different individual words and cases cited in the set of briefs on each side. (25) Each new legal issue that is raised will bring additional words and precedents into the discussion. Consequently, holding the aggregate amount of information constant, briefs that incorporate a wider vocabulary or discuss more precedents are likely to be addressing a broader set of issues. Conversely, a decrease in these measures indicates a higher degree of repetition.
Finally, we incorporate other variables that may affect the Justices' votes. Because many commentators have noticed the success of the Office of the Solicitor General in Supreme Court litigation, we control for whether the Solicitor General filed a brief on behalf of the petitioner or respondent. Next, we account for the ideology of the Supreme Court Justices. In order to quantify the ideological alignment between a Justice and the petitioner we use a standard measure of Supreme Court ideology, Martin-Quinn scores. (26) When the petitioner is seeking a conservative outcome, the variable Ideological Alignment is the Martin-Quinn score of the Justice, because higher Martin-Quinn scores denote greater conservatism. When the petitioner is advocating a liberal outcome, Ideological Alignment is the Martin-Quinn score of the Justice multiplied by -1. Finally, we control for whether there was unanimity in the lower court proceedings. The lack of a dissenting opinion below may indicate somewhat greater legal certainty in the case, which would potentially make it more difficult for petitioners to obtain votes in their favor. We measure this variable as one if there was no dissent below and zero otherwise.
Summary statistics for all of our explanatory variables are available in Table 1:
Table 1: Summary Statistics Percentile (27) 25th % 50th % 75th % Difference in # of Briefs -1 1 3 Difference in In (Word Count) -0.23 0.19 0.60 Difference in Total S.Ct. Cites -29 27 94 Difference in # of Unique Words -735 309 1,153 Difference in S.Ct. Cases Cited -16 5 25 Ideological Alignment w/Pet. -1.61 0.04 1.62 Percentages No Yes Solicitor General: Petitioner 95.7% 4.3% Solicitor General: Respondent 95.8% 4.2% Unanimous Below 57.2% 42.8%
III. RESULTS AND DISCUSSION
Table 2 presents the results of both models. The petitioner's advantage in terms of the number of briefs is statistically significant in both models, but the sign changes direction. Model 2, which excludes measures of the relative content of the briefing on each side, estimates that having a greater number of briefs has a positive impact on the probability of obtaining votes. However, this well-known finding flips sign when we account for the contents of the briefs. While our measures of the breadth of information in the briefing are not statistically significant, a relative advantage in total aggregate information measured in terms of both words and cites is significantly associated with a higher probability of getting a Justice's vote. Moreover, the size of both of these hypothesized effects is large. Increasing the petitioner's advantage from its 25% value in the data to its 75% value results in the predicted probability of a favorable vote increasing by a somewhat modest 0.03 for total citations, but the increase is a more substantial 0.10 for word count.
Our findings do not necessarily turn the conventional wisdom on its head; they simply offer more nuanced insight into why a greater number of briefs appears to translate into greater success. The side with more briefs is often presenting more information to the Court. However, the advantage is evidently driven by the overall quantity of information rather than the raw number of briefs presented. In fact, in a case where the two sides submit briefing of the same length and with the same number of citations, the side that manages to do so in fewer briefs has the advantage. Figure 1 provides an illustration of how the number of briefs and total information work together: it shows the predicted probability of a vote for the petitioner over the range of possible values for the petitioner's numerical advantage (or lack thereof) in terms of both the number of briefs and total citations in turn while holding the other variable at its minimum, median, and maximum. For example, when a petitioner has the maximum advantage in terms of total citations to Supreme Court precedent, the predicted probability of securing a vote is quite high no matter where the differential in the number of briefs lies.
The remaining variables perform largely as expected, and they have quite similar effects across the two models. The presence of the Solicitor General arguing on behalf of the petitioner does not have a statistically significant effect, while the Solicitor General arguing on behalf of the respondent does result in the expected significant reduction in the probability a Justice votes for the petitioner. Unsurprisingly, Justices who are more closely aligned with the petitioner ideologically are more likely to vote for the petitioner. Finally, the petitioner faces more difficulty in obtaining votes at the Supreme Court level when all judges who heard the case in the lower courts unanimously ruled against the petitioner.
