Using Citations to Measure Influence on the Supreme Court

Abstract

I develop a new measure of interpersonal influence on the U.S. Supreme Court. Following Altfeld and Spaeth, I define influence as “the act of producing an effect on the behavior of another without the use of coercion, authority, or political control.” I propose a measure of influence based on the number of times a justice cites her colleagues’ concurring and dissenting opinions. My analysis proceeds in two stages. First, I verify that this citation-based measure is a valid method of gauging influence. From there, I use this measure to help explain justice behavior in two pivotal stages of the Supreme Court’s decision-making process: oral arguments and opinion assignments. The results demonstrate that a citation-based measure of influence can help explain and predict behavior on the Supreme Court.

Keywords

Supreme Court influence citations oral argument opinion assignment

After his death in February of 2016, many obituaries and commemorations claimed that Justice Scalia was one of the Court’s most notable jurists due to his intelligence, wit, and steadfast pursuit of his philosophical beliefs. He was, according to some measures, the most written about justice in the past 15 years (Morris, 2016). But other Court watchers noted the anomaly of Scalia’s influence: His beliefs and strongly worded opinions inspired young lawyers (Douthat, 2016) and were frequently cited by lower courts (Cross, 2010), but his passion and abilities seemingly did not influence his colleagues on the bench. As Savage (2016) notes after Scalia’s death, “[t]he same qualities that made Scalia such a transformational figure—intellectual purity, supreme self-confidence and uncompromising positions—at times made him a less effective justice.” Savage contrasts Scalia to another great justice from the past, William J. Brennan. Whereas both Scalia and Brennan sat at the extremes of the Court, the amiable Brennan was able to marshal his colleagues and effectively enact his ideological agenda (see, for example, Maltzman, Spriggs, & Wahlbeck, 2000) in ways that Scalia could not.

Court watchers and historians will likely continue to debate Scalia’s legacy and the extent to which he influenced his colleagues. This debate does not need to be waged through anecdotes alone, however. Currently, scholars measure the influence that justices have over each other based on how frequently they join each other’s concurrences or dissents (Altfeld & Spaeth, 1984; Clouatre, 1996; Davis, 1991; Williams, 2007). In contrast, I define influence as “the act of producing an effect on the behavior of another without the use of coercion, authority, or political control” (Altfeld & Spaeth, 1984, p. 237) and measure it by the number of times each justice cites the concurring and dissenting opinions of her colleagues. By introducing a sound and quantifiable measure of interjustice influence, my findings contribute to a growing literature that has reintroduced the human element to the study of the courts and judicial decision making (Corley, Sommer, Steigerwalt, & Ward, 2010; Maltzman et al., 2000; Meinke & Scott, 2007; Sommer, 2014).

This article proceeds as follows. The next section expands on the practical and theoretical importance of studying judicial influence and the grounds for measuring it by analyzing discretionary behavior like citation practices. Next, I conduct a test to determine whether citations to concurring and dissenting opinions are patterned along dimensions related to the concept of influence. To further test the validity of the measure, I use it in two models of the Court’s decision-making process: one on oral arguments and the other on opinion assignments. Finally, I offer a discussion and conclusion.

Background and Theory

Research in political science has established that, as repeat players in a shared strategic game, Supreme Court justices develop mutually beneficial strategies in an attempt to secure their desired legal policy while acting within institutional and legal constraints (Epstein & Knight, 1998; Maltzman et al., 2000; Segal & Spaeth, 2002). In making important decisions, justices must therefore take into consideration not only their colleagues’ ideological preferences but also their past practices and the likelihood that they will cooperate to achieve mutually beneficial outcomes.

Strategic and collegial models of Court decision making provide a powerful and predictive explanation of elite, hyperrational actors’ behavior. They do not, however, provide a holistic view of the justices as human actors. These models explain judicial behavior as political processes predicated on power and bargaining (Altfeld & Spaeth, 1984). A justice will pursue her own policy preferences unless a colleague can threaten or induce her to change course. But behavior may also be the product of influence and leadership. Altfeld and Spaeth (1984) define influence as “the act of producing an effect on the behavior of another without the use of coercion, authority, or political control” (p. 237). According to Altfeld and Spaeth, “influence primarily involves the ability to persuade or convince another of the correctness of one’s opinions and may thus even be thought of as a fundamentally sociological or psychological, rather than a political, process” (p. 237).

A purely rational justice may cooperate with a colleague because it facilitates mutually beneficial negotiations but realistically a justice may also be collegial because she values her colleagues’ opinions and her relationships with them (Baum, 2006; Murphy, 1964). Supreme Court justices possess some of the nation’s most brilliant legal minds, and it would be natural for them to be influenced by the force of each other’s ideas. After many years of working together, it is inevitable that they form strong relationships that go beyond strictly utilitarian considerations. A mentor or a kindred spirit might help steer a colleague’s legal and policy views, slowly and perhaps subtly, over time. Murphy (1964), for example, suggests that Justice Stone gravitated toward the policy views of Justices Holmes and Brandeis primarily because of the “warm and stimulating companionship” of those two justices. Having an understanding of whether and to what extent justices influence each other can aid models of judicial decision making because it captures some of the informal, interpersonal dynamics that purely strategy-based, rational choice models may miss.

Measuring Influence

One method of measuring interjustice influence is based on how frequently they join each other’s concurrences or dissents (Altfeld & Spaeth, 1984; Clouatre, 1996; Davis, 1991; Williams, 2007). Joining a concurrence or a dissent is theoretically discretionary because it does not carry any precedential weight nor does it alter the policy outcomes of decisions (Hansford & Spriggs, 2006). A justice cannot “force” a colleague to join her concurrence, for example, and thus the decision to do so is presumably based on a sincere belief that the concurrence or its author is “right” (Altfeld & Spaeth, 1984). It seems reasonable to infer that justices who frequently join each other’s concurrences and dissents have a special relationship that goes beyond simply sharing some legal or policy preferences.

This measurement, on its own, is limited and potentially biased. To detect and measure influential relationships on the Court, one must look for purely discretionary behavior; a behavior that is driven by a personal regard or respect for a colleague and not the behavior that is “contaminated . . . by considerations of power and authority” (Altfeld & Spaeth, 1984, p. 238). The decision to join a concurrence or dissent, however, is not purely discretionary.

First, a justice may sympathize with the arguments contained within a dissent or a special concurrence (a concurrence which agrees with the outcome but not the legal rationale) but be unable to join those separate opinions for fear of disrupting the majority coalition and altering the outcome of the case. For instance, Maltzman et al. (2000) argue that, in Pennsylvania v. Muniz (1990), Justice Brennan opted to vote with the majority despite the fact that his sincere policy preference aligned with Justice Marshall’s dissent. As Brennan explained, he voted with the majority so that, as the senior associate justice in the majority coalition, he could author and control the content of the majority opinion. Brennan may have been influenced by Marshall and may have wanted to join his opinion, but he felt forced by countervailing strategic and policy-driven considerations to join the majority.

Similarly, a justice may be reticent to join a regular concurrence (a concurring opinion that agrees with the outcome and rationale of the majority opinion but is written separately to highlight or expand upon a certain topic) for fear that doing so will weaken the long-term precedential value of the majority opinion or alter its implementation (Hansford & Spriggs, 2006; Wahlbeck, Spriggs, & Maltzman, 1999). Obergefell v. Hodges (2015) provides a recent example of this dynamic. Although each dissenting justice authored his own dissenting opinion, none of the justices voting in the majority authored a concurring opinion. Justice Ginsburg has implied that the majority justices refrained from authoring any concurring opinions so that the majority opinion would carry more weight (Totenberg, 2015).

Finally, the decision to join a separate opinion may also be driven by tit-for-tat strategies. The justices may view joining a separate opinion as a sanction against the majority opinion authors and may refrain from doing so to secure cooperation from their colleagues in the future (Maltzman et al., 2000). The bottom line is that joining a concurrence or dissent is imbued with strategic and policy-driven motivations. A measure of influence based on joining behavior is thus theoretically unsound and may be biased by including (or omitting) justices’ nondiscretionary decisions.

