Hostname: page-component-cd9895bd7-8ctnn Total loading time: 0 Render date: 2024-12-24T01:53:25.712Z Has data issue: false hasContentIssue false

The Influence of Jurisprudential Considerations on Supreme Court Decisionmaking: A Study of Conflict Cases

Published online by Cambridge University Press:  01 January 2024

Rights & Permissions [Opens in a new window]

Abstract

In this study, we assess the impact of attitudinal and jurisprudential factors on the Supreme Court's resolution of intercircuit conflicts. In doing so, we depart from earlier efforts to assess the impact of legal factors that conceptualize law as an external constraint. Instead, we view jurisprudential considerations in terms of the justices' efforts to adopt the most legally persuasive position in light of accepted methods of legal reasoning. Our analyses reveal that the justices are (1) more likely to follow the reasoning process adopted by the majority of circuits involved in the conflict, (2) less likely to adopt the conflict position marred by contrary dissents and concurrences in the circuit court opinions, and (3) more likely to adopt the conflict position endorsed by prestigious circuit court judges. Our findings suggest that jurisprudential considerations, as well as attitudinal concerns, affect the justices' decisionmaking processes in a substantial minority of cases.

Type
Articles of General Interest
Copyright
© 2006 Law and Society Association.

Introduction: Attitudinal and Jurisprudential Influences on the Judicial Choice

As a powerful policymaking body composed of unelected officials, the U.S. Supreme Court occupies a somewhat anomalous position in a representative democracy. One of the Court's most important claims to legitimacy is the proposition that its decisions are not determined solely by the justices' personal policy preferences but are influenced as well by their understandings of what “the law” requires in a given case. With so much riding on this proposition, scholars have debated it, often passionately, for years.

At this point, the position frequently labeled “attitudinalist,” holding that policy preferences overwhelm legal considerations in justices' decisions, seems dominant among quantitatively oriented scholars—and reasonably so. Not only are there firm logical grounds to doubt that precedents, statutory language, and methods of constitutional interpretation guide decisions in the extraordinary environment of the Supreme Court, but while evidence of attitudinal influences has accumulated steadily over the years, evidence of legal influences has been much harder to find. Nevertheless, many scholars continue to maintain that, even in the Supreme Court, the law matters, and “it matters dearly” (Reference Epstein and KobylkaEpstein & Kobylka 1992:302; see also Reference SmithSmith 1994; Reference GillmanGillman 2001; Reference Richards and KritzerRichards & Kritzer 2002).

We agree that the question of the law's influence is still open. Like proponents of the attitudinal model—and many other scholars—we doubt that the law often provides immutable, “right” answers to legal questions or somehow determines the outcomes of cases. Nevertheless, we suspect that justices often engage in good-faith efforts to find the most persuasive solutions to complex legal puzzles. Our aim in this study is to contribute to the law-versus-attitudes debate by reformulating the central research question and introducing a new body of evidence to address it. We ask not whether justices' decisions are traceable to specific legal texts, but whether the justices are influenced by a desire to make legally sound decisions through what they see as proper methods. We do not seek direct connections between their decisions and such manifestations of the law as statutory or constitutional language, canons of construction, or precedents. Instead, we investigate relationships between their decisions and the behavior and characteristics of other actors in the judicial process.

We proceed by examining cases decided in the 1985 through 1995 terms in which the Supreme Court resolved conflicts among the circuits. We chose to examine conflict cases for several reasons. First, as we explain below, if jurisprudential considerations play any role in the justices' decisions, they are especially likely to be evident in conflict cases. In addition, conflict cases constitute a significant portion of the Court's outputs—about one-third of its docket in any given year. And because conflict cases involve a preexisting struggle among competing legal perspectives in the circuit courts, we can specify variables that reflect jurisprudential influences and assess the impact of those variables on Supreme Court decisionmaking. We find that the justices' decisions in conflict cases are often influenced by factors that are better understood as jurisprudential considerations than as attitudinal ones.

Before proceeding, we note that while these findings are based on data from the U.S. Supreme Court, they also have implications for courts elsewhere. Not only do high courts in many other democracies face similar issues of legitimacy, but, like the Supreme Court, some other courts of last resort must resolve conflicts among competing interpretations at the lower-court level. Our research sheds some light on the manner in which that process of ultimate resolution is influenced by deliberations in the lower courts. Hierarchical institutional arrangements have the potential to influence perceptions regarding the persuasiveness of certain arguments as they percolate through judicial systems, regardless of their location.

We turn now to the theoretical underpinnings of our study and the elaboration of our hypotheses.

Legal Influences: Theoretical Considerations

Does law matter to the justices as they formulate their decisions? There are important reasons to suppose that it does. Justices may be special types of judges, but they are still judges. Judges, like other lawyers, are socialized to view certain ways of reasoning as more or less legitimate than others (Reference FishFish 1989; Reference GreenawaltGreenawalt 1992; Reference BurtonBurton 1992). Socialization begins in the first year of law school, where prospective lawyers are introduced to and expected to master “legal” methods of research, writing, and thinking. It continues throughout their career as they are continually exposed to briefs and opinions emphasizing precedents, analogical reasoning, canons of statutory and constitutional interpretation, and the like. Furthermore, when judges describe their decisionmaking in their own words, whether in interviews (e.g., Reference SaratSarat 1977; Reference HowardHoward 1981; Reference KleinKlein 2002) or books or articles they author (e.g., Reference NewmanNewman 1984; Reference PosnerPosner 1995; Reference EdwardsEdwards 1991), they tell us over and over again that they care about reaching decisions through what feel to them like professionally legitimate methods.