The literature examining the relationship between briefs and the Justices' votes argues that briefs matter because they provide information the Justices find useful when deciding cases. The empirical evidence to date demonstrates a modest relationship between the number of briefs supporting a litigant and the probability that the litigant prevails on the merits. Missing, though, were analyses that probe directly the effectiveness of the actual information contained in the briefs, rather than their mere presence.
We tried to fill this gap. Our results show that the more information presented in the briefs--both in terms of words and citations--the more likely a Justice is to vote for that side of the case. This result, importantly, withstands the presence of numerous controls, including a variable for the number of briefs supporting each side. As a result, we conclude the underlying mechanism connecting briefs and the Justices' votes is the information contained in the briefs rather than the mere presence of the briefs. This is an important result, as it is the first time a study demonstrates a direct connection between the actual content of briefs and the Justices' votes on the merits.
Morgan L. W. Hazelton, Assistant Professor, Saint Louis University Department of Political and School of Law (by courtesy).
Rachael K. Hinkle, Assistant Professor, University at Buffalo, SUNY.
James F. Spriggs, Sidney W. Souers Professor of Government and Chair of the Department of Political Science, Washington University in St. Louis.
(1.) Tony Mauro & Marcia Coyle, A Potential Gift to Politicians, from the Justices: In Case Against Bob McDonnell, Justices Troubled by Prosecutors' Broad Interpretation of "Official Act," Nat'l L. J. (May 2, 2016), http://www.nationallawjournal.com/id= 1202756391208/A-PotentialGift-to-Politicians-from-the-J ustices?slreturn=20161030181747; see also Greg Stohr, Supreme Court May Raise Bar for Corruption, BLOOMBERG Pol. (Apr. 27, 2016, 10:17 AM), http://www.bloomberg.com/politics/articles/2016-04-27/u-s-high-court-mayraise-bar-for-corruption-prosecutions.
(2.) Mauro & Coyle, supra note I; Stohr, supra note 1.
(3.) Mauro & Coyle, supra note 1; Stohr, supra note 1.
(4.) Mauro & Coyle, supra note 1; McDonnell v. United States, 136 S. Ct. 2355 (2016).
(5.) Adam Feldman, Amicus Briefs and Oral Arguments in the Roberts Court, Empirical SCOTUS Blog, (Mar. 7, 2016), https://empiricalscotus.com/2016/03/07/amicus-briefs-and-oral-argumentsin-the-roberts-court/.
(6.) See, e.g., PAUL M. COLLINS, Jr., FRIENDS OF THE SUPREME COURT: Interest Groups and Judicial Decision Making (2008); Joseph D. Kearney & Thomas W. Merrill, The Influence of Amicus Curiae Briefs on the Supreme Court, 148 U. PA. L. Rev. 743 (2000); James F. Spriggs, II & Paul Wahlbeck, Amicus Curiae and the Role of Information at the Supreme Court, 50 POL. RES. Q. 365 (1997).
(7.) Richard Pacelle, Jr. et al" Assessing the Influence of Amicus Curiae Briefs on the Roberts Court (Annual Meeting of the Southern Political Science Association, Working Paper, 2016) (on file with authors). There is now a forthcoming version of the paper that is cited as follows: Richard L. Pacelle, Jr., John M. Scheb II, Hemant K. Sharma & David H. Scott, Assessing the Influence of Amicus Curiae Briefs on the Roberts Court (forthcoming in SOC. SCI. Q.); David H. Scott, Friendly Fire: Amicus Curiae Participation and Impact at the Roberts Court (Dec. 2013) (unpublished Ph.D. dissertation, University of Tennessee--Knoxville), http://trace.tennessee.edu/cgi/viewcontent.cgi?article=3448&context=utk_gr addiss; R. Reeves Anderson & Anthony J. Franze, Commentary: The Court's Increasing Reliance on Amicus Curiae in the Past Term, Nat'L L. J. (Aug. 24, 2011), http://www.nationallawjournal.com/id=1202512150444 /Commentary-The-Courts-increasing-reliance-on-amicus-curiae-in-the-pastterm.