Beyond joining behavior, others have used an alternative method of measuring influence that relies on justices’ citations to each other’s opinions (Kosma, 1998; Landes, Lessig, & Solimine, 1998). This follows a similar line of logic in that it captures mostly discretionary behavior. A justice has a great deal of leeway in determining what cases are cited so the decision to cite one case over another indicates that the justice finds it persuasive. As an extension of this logic, a justice whose opinions are frequently cited has presumably generated influential ideas. The main advantage of this measure over one based on joining concurring and dissenting opinions is that it produces a considerably larger data set that should provide for more robust results. Unfortunately, this measure is limited because legal precedent constrains both legally minded justices and policy-oriented justices concerned with preserving institutional legitimacy (Cross & Spriggs, 2010; Hansford & Spriggs, 2006; Richards & Kritzer, 2002). This method is thus more aptly suited to measuring a justice’s influence over legal development and not her influence over her colleagues.¹

I combine the strengths of the two dominant (yet individually flawed) measures of influence to analyze the number of times justices cite each other’s concurring and dissenting opinions.² The decision to cite a concurring or dissenting opinion does not carry the potential costs of joining one. It would be less likely to be seen as a sanction against the majority opinion authors, and indeed it does not imply any direct trade-off between the vitality of a majority opinion and separate opinions. Moreover, a justice has the opportunity to cite thousands of cases over the course of her career whereas she may only have the opportunity to join a few hundred concurrences and dissents.³ Data on discretionary citations should therefore be a more robust and valid measure of influence than one based on joining concurrences and dissents.

Furthermore, although a justice may feel bound for precedential reasons to cite certain majority opinions (Cross & Spriggs, 2010; Hansford & Spriggs, 2006), concurrences and dissents do not carry precedential weight. A strategic justice may cite a majority decision to browbeat her colleagues and to persuade them that they have deviated from controlling precedent, but because concurrences and dissents lack precedential value, citing them can be only as persuasive as the ideas they express. This means that citing a concurrence or dissent may be less of a strategic act and more of a demonstration of having been persuaded by the ideas expressed within those opinions. Disproportionately citing the concurring or dissenting opinions of a particular colleague over time may indicate that the citing justice is influenced by the general legal and policy views of the cited justice or that the justice simply respects the author of those separate opinions. My general hypothesis, then, is that interpersonal influence on the Court matters and that it can be measured via the citation practices of the justices. The next two sections use empirical tests of this hypothesis.

A Citation-Based Model of Influence

If citations to concurring and dissenting opinions tap into something substantively important, then they will not be random. Rather they will be patterned across dimensions relevant to the concept of influence. To validate my measure of influence, I estimate a model to predict the number of times a justice will cite the separate opinions of a given colleague using variables that tap into ideological and interpersonal forces. In so doing, I focus on several specific hypotheses.

First, previous studies indicate justices are influenced by colleagues ideologically similar to themselves (Altfeld & Spaeth, 1984; Segal & Spaeth, 2002). Justice Scalia, for example, shared more legal and policy views with Justice Thomas than he did with Justice Breyer. On an interpersonal level, Scalia and Breyer may at times have found it difficult to interact with each other after having a heated disagreement over a case. In contrast, the frequent agreements between Scalia and Thomas likely facilitated a strong working relationship. Simultaneously, on an intellectual level, it is natural to respect, and view as more credible, opinions that are similar to one’s own (Lupia & McCubbins, 1998). Hence, Scalia was likely inclined to look upon Thomas’s legal and policy opinions with more approval than he would Breyer’s. This influence is likely borne out in their decision to cite each other’s concurring or dissenting opinions. This leads me to predict the following:

Ideology Hypothesis: Justices will cite the concurring or dissenting opinions of colleagues who are ideologically similar to them at a higher rate than colleagues who are ideologically dissimilar.

Despite the intuitive nature of this relationship, ideological congruence alone is not a sufficient basis to form an influential relationship. Instead, it takes time for a justice to grow to respect a colleague’s views or to form a strong interpersonal relationship (Murphy, 1964). Indeed, sitting on the Court together provides opportunities to interact on a professional and personal level, above and beyond the sort of indirect communication seen in written opinions. During opinion writing, for example, justices can encourage each other and provide positive criticism in their notes on slip opinions (Murphy, 1964). The combined effects of informal interactions should help justices form bonds with one another, bonds that might be revealed by the decision to cite each other’s separate opinions more often.

Shared Tenure Hypothesis: The rate at which a justice cites a colleague’s concurring or dissenting opinions will increase as their shared tenure increases.

As an extension of this logic, a justice should be more likely to cite the separate opinion of retired colleagues if she previously served with those colleagues. This is an important test of the extent to which interpersonal forces structure influential relationships. That is, if the only reason a justice cites a concurring or dissenting opinion is to strengthen an argument, it should not matter whether the citing justice had ever served with the author of the separate opinion. Citations to retired justices should instead either be random or be primarily driven by the extent to which the author and the retired justice have similar views on legal policy. If, however, justices are more likely to cite the opinions of former colleagues than they are to cite the opinions of justices with whom they never served, it would indicate that there must have been some interpersonal element that draws the citing justice to the views of the cited justice.

Served Together Hypothesis: Justices will cite the concurring or dissenting opinions of retired justices more if they served on the bench with that retired justice.

Similarly, if the only factor that drives a justice to cite a colleague’s separate opinions is ideological proximity or having similar views on legal policy, the rate at which the justice cites her colleague’s separate opinions should not change after the colleague retires. If, however, interpersonal dynamics matter on the Court and drive justices’ citation practices, the lack of personal contact should make that colleague increasingly less influential after she retires. Thus, I expect that the rate at which a justice cites the concurring and dissenting opinion of another justice will decline after that justice retires.

Retirement Hypothesis: The rate at which a justice cites a colleague’s concurring or dissenting opinions will decrease after the colleague retires.

As the most explicit test of whether interpersonal forces matter, I hypothesize that justices will be influenced more by colleagues with whom they share personal traits or experiences. In particular, I hypothesize that justices will be influenced by colleagues who have similar personal, educational, or professional backgrounds (Hartup & Stevens, 1997; McPherson, Smith-Lovin, & Cook, 2001).

Personal History Hypothesis: Justices will cite the concurring or dissenting opinions of justices more if they share personal, educational, and professional backgrounds.

Finally, research demonstrates that the justices are strategic actors who are adept at maneuvering within institutional constraints to pursue their policy outcomes (Epstein & Knight, 1998; Johnson, Spriggs, & Wahlbeck, 2005; Maltzman et al., 2000). Murphy, for example, paints a picture of the justices as being simultaneously shrewd policy maximizers and vulnerable humans (Murphy, 1964). He suggests justices’ personal esteem for each other is tied to self-interest, stating that “[w]e tend to like those whose actions have benefited us in the past, to interpret as beneficial the actions of those whom we like, and, in turn, to help those whom we like” (Murphy, 1964, p. 38). This argument suggests that justices would be naturally drawn to members of the Court whose opinions and decisions are pivotally important. Significant research demonstrates that the median member of the Court possesses disproportionate power in terms of dictating case outcomes (Bonneau, Hammond, Maltzman, & Wahlbeck, 2007; Hammond, Bonneau, & Sheehan, 2005; McGuire, Vanberg, & Yanus, 2007; Schwartz, 1992). I thus hypothesize that the justices will be more influenced by the median member of the Court than by other colleagues.

Median Member Hypothesis: Justices will cite the concurring or dissenting opinions of justices who are the median member of the Court more frequently than those who are not.

These hypotheses focus on the extent to which the shared ideas, shared interests, or shared trust between justices create measurable levels of influence. Rejecting the null hypothesis in favor of my alternative hypotheses would provide substantial evidence that justices influence each other and that they demonstrate this influence in their citation practices.