Obviously, these points are far from conclusive. The second point, for instance, must be treated with the same skepticism that we would bring to any self-serving description of one's own behavior. More fundamentally, many important legal questions are hard to answer. Statutes and constitutions are riddled with ambiguous language, and judges called on to interpret them frequently must choose from a menu of interpretive methods, canons of construction, and arguably relevant precedents, many of which suggest contradictory answers. No universally accepted rules exist for choosing one method, canon, or precedent to reach a decision. Nor, for that matter, do judges consider themselves legally bound to follow all conceivably applicable precedents; distinguishing precedents and narrowly interpreting holdings are generally regarded as legitimate practices. Finally, there is no guarantee that any one method, canon, or precedent will suggest the same answer to all judges who rely on it. (See Reference Segal and SpaethSegal and Spaeth 2002: Ch. 2 for an extended discussion of this argument.)

We grant these points. But a large leap is required to get from them to the conclusion that the law provides no important guidance to judges in the decisionmaking process. For even where various interpretations and approaches to a problem are possible, some will frequently be more plausible than others—not because the law mandates a particular conclusion, but because the methodological and professional conventions of judging make some arguments more persuasive than others. If this is the case with legal issues, then even if scholars cannot find reliable means to identify the more plausible positions, and even if judges often reach different conclusions, honest efforts to choose more plausible positions can still lead judges to decide cases differently than they would if they relied solely on their personal policy preferences. Indeed, the recognition that what the law requires will not always be obvious and that its effects will often be subtle has led scholars who believe that “the law” matters to argue for a shift in focus from external influences to the process of legal reasoning (e.g., Reference SmithSmith 1994; Reference GillmanGillman 2001). According to Gillman, for instance, “[a] legal state of mind does not necessarily mean obedience to conspicuous rules; instead, it means a sense of obligation to make the best decision possible in light of one's general training and sense of professional obligation” (2001:486).

A skeptic might concede that this emphasis on motivations and process makes the search for legal influences less quixotic for courts generally but still deny that it has much relevance to the Supreme Court. Reference Segal and SpaethSegal and Spaeth (2002:92–6) pointed to several key characteristics that set the Supreme Court apart from most other courts in ways that lessen the likelihood that law matters there. These include the Court's control over its own docket, its lack of electoral and political accountability, the justices' lack of ambition for higher office, and the absence of a higher court. Using Reference BaumBaum's (1997) terminology, one could argue that even if the accurate interpretation of the law is an inherent goal of justices, the unique situation of Supreme Court justices renders that goal inoperative.

We view this argument as strong but not conclusive. For one thing, many other judges possess similar levels of immunity to electoral or political repercussions. Second, lack of ambition for higher office and the absence of a higher court may reduce the significance of the goal of legally sound decisionmaking without eliminating it from the set of operative goals. Judges may pursue legal soundness not simply to avoid reversal or get promoted, but also because they see the pursuit as valuable or satisfying in its own right or because they desire the good opinion of “fellow jurists, lawyers and the public” (Reference Shapiro and LevyShapiro & Levy 1995:1058; see also Reference BaumBaum 1997:47–56).

The most important piece of the skeptic's argument is the Court's control of its docket. Together with the litigants' own de-selection of easier cases as they proceed through the judicial hierarchy, the discretionary docket ensures that the typical case decided by the justices will be far more legally ambiguous and more politically salient than the typical cases found on other courts' dockets. Still, the funneling process and discretionary docket do not guarantee that all cases reaching the Court are perfectly ambiguous. Furthermore, even a cursory examination of the Supreme Court's docket reveals that the policy implications of the cases are not always salient enough to overwhelm all other considerations. Take, for example, the particular subject of this study—conflict cases. Issues giving rise to intercircuit conflicts often involve highly technical statutory matters. It seems likely that the justices choose to hear them not because they care so much about the policies involved but in order to clarify federal law and promote uniformity in its implementation.