(8.) See Lee Epstein & Joseph F. Kobylka, The Supreme Court and Legal Change: Abortion and the Death Penalty 309-10 (1992); Paul M. Collins, Jr., Friends of the Court: Examining the Influence of Amicus Curiae Participation in US. Supreme Court Litigation, 38 L. & SOC. Rev. 807 (2004) [hereinafter Friends]', Paul M. Collins, Lobbyists Before the US. Supreme Court: Investigating the Influence of Amicus Curiae Briefs, 60 POL. Res. Q. 55 (2007) [hereinafter Lobbyists]; Collins et al., Me Too? An Investigation of Repetition in US. Supreme Court Amicus Curiae Briefs, 97 JUDICATURE 228 (2014); Pamela C. Corley, The Supreme Court and Opinion Content: The Influence of Parties' Briefs, 61 POL. RES. Q. 468 (2008); Thomas G. Hansford, Information Provision, Organizational Constraints, and the Decision to Submit an Amicus Curiae Brief in a U.S. Supreme Court Case, 57 POL. Res. Q. 219 (2004); Thomas G. Hansford & Kristen Johnson, The Supply of Amicus Curiae Briefs in the Market for Information at the U.S. Supreme Court, 35 JUST. SYS. J. 362 (2014); Kearny & Merrill, supra note 6; Spriggs & Wahlbeck supra note 6.
(9.) See Lee Epstein & Jack Knight, Mapping Out the Strategic Terrain: The Information Role of Amici Curiae, in SUPREME COURT DECISION Making: New Institutional Approaches (Cornell W. Clayton & Howard Gillman eds., 1999); John Szmer & Martha Humphries Ginn, Examining the Effects of Information, Attorney Capability, and Amicus Participation on US Supreme Court Decision Making 42 Am. Pol. Res. 441 (2013); see also Timothy R. Johnson, Information, Oral Arguments, and Supreme Court Decision Making, 29 Am. POL. Res. 331, 333 (2001) (discussing the role of litigant and amicus briefs in informing the justices as part of a larger article focused on the role of oral arguments).
(10.) Spriggs & Wahlbeck, supra note 6.
(11.) See, e.g., Friends, supra note 11; Lobbyists, supra note 11 (finding that the difference in the number of briefs on both sides is significant).
(12.) Allison Orr Larsen & Neal Devins, The Amicus Machine, Va. L. Rev. (forthcoming); see also Adam Liptak, Study Shows How Much Work It Takes to Be Supreme Court's Friend, N.Y. Times (Mar. 7, 2016), https://www.nytimes.com/2016/03/08/us/politics/study-shows-how-muchwork-it-takes-to-be-supreme-courts-friend.html?_r=0.
(13.) See Epstein & Knight, supra note 9; Friends, supra note 8; Johnson, supra note 9 at 333; Johnson et al., supra note 9; Szmer & Ginn, supra note 9.
(14.) Kearney & Merrill, supra note 6, at 830.
(15.) See supra note 9; Friends, supra note 8; Lobbyists, supra note 8; but see Collins, supra note 6, at 120 (arguing that amicus briefs increase ambiguity and result in increased variability in decision-making).
(16.) Eileen Braman, Law, Politics, & Perception: How Policy Preferences Influence Legal Reasoning (2009); Jeffrey A. Segal & Harold J. Spaeth, The Supreme Court and the Attitudinal Model Revisited (1993); Eileen Braman & Thomas E. Nelson, Mechanism of Motivated Reasoning? Analogical Perception in Discrimination Disputes, 51 Am. J. OF POL. SCI. 940 (2007); Avani Mehta Sood, Motivated Cognition in Legal Judgments--An Analytic Review, 9 Ann. Rev. L. & SOC. SCI. 307 (2013); but see Collins, supra note 6 (finding that for 95% of observation Justice ideology does not appear to mediate how information is processed); Barbara A. Spellman, Judges, Expertise, and Analogy, in THE PSYCHOLOGY of Judicial Decision Making (David Klein & George Mitchell eds., 2010) (arguing that bias in processing information may be a function of experience); Chris Guthrie et al., Inside the Judicial Mind, 86 CORNELL L. REV. Ill (2000) (asserting that variation may be the result of systematic errors inherent to cognition).
(17.) Friends, supra note 8; Scott, supra note 7; Linda S. Simard, Empirical Study of Amici Curiae in Federal Court: A Fine Balance of Access, Efficiency, and Adversarialism, 27 REV. Lit. 669 (2007).