Data and Analysis

My dependent variable—citations—reports the number of times in a given term a justice cites another justice’s separate opinions by name. Each observation is a citing pair for a given term (e.g., the number of times Scalia cites Marshall in 1988). I collect citing data for every justice sitting on the court on the 1983 term and every justice appointed after that date up until the 2013 term.⁴ To create counts for citations, I obtained every written opinion from my sample period. I then used the tm package R (Feinerer, Hornik, & Meyer, 2008) to search for instances in which a justice cites the concurring or dissenting opinions of a colleague by name. To determine how many times Scalia cited one of Marshall’s concurring and dissenting opinions, for example, I search every opinion Scalia wrote for the phrase “Marshall, j., concurring” or “Marshall, j., dissenting.” I use the cases’ Lexis citations and the Supreme Court Database to link these counts to each term.⁵

My data set consists of 3,777 observations. Table 1 provides summary statistics for the sample included in my model. Of note, the mean number of citations per cited justice per term is fairly low at 1.81. While collecting the data, it became apparent that justices have unique citation patterns. Some justices, like Scalia and Stevens, cited many of their colleagues’ concurring and dissenting opinions. Others, like Rehnquist and Burger, showed relatively more constraint in their citation decisions. This is consistent with previous research suggesting that there are idiosyncratic writing and citation conventions for each justice (Cross & Spriggs, 2010).

Table 1.

Within-Sample Summary Statistics.

	M	SD	Minimum	Maximum
Citations	1.81	2.76	0.00	29.00
Ideological distance	2.56	1.77	0.00	8.11
Shared tenure	8.12	7.04	0.00	29.00
Retired	0.41	0.49	0.00	1.00
Biography	1.57	1.31	0.00	6.00
Median justice	0.07	0.25	0.00	1.00
Concurring or dissenting opinions	351.35	238.22	0.00	1,090.00
Clerks’ ideology	0.44	0.29	0.00	1.26
Freshman	0.07	0.26	0.00	1.00
Term	1999	—	1983	2013
Observations	3777

Note. These data do not include instances of justices citing their own concurring or dissenting opinions.

I measure the ideological distance of citing and cited justice pairs using the 2014 data files of the Martin–Quinn (MQ) mean judicial ideology scores (Martin & Quinn, 2002). MQ data are available for all justices from the 1937 term to 2014 term and place each individual justice on a left to right ideological space where negative scores indicate the justice is liberal and positive scores indicate she is conservative. The scores are dynamic and measure changes in the ideological preferences of a justice over time. To create my specific ideological distance variable, which measures the degree to which the citing and cited justices are ideologically congruent, I take the absolute value of the difference between their MQ scores. For those instances in which a justice is citing someone who has retired, I use the mean of the retired justices’ MQ scores over the time period within my sample.

I create a continuous variable, shared tenure, which measures the number of terms the cited and citing justices have sat on the bench together.⁶ This allows me to test whether, over time, a justice is more likely to grow close to (or away from) a colleague. I also create a variable, retired, wherein a score of 0 indicates that the justice is active and a 1 indicates that she has retired.⁷ I interact shared tenure and retired to determine whether justices are more likely to cite the opinions of retired colleagues if they had previously served with them on the bench.

Biography measures the extent to which the citing and cited justices share personal characteristics. These data come from The Supreme Court Compendium (Epstein, Segal, Spaeth, & Walker, 2012). Appendix B contains details about how this variable is coded. Briefly, however, biography is a count of the number of categories of personal background characteristics located within the Compendium in which the citing and cited justices share some relevant characteristics.⁸ For example, Alito and Kagan both went to the same undergraduate school (Princeton), and both served in some capacity in the federal government (Alito as assistant solicitor general and Kagan as the solicitor general). Hence, Alito and Kagan have a biography score of 2. Biography is meant to capture, in broad brush strokes, the extent to which two justices are similar. It runs from a hypothetical minimum of 0 to a maximum of 19, but in practice, the highest score is a 6 (Justices Stevens and White).⁹

To determine whether the justices are more likely to be influenced by colleagues who are strategically valuable, I generate a variable, median justice, that measures whether the cited justice is the ideologically median justice (1) or not (0) as determined by the nine sitting justices’ MQ scores during that term. Recent research by Lauderdale and Clark (2012) indicates that a better measure of the median justice would take into consideration the voting records in different issue areas. My data, unfortunately, neither identify the issue areas covered in the cases authored by the citing justices nor capture the issue areas of the cases that they are citing. However, the median justices included in my data set—White, Powell, Souter, O’Connor, and Kennedy—are also identified by Lauderdale and Clark as being among the most common median justices by issue area in the time period I analyze.

To better measure influence, however, I must take into account several potentially confounding factors. First, I control for the number of concurring or dissenting opinions authored by each justice. An influential justice who has not penned very many concurring or dissenting opinions will be cited less frequently if only because it is more difficult to work her opinions into a decision. Concurring or dissenting opinions measures the number of concurring and dissenting opinions the cited justice has authored. These data come from the Spaeth Supreme Court Justice Centered Database (Spaeth et al., 2014). Because my dependent variable does not capture case-by-case data and instead reports only the aggregate number of citations in a given year, the concurring or dissenting opinions variable is time lagged by 1 year to avoid controlling for concurring or dissenting opinions that have not yet been authored.

Second, the justices’ clerks play a substantial role in crafting and editing the Court’s written opinions (Kromphardt, 2015; Wahlbeck, Spriggs, & Sigelman, 2002), and citation decisions may be influenced by the views and perspectives of the clerks. I thus control for the clerks’ ideologies as well. Data on clerks’ ideologies come from Kromphardt (2015) and are based on the judicial common space (JCS) scores (Epstein, Martin, Segal, & Westerland, 2007) of the lower court judges with whom clerks served prior to clerking on the Supreme Court. I average the ideologies of each justice’s clerks and then calculate the absolute difference between that ideological score and the JCS score of the cited justice.¹⁰

In addition, I control for a freshman effect (Hagle, 1993). It may take time for a new justice to develop a coherent legal philosophy or establish a distinct writing style. The rate at which a justice cites her colleagues during her first years on the bench is likely to differ from the rate at which she cites her colleagues as a more tenured member of the Court. Freshman is coded 1 if the citing justice is in her first two terms on the Court and 0 otherwise. Finally, the justices’ citation practices likely change over time. To control for any temporal effects, I control for each term that the citing justices sit on the bench.¹¹ Term is a continuous variable equal to the term in which the citing justice cites the cited justice.

Results

Table 2 provides preliminary evidence that the justices’ citations to the concurring and dissenting opinions of their colleagues are not random. The columns contain the citing justices, and the rows contain the cited justice. These data represent the total raw count of times that each justice has cited the concurring and dissenting opinions of every other justice over the terms covered by my data. The lower left cell, for example, shows that Thomas has cited Justice Burger’s concurring or dissenting opinions 12 times. The cells are shaded by percentiles from lightest gray (10th percentile) to the darkest gray (95th percentile) with shade variants ranging from the 25th, 33rd, 50th, 66th, 75th, and 90th percentiles.

Table 2.

Raw Count of Citations by Citing and Cited Justices.

	Burger	Powell	Brennan	Marshall	White	Blackmun	Rehnquist	O’Connor	Souter	Stevens	Alito	Breyer	Ginsburg	Kagan	Kennedy	Roberts	Scalia	Sotomayor	Thomas
Burger	2	11	8	2	6	3	3	5	0	7	0	0	0	0	0	0	0	0	0
Powell	7	61	18	10	22	9	6	7	0	17	0	0	0	0	0	0	0	0	0
Brennan	11	64	257	54	57	38	40	63	0	79	0	0	0	0	5	1	14	0	0
Marshall	2	31	42	134	24	15	12	38	1	59	0	0	0	0	7	0	14	0	0
White	6	24	22	6	77	18	11	18	1	26	0	0	0	0	1	0	11	0	0
Blackmun	5	23	39	15	21	23	7	41	3	38	0	0	1	0	9	0	17	0	1
Rehnquist	15	39	12	2	37	13	38	49	6	30	0	2	1	0	28	0	32	0	6
O’Connor	15	66	45	27	71	42	37	283	52	100	0	25	10	0	58	0	68	0	44
Souter	3	20	22	14	20	26	20	102	65	54	0	21	14	0	29	1	65	0	9
Stevens	29	78	94	57	79	44	66	167	62	707	5	51	29	0	96	5	118	3	45
Alito	1	6	6	4	6	1	6	8	4	18	13	19	9	5	21	1	32	3	30
Breyer	1	17	12	14	5	9	20	76	63	111	7	108	26	2	67	6	79	8	38
Ginsburg	9	5	24	6	8	7	11	39	30	87	13	21	60	3	30	0	58	0	23
Kagan	0	2	0	0	0	0	1	2	1	2	4	7	1	0	3	5	16	3	13
Kennedy	3	24	28	17	26	24	22	92	14	65	6	29	10	1	129	3	74	2	38
Roberts	1	4	3	2	4	2	15	17	10	25	4	15	7	2	24	1	25	5	14
Scalia	8	36	36	18	45	23	51	94	29	108	11	27	15	1	62	3	387	3	35
Sotomayor	0	4	1	4	0	1	4	7	3	10	5	17	10	0	11	6	17	2	15
Thomas	12	24	15	23	42	26	42	127	26	97	14	22	15	4	45	12	168	1	372
Total	130	539	684	409	550	324	412	1,235	370	1,640	82	364	208	18	625	44	1,195	30	683

As the table indicates, many justices frequently cited O’Connor, Scalia, and Stevens but Powell and White are also well cited. Table 2 thus indicates that the most prolific writers tend to generate the most citations but those justices whose votes are often pivotal garner many citations per concurrence or dissent as well. A multivariate analysis of citation behavior provides a more formal test of these initial impressions.