Reference PerryPerry's (1991) research supports this view. In particular, Perry has argued that justices' decisionmaking on certiorari can usefully be thought of as falling into one of two categories: the “outcome mode” and the “jurisprudential mode.” In the outcome mode, certiorari votes are determined primarily by the justices' views of how a case should be decided on the merits. In the jurisprudential mode, the justices' preferences concerning policy outcomes are subordinated to such considerations as whether a circuit court conflict has arisen that must be resolved in order to ensure uniformity in federal law. Although Perry did not investigate decisions on the merits, it is reasonable to infer that the mix of considerations that leads justices to place a case on the docket will carry over, at least in part, to the vote on the merits. If so, then the influence of policy preferences will likely be less dominant in some cases than in others, leaving more room for legal goals to play a role.Footnote 1

Legal Influences: Existing Evidence and Our Approach

The foregoing discussion may provide grounds for believing that legal goals can be operative even at the Supreme Court, but it falls far short of proving that legal goals actually do affect justices' decisions. Proof must come from the empirical evidence. Quantitative evidence has offered little comfort to those who believe legal goals matter. For one thing, the myriad studies showing strong relationships between judges' voting and measures of ideology cannot help but generate skepticism about the law's role. For another, Reference Spaeth and SegalSpaeth and Segal (1999) demonstrated convincingly that most justices are unwilling to treat precedents from which they dissented as binding. True, the authors searched for only one type of legal influence, but theirs is the most systematic such search to date, and it found little evidence of any. Reference Howard and SegalHoward and Segal's (2002) finding that arguments concerning plain meaning or framers' intent have had little systematic influence on the justices' votes similarly encourages skepticism about legal explanations.

The strongest arguments for the importance of legal influences have typically come from interpretive qualitative studies. To take a few prominent examples, Reference Epstein and KobylkaEpstein and Kobylka (1992) concluded from their in-depth analysis of the Supreme Court's death penalty and abortion decisions that the early stages of doctrinal development in these ideologically charged areas were substantially influenced by the arguments the justices heard and made (1992:302), while Reference GillmanGillman's (1993) and Reference CushmanCushman's (1998) historical analyses and close readings of controversial decisions in the Lochner and New Deal eras bolstered claims that these decisions did not merely reflect justices' policy preferences but arose from honest attempts to apply broader legal principles.

On the quantitative side, some evidence of legal influences comes from Reference WahlbeckWahlbeck's (1997) discovery that the “weight” of precedent had an effect on justices' decisions to invoke legal change; as the number of consistent legal rulings on a given issue increased, the likelihood that the Court would initiate a change in legal norms was reduced. Most intriguingly, Reference Richards and KritzerRichards and Kritzer (2002) uncovered evidence that a “jurisprudential regime change” affected justices' votes in First Amendment cases. Comparing opinions before and after the Court's decision in Grayned v. Rockford (1972), which changed the legal parameters for evaluating governmental restrictions on speech, they found significant changes in the effects of some variables; for example, the fact that restrictions were content-based had more effect on the likelihood that they would be invalidated after Grayned than before.

Our aim in this study is to generate new quantitative evidence about the importance of legal influences. Both our theoretical perspective, described earlier, and our reading of existing evidence lead us to believe that we may learn more by seeking evidence of legal influences in large patterns than by examining connections between specific legal texts and decisions. Hence, instead of assessing the changing influence of case factors or seeking other direct evidence of causal links between sources of law and justices' decisions, we begin with the idea that the justices might be motivated to reach legally sound decisions and ask what behaviors and relationships we would expect to observe if they were. In doing so, we assume that judges who are so motivated are likely to be influenced by the persuasiveness of legal argumentation and to view themselves as members of an interpretive community—that is, as part of a group of people capable of and willing to evaluate legal arguments according to some shared standards. In the next section, we explain how these somewhat abstract ideas can be translated into operational hypotheses.

Hypotheses

Our aim in this article is to test the proposition that justices' voting behavior is influenced by their desire to reach legally sound decisions. To do so, we examine cases in which the Court resolved an intercircuit conflict by choosing the legal rule favored by one set of circuit courts over that favored by another.Footnote 2 Since the overwhelming majority of conflicts involve only two viable legal positions, we treat the Supreme Court's decision as a choice between two teams. The Respondent's Team includes the circuit that decided the case before the Court and all other circuits that took the same position on the conflict issue. The Petitioner's Team is made up of those circuits disagreeing with the decision in the case below. Concentrating primarily on team composition, we identify team characteristics that should be related to the justices' choice of one position over the other if they are acting on the goal of legal soundness.

Jurisprudential Hypotheses

Our first hypothesis is that justices will favor the position favored by the larger team. More formally,

H1: The greater the number of circuits on the Petitioner's Team relative to the number of circuits on the Respondent's Team, the more likely justices are to vote for the legal position espoused by the Petitioner's Team.

There are two chief reasons for believing that this relationship should hold if the justices are trying to make good law. First, given that all judges have received similar training in how to evaluate legal arguments, if circuit judges and justices are engaged in a common enterprise to make legally sound decisions, then a majority of the justices should tend to decide in the same way as a majority of the circuits just because they view the issues as the circuit judges do. Second, the justices may be directly influenced by what happens in the lower courts. Most obviously, they may be persuaded by circuit judges' arguments. The more judges writing opinions in defense of a position, the better the chance that one of them will write something that convinces a justice. In addition, the numbers of circuits on each side of a conflict may serve as a cue to the justices. For instance, justices facing a difficult issue might view a 5-1 circuit split as strong evidence that the majority position is more legally defensible.