(18.) Kearney & Merrill, supra note 6.
(19.) See Friends, supra note 8; Collins et al., supra note 8; Kearney & Merrill, supra note 6; Claire B. Wofford, Assessing the Anecdotes: Amicus Curiae, Legal Rules, and the U.S. Supreme Court, 36 JUST. SYS. J. 274, 279 (2015) (discussing repetition of a specific legal rule); see also Spriggs & Wahlbeck, supra note 6 (discussing their finding that "the Court less frequently utilizes amicus briefs' arguments when they exclusively add information not contained in the party's brief'). This theory and corresponding empirical results are particularly interesting in light of the stress placed on the importance of new information in amicus briefs found in both the Supreme Court's rules (Rule 37) and guides for writing effective amicus briefs. See Antonin Scalia & Bryan A. Garner, Making Your Case: The Art of Persuading Judges (2008); Kelly J. Lynch, Best Friends?: Supreme Court Law Clerks on Effective Amicus Curiae Briefs, 20 J.L. & POL. 33 (2004).
(20.) Based on work we have done assessing the completeness of the Westlaw database in years immediately preceding the Roberts Court, we estimate that this database is 95% complete.
(21.) Harold J. Spaeth et al., SUPREME COURT DATABASE (2016), http://supremecourtdatabase .org.
(22.) John D. Huber & Charles R. Shipan, Deliberate Discretion: The Institutional Foundations of Bureaucratic Autonomy (2002); John D. Huber, Charles R. Shipan & Madelaine Pfahler, Legislatures and Statutory Control of Bureaucracy, 45 Am. J. OF POL. SCI. 330 (2001) (using word counts to measure the amount of information in statutes).
(23.) See, e.g., Collins, supra note 6; THOMAS G. HANSFORD & JAMES F. Spriggs, The Politics of Precedent on the U.S. Supreme Court (2008).
(24.) A python script was used to extract all text strings matching the format of a citation to a U.S. Supreme Court case. Extracted cites were compared against a list of valid citations to Supreme Court cases decided on the merits, and only validated citations were used.
(25.) For convenience, the number of unique words on each side is normalized by one thousand, and the number of unique cases cited on each side is normalized by one hundred.
(26.) 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).
(27.) The numbers for percentiles of the distribution indicate the values for each variable when the observations are ordered from lowest to highest. Specifically, the values at the bottom quarter (25%), median (50%), and top quarter (75%) are provided.
Caption: Figure 1: Predicted probabilities of a vote for the petitioner across the range of Difference in # of Briefs (left panel) and Difference in Total S. Ct. Citations (right panel) when holding the other at its minimum (dashed line), median (solid line), and maximum value (dotted line) in turn. All remaining variables are held at their median.
Table 2: Probit model results with fixed effects for term and Justice (not shown). * indicates a p-value < .05. Model 1 Model 2 Coef. (S.E.) Coef. (S.E.) Difference in # of Briefs - (0.00 (0.00 0.015 * 8) 0.020 * 3) Difference in ln(Word Count) (0.06 0.361 * 6) Diff. in Total S.Ct. Cites/100 (0.02 0.078 * 4) Diff. in Unique Words/1000 - (0.04 0.036 0) Diff. in S.Ct. Cases Cited/100 - (0.08 0.020 9) Solicitor General: Petitioner (0.08 (0.08 0.027 6) 0.079 4) Solicitor General: Respondent - (0.08 - (0.08 0.427 * 9) 0.473 * 7) Ideological Alignment w/Pet. (0.00 (0.00 0.172 * 9) 0.169 * 9) Unanimous Below - (0.03 - (0.03 0.233 * 5) 0.219 * 4) Intercept (0.09 (0.09 0.258 * 1) 0.273 * 0) N 6,025 6,025
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||The Judicial Behavior of the Roberts Court|
|Author:||Hazelton, Morgan L.W.; Hinkle, Rachael K.; Spriggs, James F., II|
|Publication:||Washington University Journal of Law & Policy|
|Date:||Mar 22, 2017|
|Previous Article:||"They've Got ... Personality!" Goals, Traits, and Behavior on the U.S. Supreme Court.|
|Next Article:||The Roberts Court and Oral Arguments: A First Decade Retrospective.|