Because my dependent variable is a count of citations, I use a count model.¹² Furthermore, because statistical tests indicate that the data are overdispersed, I use a negative binomial model (King, 1989). Such models have been used in past analyses of Supreme Court behavior (e.g., Black, Johnson, & Wedeking, 2012). Finally, each justice exhibits idiosyncratic writing styles and citation practices (Cross & Spriggs, 2010). Over the course of their careers, for example, Stevens cited the concurring and dissenting opinions of almost all of his colleagues more than Chief Justice Rehnquist cited any of his colleagues. I thus estimate a fixed effects model to account for intrajustice correlation and interjustice heterogeneity of errors (Clark & Linzer, 2015).¹³

All of my key variables of interest are statistically significant and signed in the predicted direction (see Table 3). The only variable that fails to meet traditional levels of statistical significance is freshman. The results indicate that influence is clearly a function of colleagues’ policy preferences and strategic value, providing strong support for my Ideology Hypothesis and Median Member Hypothesis. As Figure 1 demonstrates, the closer justices are to each other ideologically, the more likely they are to cite each other’s concurring and dissenting opinions.¹⁴ Holding all other values at their means and modes, moving from the most distant ideological pair of justices (which would be justices Thomas and Marshall in 2005) to the closest (justices Breyer and Souter in 2000) increases the predicted mean count of citations to concurring dissenting opinions by more than 36% (from 1.1 to 1.5 citations per year). Furthermore, the median justice garners about 40% more citations than similarly situated colleagues (1.4 to 2 predicted mean citations per year).¹⁵ As Murphy (1964) suggests, there seems to be a reinforcing relationship between personal regard and self-interest. The justices view ideological allies and strategically valuable colleagues positively and hence are more likely to choose to cite their past concurring and dissenting opinions.

Table 3.

Fixed Effects Negative Binomial Count Model for Citations.

	Citations
Ideological distance	−0.036*
Ideological distance	(0.015)
Shared tenure	0.009*
Shared tenure	(0.004)
Retired	−1.155***
Retired	(0.075)
Retired × Shared Tenure	0.017**
Retired × Shared Tenure	(0.006)
Biography	0.030*
Biography	(0.014)
Median justice	0.364***
Median justice	(0.059)
Concurring or dissenting opinions	0.001***
Concurring or dissenting opinions	(0.000)
Clerks’ ideological distance	−0.323***
Clerks’ ideological distance	(0.090)
Freshman	−0.010
Freshman	(0.080)
Term	−0.024***
Term	(0.003)
Constant	48.434***
Constant	(6.370)
Observations	3,777
Log likelihood	−6,134.181
Wald χ²	979.14

Note. Standard errors in parentheses.

p < .05. **p < .01. ***p < .001.

Figure 1.

The graph presents the predicted mean count of citations to concurring or dissenting opinions given the ideological distance between the citing and cited justices.

The results also provide strong evidence that interpersonal relationships play a large role in driving influence. For example, going from a value of 1 for shared tenure (indicating that justices have been on the bench together for just 1 year) to the in-sample maximum of 29 (indicating that the justices sat on the bench together for 29 years, as was the case with Justice Stevens and Chief Justice Rehnquist) increases the predicted mean count of citations by 30% (from 1.3 to 1.7 citations per year).¹⁶ The biography variable is also statistically significant—going from the in-sample minimum to maximum increases the predicted mean count of citations by 31% (from 1.3 to 1.7 citations per year).¹⁷ If citations to concurring or dissenting opinions were purely a function of strategic or policy-driven calculations, there is little reason to suppose that a justice would cite the opinions of a long-term colleague above those of a similarly situated newcomer to the bench and even less reason to expect that the personal history of the justices would have any impact whatsoever on their citation patterns.

Further evidence of the interpersonal nature of influence comes from behavior toward retired justices. A justice retiring decreases the predicted mean count of citations that she will receive by about 65% (from 1.4 to 0.5 citations per year). The positive coefficient for the interaction between retired and shared tenure indicates that the effect of shared tenure is more pronounced for citations to retired justices than to current colleagues. Having previously served with a now-retired justice for the average amount of shared tenure within this sample (about 8 years) increases the predicted mean count of citing her concurring or dissenting opinions by about 23% compared with other retired justices with whom she has never served (roughly 0.41 to 0.50 citations per year). If citations to concurring or dissenting opinions were purely driven by ideological compatibility, there would be little reason to cite the opinions of colleagues you have worked with above those of more ideologically congruent retired justices. Combined, these findings indicate that interpersonal relationships play a large role in driving influence.

In total, these models suggest that a measure of influence based on the number of citations to the concurring and dissenting opinions of colleagues is sound. Justices do not cite their colleagues’ concurring and dissenting opinions in a random fashion. They are more likely to cite those written by ideological allies, strategically important colleagues, their long-term coworkers, and by colleagues who are personally similar to themselves. Justices develop interpersonal relationships and intellectual respect with their colleagues. This respect is evident in and can be measured by their citation practices, and it likely influences their behavior in substantive and important ways. I test that proposition in the next section.

Applying the Measure

To test whether this measure of influence is valid and useful, I include it in two models predicting the justices’ behavior during important parts of the decision-making process. In the first, I use my citations measure in the Black et al. (2012) model of interruptions during oral argument. In the second, I use my citations measure in the Wahlbeck’s (2006) model of Chief Justice Rehnquist’s opinion assignment decisions.

Citations and Oral Argument

Previous research has demonstrated that the justices use oral arguments to gather information (Johnson, 2004; Johnson & Ringsmuth, 2013) and that the information presented during oral argument can ultimately change the justices’ opinions about the case (Bryan, Johnson, & Ringsmuth, 2013; Johnson, 2004; Johnson, Wahlbeck, & Spriggs, 2006; Wedeking, 2010). Further research has also demonstrated that the justices use oral argument as a sort of “preconference” during which they can begin to identify and form coalitions (Black et al., 2012).

Black et al. (2012) analyze one mechanism by which the justices strategically use oral arguments to craft eventual voting coalitions: interruptions. They hypothesize that the justices will strategically interrupt ideological opponents to prevent them from advancing their arguments and from communicating positions to their like-minded colleagues (Johnson, Black, & Wedeking, 2009). In contrast, justices should be less likely to interrupt ideological allies because such speakers are presumably advancing the justice’s own position. In addition to strategic and ideological goals, Black et al. (2012) also demonstrate that the justices care about norms of collegiality to the extent that a justice is much more likely to interrupt a colleague who has previously interrupted her. This sort of “tit-for-tat” strategy indicates that interpersonal dynamics affect the justices’ behavior during even the most public part of their decision-making process, making interruptions during oral argument an ideal area in which I can test the predictive power of my citation-based measure of influence.

Just as justices are less likely to interrupt ideological allies or colleagues who have not interrupted them before, a justice should be less likely to interrupt a colleague who she respects. If a justice cites the concurring and dissenting opinion of a colleague more often than anyone else’s, she likely shares that colleague’s views on legal policy and respects that colleague.¹⁸ Justices, like normal people, probably avoid interrupting colleagues who they respect and admire. I thus hypothesize that a speaking justice is less likely to interrupt a colleague who she is influenced by as measured by having a higher rate of citing that colleague’s concurring or dissenting opinions. The rate of citations should be negatively related to the rate of interruptions even when controlling for ideological congruence and the other key variables included in the original model used by Black et al.(2012).