For our next hypothesis, we turn from consensus to dissensus. Dissents and concurrences are much rarer events in the courts of appeals than in the Supreme Court. For example, more than 84 percent of the circuit court decisions we examined were unanimous, even though every one involved an issue that was contentious enough to split the circuits. The rarity of separate opinions is probably due at least in part to the larger caseloads in the circuit courts; it may also reflect a norm of comity (see Reference HettingerHettinger et al. 2003). Whatever the reason, we would typically expect to find judges expressing disagreement only when they feel quite strongly that the majority is wrong. Naturally, their feelings will often have an ideological basis, but if judges act on the goal of legal soundness, then they will frequently reflect legal reasoning as well.

What are the implications for justices who care about legal reasoning? First, a dissent or disagreeing concurrence might signal a problem, encouraging the justices to view the majority's argument more skeptically. Review of the Supreme Court's opinions in these cases reveals that the justices are indeed cognizant of dissenting (and sometimes concurring) opinions that attack a particular panel's legal reasoning. Second, even when the justices do not take special note of disagreeing opinions, if those opinions reflect weaknesses in a team's position, then that position may be less likely to persuade the justices. For both reasons, if justices attempt to choose the more legally convincing position, then:

H2: In resolving an intercircuit conflict, justices are less likely to side with the conflict team whose position has generated more contrary dissenting and concurring opinions.

In addition to noting concurrences and dissents in the circuits, the Supreme Court's opinions occasionally direct special attention to individual circuit court judges' opinions (majority or separate). In light of previous research concerning the influence of prestigious judges and courts (Reference CaldeiraCaldeira 1985; Reference Klein and MorrisroeKlein & Morrisroe 1999), these references are not surprising. Apparently, some judges enjoy a special respect among their colleagues, and their opinions carry greater clout. If the justices are motivated to find legally persuasive solutions, they might reasonably choose to give extra weight to the views of these more respected judges. Even if they do not, to the extent that circuit judges' reputations reflect their skills in reasoning and argument, their opinions may be more persuasive to the justices. At the least, highly respected judges may be less apt to adopt positions that other judges (including the justices) find implausible. For these reasons, our third jurisprudential hypothesis is this:

H3: The greater the prestige of the judges on the Petitioner's Team relative to those on the Respondent's Team, the more likely justices are to side with the Petitioner's Team.

Attitudinal Hypotheses

As explained in “Legal Influences: Theoretical Considerations” above, conflict cases may elicit a less pronounced ideological response from justices than other cases do. Nevertheless, because ideology is such a powerful force in Supreme Court decisionmaking generally, we expect it to have a significant effect on the decisions examined here. In the context of intercircuit conflicts, we conceptualize the influence of Supreme Court ideology in a straightforward manner: we expect that justices will be more likely to support the team whose policy position on the conflict issue is most consistent with the justices' policy predilections. Accordingly, we hypothesize that:

H4: In resolving an intercircuit conflict, justices are more likely to side with the conflict team whose position is ideologically closer to their policy preferences.

The Solicitor General

Our final hypothesis is that:

H5: In resolving an intercircuit conflict, justices are more likely to side with the team, if any, that is supported by the Solicitor General.

It is well established that the Solicitor General (SG), the federal government's litigator in the Supreme Court, enjoys considerable success at the agenda-setting stage. Some research indicates that the SG's success continues at the merits stage as well, whether as party or amicus (Reference Handberg and HillHandberg & Hill 1980; Reference Spaeth, Teger, Halpern and LambSpaeth & Teger 1982; Reference SheehanSheehan et al. 1992; Reference McGuireMcGuire 1995). The SG's influence may be related to a number of factors. For instance, justices may give greater deference to the views of the executive branch than to those of other parties. Reference McGuireMcGuire (1998) has argued that the SG's success is attributable largely to the expertise of lawyers in the SG's office. As we discuss in more detail later, the variety of explanations for the SG's success keeps us from designating the SG's position as an attitudinal or jurisprudential influence prior to the analysis. We present empirical tests of different explanations following the primary analyses.

Data and Methods

Dependent Variable

To test the hypotheses set forth above, we first used the Spaeth Supreme Court Database to identify all cases in which the Supreme Court granted certiorari to resolve an intercircuit conflict for the terms 1985 to 1995. Our initial search yielded 384 cases. We eliminated any that did not involve a conflict among the circuit courts (such as where a circuit was in conflict with a state court), and any in which the Supreme Court did not adequately identify the circuits involved in the conflict itself. In the end, we had 338 conflict cases, yielding 2,988 individual votes for analysis.Footnote 3 Our dependent variable was scored 1 if the justice voted for the position taken by the petitioner and 0 if the justice voted for the respondent's position.

Independent Variables

To construct our independent variables, we began by recording all circuit court cases identified by the Supreme Court's opinion as involved in the conflict. After reading each circuit court case (a total of 1,501, or about 4.5 per conflict), we assigned it to either the Petitioner's or the Respondent's Team, depending on whether the court's decision in the case was consistent with the petitioner's argument in the Supreme Court or the respondent's. Since our dependent variables were structured with reference to the Petitioner's Team, we calculated our independent variables by comparing the Petitioner's Team with the Respondent's Team on a number of different dimensions. We explain each in turn.

Team Size (H1)

To construct the variable Team Size Differential, we simply subtracted the number of circuits on the Respondent's Team from the number of circuits on the Petitioner's Team. Where different panels from the same circuit took different sides on the issue, we counted that circuit for each team. We expected this variable to have a positive effect on the probability that the Petitioner's Team would win the vote of individual justices.