Black et al. (2012) use data from 681 orally argued cases between 1998 and 2007. The dependent variable is whether a justice interrupts a colleague (1) or not (0), making each justice’s response to the speaking justice the unit of observation. For example, each time a justice speaks, all eight other justices have an opportunity to interrupt her, producing eight observations. The 681 cases yielded 694,496 observations. I add to this model a count of how many times each of the nonspeaking justices have cited the concurring or dissenting opinions of the speaking justice in the previous term. I use a lagged citation count to avoid including concurring or dissenting opinions that had not yet been authored.¹⁹ I also estimate a separate model using the lagged number of separate opinions justices join as a measure of influence. The results are reported in Appendix D.

The addition of my citation measure does not substantively alter the performance of any of the original variables in Black et al. (2012) model. Ideological distance and being previously interrupted remain key predictors of whether a justice interrupts a speaker. Interjustice influence, however, as measured both by joining and citing to separate opinions, has a statistically significant and substantively stronger impact on the probability that a justice interrupts a speaker than does ideological distance.²⁰ As Figure 2 demonstrates, moving from the in-sample minimum of citations to the in-sample maximum produces a 50% change in the probability of interrupting the speaking justice (from .006 to .003).

Figure 2.

The graph presents the predicted probability that a specific colleague will interrupt a speaking justice given the number of times that the specific colleague has cited the concurring or dissenting opinions of the speaking justice.

The predicted probabilities are small because the actual incidents of interruptions are quite small.²¹ One way to interpret the substantive magnitude of these effects would be to compound them over a period of time. Justice Stevens’s treatment of the Court’s two moderates, Justices Kennedy and O’Connor, is informative. From 1999 to 2004, Stevens cited 52 of O’Connor’s concurring and dissenting opinions and 34 of Kennedy’s. Presumably, Stevens is more influenced by O’Connor and her views on legal policy than he is by Kennedy. From the model, it is possible to extrapolate the impact that interjustice influence has on Stevens’s propensity to interrupt the Court’s moderate justices. Using the real citation counts while holding all other variables at their means and modes, including ideological distance, translates into a 47% higher probability of Stevens interrupting Kennedy than O’Connor from 2000 to 2005 (.0028 probability vs. .0019, respectively). And, indeed, from 2000 to 2005, Stevens interrupted Kennedy 52 times and O’Connor only 12 times. Other factors certainly play into whether a justice interrupts a colleague, but these data suggest that influence—as measured by citations to concurring and dissenting opinions—is substantively important. I look at the role of influence in another pivotal stage of the Court’s decision-making process next.

Citations and Opinion Assignment

Wahlbeck (2006) analyzes a key stage in the Court’s agenda setting process: determining which justice will draft the majority opinion and hence control the policy outcome of a case. When voting in the majority coalition, convention dictates that the Chief Justice assign the task of drafting the majority opinion to a colleague or to himself. Wahlbeck theorizes that the Chief Justice attempts to set the agenda by strategically assigning opinions so as to secure a majority opinion that is closest to the Chief Justice’s preferred outcome. The Chief Justice pursues this goal within the “twin constraints” of securing the majority coalition and maintaining efficient operation of the Court. Hence, Chief Justice Rehnquist could not assign an opinion to an ideological ally if doing so would risk fracturing the majority coalition—doing otherwise would risk moving the opinion even further from Rehnquist’s preferred outcome. Similarly, Rehnquist could not assign an opinion to an ideological ally if that justice was overworked and late in submitting opinions for other cases—doing otherwise would risk the efficiency of the Court as a whole.

Using data from Justice Blackmun’s notes, Wahlbeck estimates a random effects probit model to test each of his key hypotheses regarding Rehnquist’s opinion assignments. The dependent variable is whether Rehnquist assigned the majority opinion to a justice, and there is one observation for each justice who was eligible to receive the assignment by virtue of voting within the majority coalition. Wahlbeck finds that Rehnquist spreads assignments around to ensure the smooth functioning of the Court, that he assigned cases to ideological moderates when there was a minimum winning coalition, and that he generally did not make a habit of assigning opinions to ideological allies except in politically salient cases (Wahlbeck, 2006). The question remains whether Rehnquist’s assignment choices in politically salient cases were predicated solely on ideological similarity with the assignee or whether interpersonal influence mattered as well.

I use Wahlbeck’s replication data to test whether my measure of influence can make a substantive contribution to understanding Chief Justice Rehnquist’s assignment behavior. Per Wahlbeck’s original hypotheses, I hypothesize that in politically salient cases, Rehnquist will be more likely to assign opinions to those colleagues who exert the most influence on him. As a measure of influence, I use a simple count of the number of times in the past year that Rehnquist has cited any of his colleague’s concurring or dissenting opinions. I additionally estimate a model that measures influence by the number of times Rehnquist has joined the separate opinions of his colleagues in the previous year, reported in Appendix E. I interact this measure with Wahlbeck’s measure for politically salient cases (Epstein and Segal’s (2000) New York Times measure). Per Wahlbeck’s original theory, the outcome of a case likely matters more when it deals with a politically salient issue. I hypothesize that Rehnquist should be more likely to assign such cases to influential colleagues and thus anticipate a positive coefficient for the interactive variable.

With the addition of my influence measure, ideological distance retains a direct and positive effect on Rehnquist’s assignment decisions.²² Per Wahlbeck’s theory, Rehnquist strategically attempted to spread the burden of authoring opinions while also saving the task of authoring important opinions for those colleagues he trusted most. But Rehnquist had perhaps a more nuanced view of his colleague’s legal policy preferences and abilities than a simple measure of ideological congruence can capture. The inclusion of the citation-based influence variable results in the interaction effect between ideological distance and salience losing its statistical significance. That is to say, ideological distance has the same effect on Rehnquist’s decision in salient cases as it does in nonsalient cases.

In support of my hypothesis, the interaction between my citation-based measure of influence and the New York Times measure of salience is statistically significant at the p = .01 level and signed in the predicted direction.²³ Influence as measured by Rehnquist’s joining behavior is not statistically significant. As Figure 3 demonstrates, moving from the in-sample minimum of 0 citations to the maximum of 12 citations (Justice Powell in 1986) increases the probability of being assigned the task of authoring an opinion in salient cases almost fourfold (from .13 to .44). Chief Justice Rehnquist is more likely to assign an opinion to an influential colleague in politically salient cases even when controlling for ideological similarity between the Chief Justice and the assignee. This suggests that a fairly blunt measure of ideology, standing alone, is insufficient to discern the interpersonal and sometimes idiosyncratic behavior of the justices.

Figure 3.

The graph presents the probability of Chief Justice Rehnquist assigning the task of authoring an opinion to a specific colleague given the degree to which that colleague influences Rehnquist as measured by the number of times Rehnquist has cited that colleague’s concurring or dissenting opinions.

To get a better sense of the substantive import of influence on the justices’ behavior, it is again instructive to determine how the moderate members of the Court are treated. Justices White and Kennedy were both ideologically proximate to Chief Justice Rehnquist (for the years in this sample, the ideological distance between Rehnquist and White ranged from 1.31 to 1.97 whereas the distance between Rehnquist and Kennedy ranged from 0.92 to 1.61). Within the sample studied by Wahlbeck, Rehnquist cited the concurring and dissenting opinions of White and Kennedy an average of 3.6 and 0.3 times per term, respectively. Holding all other variables at their means and modes, including ideological distance, the model estimates that this difference in citations would be associated with a 7% increase in the probability that Rehnquist would assign the opinion to White in a salient case compared with Kennedy (a .21 probability compared with .14).

And, indeed, Rehnquist did assign salient cases to White an average of 3 times per term compared with Kennedy’s 1.7 times per term within this sample. To emphasize the magnitude of this effect, White was in the conference majority and hence eligible to receive the opinion assignment for 89 salient cases within this time period and Kennedy for 88. Rehnquist assigned White the task of authoring 21 such salient cases, or nearly one quarter of the possible salient cases that White could author. Kennedy was assigned 12, or just 13% of the possible salient cases he could author.²⁴ Rehnquist was certainly driven by a desire to ensure his preferences for legal policy were made into law but ideology alone cannot explain the difference between how he treated Justices White and Kennedy (Kennedy was, after all, more ideologically proximate to Rehnquist over that period). I argue that Rehnquist was likely persuaded and influenced by the weight of White’s legal arguments and by his personal and intellectual regard for him. Kennedy and Rehnquist may vote similarly but Rehnquist seemed to have a deeper “meeting of the minds” with White such that he trusted him more with sensitive and important opinions. Together, these results provide additional support that my measure of influence is tapping into a more nuanced measure of ideological congruence while also capturing some of the interpersonal dynamics on the Court in ways that vote-based measures of ideology do not.

Discussion and Conclusion

The decision to cite a concurring or dissenting opinion is purely discretionary. There are no legal, strategic, or policy-driven calculations that dictate such citation practices. If justices were citing a concurring or dissenting opinion purely to provide supplemental legal support for their argument or for rhetorical flourish, there is little reason to expect them to do so in a patterned fashion. These results, however, suggest that citations to concurring and dissenting opinions reveal just such a pattern: The opinions of colleagues who are ideologically proximate, who are strategically valuable, who are personally similar, and with whom the justice has a long working relationship receive a disproportionate number of citations. Furthermore, influence as measured by citations is a significant predictor of strategic decisions that the justices make. Justices are less likely to interrupt their influential colleagues, and Chief Justice Rehnquist was more likely to assign opinions to colleagues who influenced him.

This article provides an important extension to past work on judicial influence. Measuring influence based on justices’ opinion-joining behavior (Altfeld & Spaeth, 1984) or based on their citations to majority precedent (Kosma, 1998) is theoretically unsound as both forms of behavior are constrained by strategic and institutional forces. A justice is neither wholly free to join any and every separate opinion nor free to ignore controlling precedent from majority opinions. Both measures are thus contaminated with nondiscretionary behavior and may be biased. As my empirical tests further demonstrate, a measure based on citations to separate opinions is both theoretically and practically preferable: My measure of influence produces statistically significant effects in both the Black et al. (2012) extension and the Wahlbeck (2006) extension whereas the joining-based measure of influence was significant only in the former. This theoretically and empirically sound method of measuring influence can be used to advance studies of the Court and elite behavior in several ways.

First, influence can be used to better predict strategic decisions made by justices leading up to and following a vote on the merits. A justice may be more likely to assign an opinion to, make policy accommodations for, or join a concurring or dissenting opinion with an influential colleague (Maltzman et al., 2000). The results from the Black et al. (2012) and Wahlbeck (2006) models confirm that a measure of influence can supplement, and in some situations exceed, the power of ideological proximity in models that attempt to predict and explain the justices’ strategic decisions. This indicates that interpersonal relationships, respect for a colleague’s intellectual abilities, and affinity for the content—not just the ideological direction—of a colleague’s opinions can actually guide the choices that justices make in pivotal stages of decision-making process. Using a quantifiable measure of influence can thus improve our predictive models of strategic behavior on the Court.

Second, further development in the study of judicial influence could help determine the specific content areas in which justices are most influential and develop a more nuanced view of the justices’ ideological preferences. Clark and Lauderdale (2010) use citations to precedential cases to get a better approximation of the ideological content of opinions, and Lauderdale and Clark (2014) use text analysis of opinions to get a better approximation of the justices’ ideological preferences. Combining elements of both methods, we can use text analysis of citations to nonprecedential opinions to augment measures of the justices’ preferences, especially if such an analysis controls for the issue area both in the authored opinion and the cited opinion. For example, Justice Thomas positively cited Justice Scalia’s concurring and dissenting opinions on procedural due process issues²⁵ but largely ignored or even criticized Scalia’s opinions on the Commerce Clause.²⁶ Collecting data on citations to separate opinions can help pinpoint the justices’ preferences relative to each other in specific issue areas. This, in extension, may enhance our ability to predict the actual content of written opinions. Recent work has begun to explore in-depth who influences the actual content of the Court’s opinions (Corley, 2008; Corley, Collins, & Calvin, 2011; Corley, Collins, & Hamner, 2013). Having a better sense of the Court’s thought leaders and who influences whom and in what areas could advance our understanding of where legal policy may be moving. For example, when authoring an opinion, Thomas may have been more inclined to accommodate Scalia’s revision requests in a procedural due process case than a case involving the Commerce Clause as he was swayed by Scalia’s opinions regarding the former but not the latter.

Finally, similar methods as those used here can be used to measure influential relationships on lower courts (see, for example, Klein & Morrisroe, 1999; Landes et al., 1998) and in other institutional contexts. Lower court judges have a wide array of case law to cite, and it would not be surprising if they opted, when possible, to cite precedent from cases authored by thought leaders (Weiser, 2015) or by colleagues who share their values. Similarly, legislators who make explicit references to colleagues during legislative proceedings or campaign speeches may be identifying thought leaders and influential members of Congress. Using similar methods of identifying influential relationships in other institutional contexts can help identify informal connections between elite actors—connections that exist outside of traditional power structures but which nonetheless shape actors’ behavior.

Supreme Court Justices are undoubtedly strategic political actors who attempt to maximize the likelihood that each vote will lead to their preferred policy outcome, but they are also human actors who can be influenced by the force of their colleagues’ views and personalities. The interpersonal element of judicial decision making is, on some level, more amenable to the work of biographers than political scientists. But the study of justices’ interpersonal relationships need not be dominated by anecdotal data. This article has demonstrated that systematically tracking a justice’s discretionary decision to cite a colleague’s nonprecedential decisions can be a useful and quantifiable proxy measure for influence on the Supreme Court. Such a measure can augment models of Supreme Court decision making by adding a human depth to them.

Footnotes

Appendix A

Appendix B

Appendix C

Fixed Effects Negative Binomial Count Model for Citations.

	Ideological distance–squared	No clerks	Shared × Ideological Distance	Biography subgroups
Ideological distance	−0.012	−0.072***	0.004	−0.037*
Ideological distance	(0.035)	(0.011)	(0.022)	(0.015)
Ideological Distance × Ideological Distance	−0.004
Ideological Distance × Ideological Distance	(0.005)
Shared tenure	0.009*	0.011*	0.020***	0.010*
Shared tenure	(0.004)	(0.004)	(0.006)	(0.004)
Retired	−1.153***	−1.127***	−1.147***	−1.143***
Retired	(0.075)	(0.074)	(0.075)	(0.076)
Retired × Shared Tenure	0.017**	0.015*	0.016**	0.017**
Retired × Shared Tenure	(0.006)	(0.006)	(0.006)	(0.006)
Ideological Distance × Shared Tenure			−0.004*
Ideological Distance × Shared Tenure			(0.002)
Biography	0.030*	0.030*	0.029*
Biography	(0.014)	(0.014)	(0.014)
Median justice	0.362***	0.388***	0.365***	0.372***
Median justice	(0.059)	(0.059)	(0.059)	(0.061)
Concurring or dissenting opinions	0.001***	0.001***	0.001***	0.001***
Concurring or dissenting opinions	(0.000)	(0.000)	(0.000)	(0.000)
Clerks’ ideologies	−0.317***		−0.348***	−0.318***
Clerks’ ideologies	(0.090)		(0.090)	(0.090)
Freshman	−0.010	−0.014	0.009	−0.010
Freshman	(0.080)	(0.080)	(0.081)	(0.080)
Term	−0.024***	−0.025***	−0.025***	−0.025***
Term	(0.003)	(0.003)	(0.003)	(0.003)
Personal similarities				0.049
Personal similarities				(0.030)
Professional experience similarities				0.053^†
Professional experience similarities				(0.032)
Educational similarities				0.014
Educational similarities				(0.027)
Military similarities				−0.003
Military similarities				(0.034)
Constant	48.716***	49.274***	49.509***	49.709***
Constant	(6.379)	(6.350)	(6.394)	(6.512)
Observations	3,777	3,777	3,777	3,777
Log likelihood	−6,133.891	−6,140.690	−6,131.198	−6,133.436
Wald χ²	979.71	965.16	987.42	982.17

Note. Standard errors in parentheses. These data do not include instances of justices citing their own concurring or dissenting opinions.

†

p < .10. *p < .05. **p < .01. ***p < .001.

Appendix D

Logistic Regression Estimates of Interruptions.