Dissenting Opinions (H2)

We read each dissenting and concurring opinion in the circuit court cases to analyze whether it addressed the legal issue involved in the conflict and, if so, whether it challenged the majority's position on the issue. Occasionally, a circuit judge would dissent from the circuit's denial of an en banc hearing, and that dissent would be appended to the panel decision in the Federal Reporter. In such cases, we also recorded the dissent to the en banc denial. For each team, therefore, we were able to calculate the number of dissenting and concurring opinions that challenged the team's legal position. To construct the Dissent Differential variable, we subtracted the number of such separate opinions on the Respondent's Team from the number of separate opinions on the Petitioner's Team. Because a higher score indicated more dissension on the Petitioner's Team's side, the expected effect of this variable was negative.

Judicial Prestige (H3)

Our third hypothesis relates to judicial prestige. To measure the concept, we followed the approach of Reference Klein and MorrisroeKlein and Morrisroe (1999), with slight modifications suggested by Reference Bhattacharya and SmythBhattacharya and Smyth (2001). Because the process of constructing the variable was complicated, we give only a sketch of it here. A full description is provided in Appendix 1.

We began by calculating a prestige score for each circuit court judge who wrote a majority, concurring, or dissenting opinion in our dataset. The key element of this score was the number of times the judge was cited by name by judges outside his or her own circuit in a certain period. Put simply, the logic behind the measure is as follows: references to other judges by name when their opinions are cited are rare and typically unnecessary. Therefore, it is likely that the choice to cite a judge by name reflects something special about that judge. The most plausible explanation is that the cited judge is particularly respected and the citing judge wishes to associate the prestige of the cited judge with his or her own opinion.

To generate a score for each judge, we searched the Lexis database for all cases in which the judge was cited by name in another circuit court opinion during the period 1989–1991,Footnote 4 distinguishing between citations to majority opinions and citations to separate opinions. Following Reference Klein and MorrisroeKlein and Morrisroe (1999), we valued a named citation to a separate opinion as equal to 0.27 of a named citation to a majority opinion. We then divided the total number of named citations for each judge by the opportunities that each judge had to be cited—that is, by the number of opinions he or she had written by 1990, discounted according to the age of the opinions. Two small adjustments, as described in Appendix 1, completed the process.

At this point, for each conflict we had prestige scores for all judges who wrote opinions (majority, concurring, or dissenting) defending the Petitioner's Team's position and all those who wrote opinions in support of the Respondent's Team. We would not expect justices to look any more skeptically on a position just because a little-known or little-respected judge happened to support it, nor that this judge's arguments could somehow make the position less persuasive. It is the positions of the most prestigious judges that should matter most. Thus, we subtracted the highest prestige score among the judges on the Respondent's Team from the highest prestige score among the judges on the Petitioner's Team. The resulting measure, Prestige Differential, was expected to have a positive effect on the Petitioner's Team's odds of winning a justice's vote.

Ideology (H4)

We hypothesized that the justices' ideologies would be related to their support for a conflict team's legal position. We investigated a number of different Supreme Court ideology measures, including Poole's Presidential W-Nominate scores for the justices' appointing president (see Reference Poole and McCartyPoole & McCarty 1995) and the “dynamic” ideal point measures for individual justices developed by Reference Martin and QuinnMartin and Quinn (2002). The results presented here are from the model using the Martin-Quinn measures because they provide the most generous measure of ideology, given that they are constructed from the justices' voting behavior itself.Footnote 5

To test Hypothesis 4, we coded the Petitioner's Team's position on the conflict issue employing the traditional criterion of the party favored by the decision. (Our coding rules were the same as those used by Spaeth in constructing the Supreme Court Database, except that our coding focused on the legal rule adopted and the party that would tend to benefit from it, rather than on the actual outcome of the case for the parties; the number of different coding outcomes was minimal and had no effect on the results.) Conservative positions were scored as 1 and liberal positions as −1, to match the Martin-Quinn scoring. We constructed the variable Ideology by multiplying the score for the petitioner's position by the justice's Martin-Quinn score for the preceding term (or for the present term, if it was a justice's first). As a result, the more highly positive values on this variable were, the more likely it was that a justice would prefer the petitioner's position on ideological grounds. A negative value would indicate that the justice's ideological preferences were inconsistent with the petitioner's position on the conflict issue. Thus, we expected this variable to have a positive coefficient.

Solicitor General

From the Court opinion, we recorded whether the SG supported a given team either as amicus or by representing the United States as a party. The variable SG for Petitioner was scored 1 if the SG supported the petitioner's position and 0 otherwise. Its predicted effect was positive. The variable SG for Respondent was scored 1 if the SG supported the respondent's position and 0 otherwise. Its predicted effect was negative. The default category was no SG participation in the case.

Descriptive statistics for our dependent and independent variables are presented in Appendix 2.

Results

Since our dependent variable was dichotomous, we selected logit as our estimation technique (Reference Aldrich and NelsonAldrich & Nelson 1984). Because justices' votes in a particular case are not necessarily independent of each other (the justices may influence each other or be commonly influenced by some case-specific variable not included in our model), conventional standard errors can be misleadingly small. To avoid this problem, we specified robust standard errors, using the “cluster” function in Stata 8, with case as the clustering variable.

The results from our model of votes in Supreme Court conflict cases are reported in Table 1. As expected, justices' ideologies strongly influence their votes. The Solicitor General also has the predicted effects. More important for this article, all three of our jurisprudential variables have statistically significant coefficients in the predicted directions.Footnote 6 A justice is more likely to vote for the petitioner's position where (1) the circuits taking that position outnumber those on the other side, (2) the number of circuit judges rejecting that position in concurrences or dissents is smaller than the number dissenting on the other side, and (3) more highly prestigious circuit judges have written opinions supporting the petitioner's position than the respondent's.Footnote 7

Table 1. Logit of Individual Justices' Support for the Petitioner's Team on Jurisprudential and Attitudinal Variables

N=338. Goodness of fit: chi-square=95.93 (df=6), p<0.0001. Significance values are calculated using one-tailed tests.

Because logit coefficients are not directly interpretable, in Table 2 we provide estimates of the impact of changing values of selected independent variables on the probability that a justice will vote for the legal position favored by the Petitioner's Team. Specifically, for the variables not involving the SG, we examined the effects of moving from the 10th percentile value of that variable to the median value and from the median to the 90th percentile value. The SG variables have only two possible values each. Aside from the variable being manipulated, all variables are held at their medians for purposes of calculating these predicted probabilities.

Table 2. Estimated Change in Probability That a Justice Will Support the Petitioner's Team, Given Specified Change in Variable

Note: All other variables are held at their medians. Estimates were generated using the Clarify program (Reference TomzTomz et al. 1998) in Stata 8.0. Probability of the Petitioner's team winning with all variables held at their median is 0.505.

To take an example, Table 2 shows that, for an otherwise typical case, the probability that a justice will vote for the Petitioner's Team is about 4 percent higher when that team is the same size as the Respondent's than when it has two fewer members. The justice is about 6 percent more likely to support the Petitioner's Team when it outnumbers the Respondent's by three than when they are the same size. The effects of dissenting opinions are about the same. The support of prestigious judges has slightly larger effects.

Initially, one might be inclined to view these effects as too small to be important. We think this conclusion would be mistaken for two reasons. First, the numbers presented reflect the result of changing just one variable at a time, with all others held constant. But in the real world, cases often differ from each other on more than one variable. Where two or more of our jurisprudential variables push in the same direction, their combined effects can be substantial. And more often than not, these variables will push in the same direction. This is particularly true of the team-size and prestige variables, which are moderately strongly correlated with each other (r=0.28). So, for example, if we begin by setting all variables at their medians and calculate the effect of moving both the team-size and prestige variables from their medians to their 90th percentile values, the estimated change in probability of a pro-petitioner vote is 0.12 (st. error=0.03). If we allow all three of our jurisprudential variables to move together (the dissent variable moving to its 10th percentile value, since its effect is negative), the estimated change in probability is 0.16 (st. error 0.04).

Second, the numbers are estimates for individual justices. Typically, nine different justices vote on a case. Predicted effects will be slightly smaller for different justices at different times, but all are predicted to be subject to at least some influence. In close cases where one or more justices are unsure of their positions, even a fairly small difference in one of our jurisprudential variables could change not only the votes of one or two justices but the outcomes of the cases. Larger differences or differences involving more than one of the jurisprudential variables are likely to be consequential in many cases.

To illustrate this point, consider two hypothetical cases decided during the Court's 1990 term, the middle term in our sample. The first is the most common type of case: the SG takes no position, the circuits and dissents on each side are equal in number, and the petitioner advocates a liberal position. In this case, the predicted probabilities of voting for the petitioner range from 0.40 (Rehnquist) to 0.71 (Marshall). The second case is different in only one way: the prestige variable is at its 90th percentile value rather than its median. Three justices who would be predicted to lean toward the respondent in the first case would now be predicted to lean very slightly toward the petitioner (Kennedy, O'Connor, and Souter). If instead we change the prestige variable from its median to its 10th percentile value instead, the model predicts a change in Blackmun's leaning from pro-petitioner to pro-respondent and predicts that Stevens would move from a pro-petitioner position (0.56) to essentially undecided (0.503). Moving the prestige and team-size variables together from their medians to either their 10th or 90th percentile values would be sufficient to change the leanings of every single justice except Marshall.Footnote 8

Our claim is decidedly not that these effects are strong enough to overwhelm attitudinal influences. After decades of research demonstrating the importance of ideology in Supreme Court decisionmaking, we would have been shocked to find such strong effects and could not have asked readers to believe them. The interesting question in a world where ideology is known to be a powerful influence on Court decisions is whether jurisprudential factors can affect justices' votes and litigant outcomes in a non-negligible number of cases. The results here suggest that they can.Footnote 9

Alternative Attitudinal Explanations for Jurisprudential Effects

Taking all of the results together, then, we believe they provide strong evidence that jurisprudential influences matter for justices' decisions in conflicts cases. Three variables—team size, dissent, and prestige—were identified as jurisprudential from the start. We would also argue that the large effects for the SG are at least partly attributable to jurisprudential considerations. In this section, we assess each variable in turn, giving reasons for our jurisprudential interpretations but also considering possible attitudinal explanations for the findings. Of course, the fact that we controlled for ideology in our analyses using a particularly generous measure (based on justices' past votes) must cast doubt on any attitudinal explanation, but it is still worth the effort to test these alternatives as directly as possible.