Did justice interrupt a speaking colleague?	Citations	Joined	Reverse citations
Ideological distance	0.058***	0.025*	0.059***
Ideological distance	(0.008)	(0.010)	(0.008)
Citations	−0.024***
Citations	(0.005)
Joined opinions		−0.050***
Joined opinions		(0.008)
Reverse citations			0.009
Reverse citations			(0.005)
Speaker previously interrupted specific colleague	3.014***	3.008***	3.014***
Speaker previously interrupted specific colleague	(0.067)	(0.067)	(0.067)
Speaker previously interrupted any colleague	0.438***	0.440***	0.439***
Speaker previously interrupted any colleague	(0.058)	(0.058)	(0.058)
Colleague’s issue area expertise	0.015***	0.015***	0.014***
Colleague’s issue area expertise	(0.002)	(0.002)	(0.002)
Speaker’s issue area expertise	−0.002	−0.001	−0.004*
Speaker’s issue area expertise	(0.002)	(0.002)	(0.002)
Case complexity	−0.011***	−0.011***	−0.011***
Case complexity	(0.002)	(0.002)	(0.002)
Speaker’s previous words	−0.000**	−0.000**	−0.000**
Speaker’s previous words	(0.000)	(0.000)	(0.000)
Colleague’s previous	0.001***	0.001***	0.001***
Colleague’s previous	(0.000)	(0.000)	(0.000)
Constant	−5.640***	−5.542***	−5.636***
Constant	(0.061)	(0.063)	(0.062)
Observations	665,780	665,780	665,780
Pseudo R²	.086	.086	.085
Log pseudolikelihood	−27,012.657	−27,009.268	−27,023.263

Note. Standard errors in parentheses.

p < .05. **p < .01. ***p < .001.

Appendix E

Random Effects Probit Regression Estimates of Assignments.

Did Rehnquist assign justice an opinion?	Citations	Joined	Reverse citations
Ideological distance	0.059***	0.063***	0.060***
Ideological distance	(0.014)	(0.014)	(0.014)
Politically salient case	0.021	0.236	0.254*
Politically salient case	(0.105)	(0.105)	(0.114)
Politically Salient Case × Ideological Distance	−0.059	−0.072	−0.069*
Politically Salient Case × Ideological Distance	(0.035)	(0.035)	(0.035)
Minimum winning conference coalition	0.160*	0.158*	0.157*
Minimum winning conference coalition	(0.079)	(0.077)	(0.079)
Minimum Winning Conference Coalition × Ideological Distance	0.084* (0.033)	0.084** (0.033)	0.084** (0.033)
Citations	−0.004
Citations	(0.011)
Politically Salient Case × Citations	0.083**
Politically Salient Case × Citations	(0.027)
Joined opinions		0.006
Joined opinions		(0.009)
Politically Salient Case × Joined Opinions		−0.015
Politically Salient Case × Joined Opinions		(0.023)
Reverse citations			0.001
Reverse citations			(0.011)
Politically Salient Case × Reverse Citations			−0.026
Politically Salient Case × Reverse Citations			(0.030)
Assignment cycle equity	−0.438***	−0.439***	−0.438***
Assignment cycle equity	(0.051)	(0.049)	(0.050)
Annual equity	0.033	0.026	0.024
Annual equity	(0.049)	(0.047)	(0.048)
Number of late majority opinion draft	−0.041*	−0.043*	−0.042*
Number of late majority opinion draft	(0.020)	(0.020)	(0.020)
Number of late dissenting opinion drafts	0.009	0.007	0.010
Number of late dissenting opinion drafts	(0.020)	(0.019)	(0.020)
Number of late votes	−0.006	−0.008	−0.007
Number of late votes	(0.012)	(0.012)	(0.012)
Constant	−1.081***	−1.103***	−1.087***
Constant	(0.053)	(0.051)	(0.052)
Observations	5,099	5,099	5,099
Wald χ²	137.67	128.05	127.97

Note. Standard errors in parentheses.

p < .05. **p < .01. ***p < .001.

Acknowledgements

Special thanks to the SPSA panel participants and panel chair, Bethany Blackstone, for their helpful comments and suggestions. Thanks also to Timothy Johnson and Paul Wahlbeck for their comments and suggestions on previous drafts. Thanks also to Benjamin Bagozzi, Amanda Bryan, Christopher Frederico, Dan Jones, Chris Kromphardt, and Adam Olson for their assistance. Finally, I want to thank the anonymous reviewers for their thoughtful comments and suggestions.

Author’s Note

A version of this article was presented at the 2015 Southern Political Science Association (SPSA) Conference.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Tom Pryor, JD/PhD, researches and writes about judicial behavior, psychology, and the law.

References

Altfeld

M. F.

Spaeth

H. J.

(1984). Measuring influence on the U.S. Supreme Court. Jurimetrics, 24, 236-247.

Baum

(2006). Judges and their audiences: A perspective on judicial behavior. Princeton, NJ: Princeton University Press.

Black

R. C.

Johnson

T. R.

Wedeking

J. P.

(2012). Oral arguments and coalition formation on the U.S. Supreme Court: A deliberate dialogue. Ann Arbor: University of Michigan Press.

Bollen

Lennox

(1991). Conventional wisdom on measurement: A structural equation perspective. Psychological Bulletin, 110, 305-314.

Bonneau

C. W.

Hammond

T. H.

Maltzman

Wahlbeck

P. J.

(2007). Agenda control, the median justice, and the majority opinion on the U.S. Supreme Court. American Journal of Political Science, 51, 890-905.

Bryan

A. C.

Johnson

T. R.

Ringsmuth

E. M.

(2013). Voting fluidity and oral argument on the U.S. Supreme Court. Political Research Quarterly, 66, 429-440.

Clark

T. S.

Lauderdale

(2010). Locating Supreme Court opinions in doctrine space. American Journal of Political Science, 54, 871-890.

Clark

T. S.

Linzer

D. A.

(2015). Should I use fixed or random effects? Political Science Research and Methods, 3, 399-408.

Clouatre

(1996). Influence relationships: Justice O’Connor and the Supreme Court. Southeastern Political Review, 24, 655-673.

10.

Corley

P. C.

(2008). The Supreme Court and opinion content: The influence of parties’ briefs. Political Research Quarterly, 61, 468-478.

11.

Corley

P. C.

Collins

P. M.

Calvin

(2011). Lower court influence on U.S. Supreme Court opinion content. The Journal of Politics, 73, 31-44.

12.

Corley

P. C.

Collins

P. M.

Hamner

(2013). The influence of amicus curiae briefs on U.S. Supreme Court opinion content. Law & Society Review, 49, 917-944.

13.

Corley

P. C.

Sommer

Steigerwalt

Ward

(2010). Extreme dissensus: Explaining plurality decisions on the United States Supreme Court. Justice System Journal, 31, 180-200.

14.

Cross

F. B.

(2010). Determinants of citations to Supreme Court opinions (and the remarkable influence of Justice Scalia). Supreme Court Economic Review, 18, 177-202.

15.

Cross

F. B.

Spriggs

J. F.

(2010). Citations in the U.S. Supreme Court: An empirical study of their use and significance. University of Illinois Law Review, 2010, 489-576.

16.

Davis

(1991). The Supreme Court: Rehnquist’s or Reagan’s? Western Political Quarterly, 44, 87-99.

17.

Diamantopoulos

Reifler

Roth

K. P.

(2008). Advancing formative measurement models. Journal of Business Research, 61, 1203-1218.

18.

Douthat

(2016, February 13). Antonin Scalia, conservative legal giant. The New York Times. Retrieved from https://www.nytimes.com/2016/02/13/opinion/antonin-scalia-conservative-legal-giant.html

19.

Epstein

Knight

(1998). The choices justices make. Washington, DC: CQ Press.

20.

Epstein

Martin

A. D.

Segal

J. A.

Westerland

(2007). The judicial common space. Journal of Law, Economics, and Organization, 23, 303-325.

21.

Epstein

Segal

J. A.

(2000). Measuring issue salience. American Journal of Political Science, 44, 66-83.

22.

Epstein

Segal

J. A.

Spaeth

H. J.

Walker

T. G.

(2012). The Supreme Court compendium: Data, decisions, and developments. Washington, DC: CQ Press.

23.

Feinerer

Hornik

Meyer

(2008). Text mining infrastructure in R. Journal of Statistical Software, 25(5), 1-54.

24.

Fowler

J. H.

Johnson

T. R.

Spriggs

J. F.

Jeon

Wahlbeck

P. J.

(2007). Network analysis and the law: Measuring the legal importance of Supreme Court precedents. Political Analysis, 15, 324-346.

25.

Gonzales v. Raich, 545 U.S. 1, 57–74 (2005).

26.

Hagle

T. M.

(1993). “Freshman effects” for Supreme Court justices. American Journal of Political Science, 37, 1142-1157.

27.

Hammond

T. H.

Bonneau

C. W.

Sheehan

R. S.

(2005). Strategic behavior and policy choice on the U.S. Supreme Court. Stanford, CA: Stanford University Press.

28.

Hansford

T. G.

Spriggs

J. F.

(2006). The politics of precedent on the U.S. Supreme Court. Princeton, NJ: Princeton University Press.

29.

Hartup

W. W.

Stevens

(1997). Friendship and adaptation in the life course. Psychological Bulletin, 121, 355-370.

30.

Jarvis

C. B.

MacKenzie

S. B.

Podsakoff

P. M.

(2003). A critical review of construct indicators and measurement model misspecification in marketing and consumer research. Journal of Consumer Research, 30, 199-218.

31.

Johnson

(2004). Oral arguments and decision making on the United States Supreme Court. Albany, NY: SUNY Press.

32.

Johnson

T. R.

Black

R. C.

Wedeking

(2009). Pardon the interruption: An empirical analysis of Supreme Court justices’ behavior during oral arguments. Loyola Law Review, 55, 331-351.

33.

Johnson

T. R.

Ringsmuth

E. M.

(2013). Supreme Court oral arguments and institutional maintenance. American Politics Research, 41, 651-673

34.

Johnson

T. R.

Spriggs

J. F.

Wahlbeck

P. J.

(2005). Passing and strategic voting on the U.S. Supreme Court. Law and Society Review, 39, 349-377.

35.

Johnson

T. R.

Wahlbeck

P. J.

Spriggs

J. F.

(2006). The influence of oral argument on the U.S. Supreme Court. American Political Science Review, 100, 99-113.

36.

King

(1988). Statistical models for political science event counts: Bias in conventional procedures and evidence for the exponential Poisson regression model. American Journal of Political Science, 32, 838-863.

37.

King

(1989). Variance specification in event count models: From restrictive assumptions to a generalized estimator. American Journal of Political Science, 33, 762-784.

38.

Klein

Morrisroe

(1999). The prestige and influence of individual judges on the U.S. Courts of Appeals. The Journal of Legal Studies, 28, 371-391.

39.

Kosma

M. N.

(1998). Measuring the influence of Supreme Court justices. Journal of Legal Studies, 27, 333-372.

40.

Kromphardt

C. D.

(2014). Fielding an excellent team: Law clerk selection and chamber structure at the U.S. Supreme Court. Marquette Law Review, 98, 289-311.

41.

Kromphardt

C. D.

(2015). U.S. Supreme Court law clerks as information sources. Journal of Law and Courts, 3, 277-304.

42.

Landes

W. M.

Lessig

Solimine

M. E.

(1998). Judicial influence: A citation analysis of federal courts of appeals judges. The Journal of Legal Studies, 27, 271-332.

43.

Lauderdale

B. E.

Clark

T. S.

(2012). The Supreme Court’s many median justices. American Political Science Review, 106, 847-866.

44.

Lauderdale

B. E.

Clark

T. S.

(2014). Scaling politically meaningful dimensions using texts and votes. American Journal of Political Science, 58, 754-771.

45.

Lupia

McCubbins

M. D.

(1998). The democratic dilemma: Can citizens learn what they need to know? New York, NY: Cambridge University Press.

46.

Maltzman

Spriggs

J. F.

Wahlbeck

P. J.

(2000). Crafting law on the Supreme Court: The Collegial Game. New York, NY: Cambridge University Press.

47.

Martin

A. D.

Quinn

K. M.

(2002). Dynamic ideal point estimation via Markov chain Monte Carlo for the U.S. Supreme Court, 1953–1999. Political Analysis, 10, 134-153.

48.

McGuire

K. T.

Vanberg

Yanus

A. B.

(2007, January). Targeting the median justice: A content analysis of legal arguments and judicial opinions. Paper presented at the Annual Meeting of the Southern Political Science Association, New Orleans, LA.

49.

McPherson

Smith-Lovin

Cook

J. M.

(2001). Birds of a feather: Homophily in social networks. Annual Review of Sociology, 27, 415-444.

50.

Meinke

S. R.

Scott

K. M.

(2007). Collegial influence and judicial voting change: The effect of membership change on U.S. Supreme Court justices. Law & Society Review, 41, 909-938.

51.

Morris

(2016, February 14). How Scalia became the most influential conservative jurist since the New Deal. FiveThirtyEight. Retrieved from https://fivethirtyeight.com/features/how-scalia-became-the-most-influential-conservative-jurist-since-the-new-deal/

52.

Murphy

W. F.

(1964). Elements of judicial strategy. Chicago, IL: University of Chicago Press.

53.

Obergefell v. Hodges, 135 S.Ct. 2071 (2015).

54.

Pennsylvania v. Muniz, 496 U.S. 582 (1990).

55.

Peppers

T. C.

Zorn

(2008). Law clerk influence on Supreme Court decision making: An empirical assessment. DePaul Law Review, 58, 51-77.

56.

Richards

Kritzer

(2002). Jurisprudential regimes in Supreme Court decision making. American Political Science Review, 96, 305-320.

57.

Savage

D. G.

(2016, February 20). Scalia’s uncompromising style at times limited his impact on the Supreme Court. Los Angeles Times. Retrieved from http://www.latimes.com/nation/la-na-scalia-legacy-20160220-story.html

58.

Schwartz

E. P.

(1992). Policy, precedent, and power: A positive theory of Supreme Court decision-making. Journal of Law, Economics, & Organization, 8, 219-252.

59.

Segal

J. A.

Spaeth

H. J.

(2002). The Supreme Court and the attitudinal model revisited. New York, NY: Cambridge University Press.

60.

Sommer

(2014). Supreme Court agenda setting: Strategic behavior during case selection. New York, NY: Palgrave Macmillan.

61.

Spaeth

H. J.

Epstein

Martin

A. D.

Segal

J. A.

Ruger

Benesh

S. C.

(2014). Supreme Court Database (Version 2014, Release 01). Available from http://supremecourtdatabase.org

62.

Streiner

D. L.

(2003). Being inconsistent about consistency: When coefficient alpha does and doesn’t matter. Journal of Personality Assessment, 80, 217-222.

63.

Totenberg

(2015, July 10). Ginsburg: Liberal justices make a point to speak with one voice. National Public Radio. Retrieved from http://www.npr.org/sections/itsallpolitics/2015/07/10/421811833/ginsburg-liberal-justices-make-a-point-to-speak-with-one-voice

64.

Wahlbeck

P. J.

(2006). Strategy and constraints on Supreme Court opinion assignment. University of Pennsylvania Law Review, 154, 1729-1755.

65.

Wahlbeck

P. J.

Spriggs

J. F.

Maltzman

(1999). The politics of dissents and concurrences on the U.S. Supreme Court. American Politics Quarterly, 27, 488-514.

66.

Wahlbeck

P. J.

Spriggs

J. F.

Sigelman

(2002). Ghostwriters on the Court? A stylistic analysis of U.S. Supreme Court opinions. American Politics Research, 30, 166-192.

67.

Wedeking

(2010). Supreme Court litigants and strategic framing. Political Research Quarterly, 5, 617-631.

68.

Weiser

(2015, May 19). Faced with legal puzzles, judges often turn to fellow jurists. New York Times. Retrieved from http://www.nytimes.com/2015/05/20/nyregion/faced-with-legal-puzzles-judges-often-turn-to-fellow-jurists.html?_r=0

69.

Weiss v. United States, 510 U.S. 516 (1994).

70.

Williams

L. A.

(2007). Measuring internal influence on the Rehnquist Court: An analysis of non-majority opinion joining behavior. Ohio State Law Journal, 68, 679-732.