Team Size

We think there are three plausible interpretations for the tendency of larger conflict teams to do better at the Supreme Court: (1) the justices consciously or unconsciously see the number of circuits on a side as an indicator of the strength of that team's position; (2) the team with more members, having more chances to hit upon a persuasive argument for its position, more often succeeds in doing so; or (3) one position gains more supporters in the first place because it has greater legal plausibility, and the justices tend to choose the more plausible position. Although slightly different, all of these interpretations share a very important implication—that circuit judges and Supreme Court justices share a desire to reach legally sound decisions and that this desire affects their decisions.

Someone who believes that ideological considerations always dominate the justices' behavior could offer two alternative explanations of the team-size effect. First, the courts in the period examined here were dominated by appointees of Republican presidents. (More than 60 percent of the circuit judges in our cases were appointed by Republicans.) Perhaps a shared conservative ideology accounts for the correspondence between the decisions of the circuits and those of the Supreme Court. We can test this explanation in our model by asking whether the team-size variable has a greater effect for conservative justices. We do this by creating an interactive term by multiplying the team-size variable by justices' ideology scores and adding this term (and the ideology scores) to our original model. If this attitudinal explanation of team-size effects is correct, the interactive term should have a positive coefficient and the effect of the team size differential should become insignificant. Neither happens. Team size is still significant, and the coefficient for the interaction term is −0.015 (st. error = 0.011). As an additional test, we collapse ideology into a dummy variable, where 1 = conservative and 0 = liberal, create a new interactive term by multiplying this dummy variable by the team-size variable, and add these two variables to the original model. Again, the interactive term should have a positive effect, but it does not. In fact, its negative coefficient is statistically significant (−0.82, st. error = 0.041)! The direct effect of team size naturally remains significant. If anything, liberal justices are more likely to side with lower court judges. The alternative explanation fails.

The second alternative is that the team-size effect reflects successful attempts by circuit judges to read and anticipate the policy preferences of Supreme Court justices and decide cases as they would. To test this alternative, we begin by recognizing that (1) there is wide variation in the amount of time that passes between a circuit court decision and the Supreme Court's resolution of a conflict issue, and (2) circuits deciding closer in time to the Supreme Court's resolution should be able to anticipate that resolution more accurately, since they possess more up-to-date information about the justices' preferences and significant membership change is less likely to intervene. Thus, if successful anticipation of the Supreme Court is producing the team-size effect, we should find that panels deciding closer in time to the Supreme Court's resolution of the conflict are more likely to be on the winning team. To evaluate this explanation, in our dataset of circuit court cases we create a variable for each case indicating the number of years that passed between the circuit judges' decision and the Supreme Court's resolution of the issue. We then run a logit with this as the independent variable, the dependent variable being whether the circuit decision was on the side that ultimately won at the Supreme Court (n = 1501). The results provide no support whatsoever for the attitudinal explanation: the coefficient, 0.0031, is not only very small, but in the wrong direction. If we replace our original independent variable with a dummy variable in which cases decided within two years of the Supreme Court decision are scored as 0 and all others as 1, the coefficient is correctly signed but indistinguishable from zero: −0.0007, st. error = 0.104, p (one-tailed) = 0.473. No other versions of the independent variable work any better. Circuit court decisions coming closer in time to the Supreme Court's decision are no more likely to be on the winning team. For this reason and the others given above, we think that our jurisprudential interpretation of the team-size effect is far more compelling than an attitudinal explanation.

Dissenting Opinions

We believe that a theoretical interpretation of the dissenting opinion effect is fairly straightforward. Dissenting opinions typically identify faults in the majority's legal analysis, thus undermining its persuasiveness. Often, a dissenting opinion will actually cause the conflict in the first place, as later circuits considering the issue adopt the reasoning in the earlier dissent. We think it unlikely that the Court counts the number of dissenting opinions in assessing the strength of the teams' legal positions. But in writing their opinions, the justices do appear to take note of individual dissenting opinions from time to time, thus indicating the justices' awareness of the separate opinions' content. Moreover, as with team size, this variable may reflect an underlying dimension related to persuasive legal argumentation: dissents are more likely to accompany less persuasive rulings.

An attitudinal explanation for the effect of dissenting opinions might go something like this: writing a separate opinion takes time and attracts attention to oneself, so it should be less attractive to judges who support a position the Court is likely to reject and—because they are more likely to support such a position—less attractive to judges who are ideologically distant from the Supreme Court. (Not only would the dissent have little chance of persuading the Court, but it might backfire by causing justices to view the dissent's author with suspicion in the future.) Because of this self-selection effect, the dissents would tend to appear where the Supreme Court is already likely to oppose a position on ideological grounds.

Two simple tests can help us determine how credible this interpretation is. First, since the Supreme Court's leanings were clearly conservative across the entire time period studied here, the attitudinal interpretation would predict that minority opinions came predominantly from conservative judges. We test this proposition by assigning each circuit judge in our sample his or her appointing president's W-Nominate score developed by Poole (see Reference Poole and McCartyPoole & McCarty 1995). These scores are centered around 0, with liberal presidents receiving negative scores and conservative presidents receiving positive scores. Contrary to the prediction of the attitudinal alternative, dissenters were no more conservative than majority opinion writers. The mean ideology score of the former group, 0.069, is virtually identical to the mean ideology score of the latter, 0.073.

The second test asks whether circuit judges dissented more often from liberal rulings than conservative ones, as we would expect if they were acting strategically with regard to Supreme Court ideology. They did not. They dissented from 15.8 percent of liberal decisions and 15.2 percent of conservative decisions. This difference does not approach statistical significance (chi-square = 0.11; p = 0.74, n = 1501). Again, the data provide no support for the attitudinal explanation.

Prestige

We see two possible interpretations of the finding that the team with the most prestigious judge does better at the Supreme Court. First, the Court could use the identities of highly prestigious judges as cues in the decisionmaking process. When the justices see that Judge Friendly, Wisdom, or Posner has decided a conflict issue in a particular way, they may see that as a point in favor of that judge's team. However, we may assume that prestigious judges are prestigious for a reason: because their legal reasoning and/or argumentative skills are superior to those of less prestigious judges, or for other reasons. If so, they should be more likely to choose legally sound positions and should have a greater capacity to persuade other judges, regardless of whether those other judges know of their reputations. Especially as Reference Klein and MorrisroeKlein and Morrisroe (1999) have shown that their measure of prestige is unrelated to ideology, we can think of no plausible attitudinal explanation for this finding.

Solicitor General

As we noted earlier, the SG's success before the Court could be explained in several ways. These explanations might usefully be grouped into three categories: jurisprudential, attitudinal, and what we call institutional. One jurisprudential explanation would be that the exceptionally able and experienced lawyers in the SG's office are more likely to persuade the justices. Another would be that these lawyers are more likely to identify and adopt the legally stronger position. An institutional explanation might hold that the SG's position as representative of the U.S. government influences the justices' decisions, because they believe it proper to defer to the federal government or because they fear some kind of retaliation from the executive. We cannot distinguish between these explanations empirically here. But we can assess the persuasiveness of attitudinal explanations for the SG's success.

Two attitudinal explanations are most likely: (1) because of presidential appointments, the policy preferences of the SG and the justices tend to correspond; and (2) the SG acts strategically, choosing not to defend positions that are incompatible with the Court's ideological leanings. Both explanations can be tested using interactive terms.

To see whether ideological agreement accounts for the SG's influence, we start by creating a new dummy variable with the value 1 for conservative justices when the SG is appointed by a Republican and for liberal justices when the SG is a Democrat, and 0 otherwise. We then collapse our two SG measures into one, scored −1 if the SG supports the respondent, 0 if he or she takes no position, and 1 if he or she supports the petitioner. We then multiply these two variables together to form an interactive term. (The single measure of the SG's position is technically less appropriate than the two separate measures, but it provides a fairer test for the interaction by avoiding excessive collinearity.) If the attitudinal explanation is correct, then the interactive term should have a positive effect and the direct effect of the SG positions should no longer be significant. The results provide some support for the attitudinal alternative: the interactive term has a significant positive effect (0.238, st. error = 0.114). But the direct effect remains significant and is at least as strong (0.308, st. error = 0.113). This means that the SG's effect is especially strong for like-minded justices, but even justices who are presumably ideologically distant from the SG are quite likely to vote for the side he or she supports.

One way to test the second explanation—strategic choices by the SG—is to ask whether the federal government's effect is limited to, or at least much stronger in, cases where it participates voluntarily. We code the SG's participation as involuntary (0) where the United States is the respondent or the Court “invites” the SG to submit an amicus curiae brief and voluntary (1) in all other cases where the SG submits an amicus brief and all cases where the government is the petitioner. Once again, we create an interactive term, this time by multiplying the voluntariness dummy by the trichotomous SG variable introduced in the previous paragraph. The results provide no support at all for the proposition that the SG is on the winning side more often simply because he or she chooses to participate in cases where his or her position is likely to coincide with a majority of the justices'. The direct effect of the SG position remains significant while the interaction term, contrary to prediction, is negative. The SG's position is no more likely to be adopted when put forward voluntarily.

We are not confident that this test is conclusive; it still may be that strategic ideological choices play an important role in the SG's influence. However, given the results of both tests, we think it highly likely that jurisprudential and/or institutional considerations account for some of the SG's success in conflict cases.

Conclusions

The objective of this study was to investigate the impact of jurisprudential considerations on Supreme Court decisionmaking. We chose to study cases in which the Supreme Court resolved conflicts because we thought the influence of jurisprudential considerations was likely to be greater in such cases and because they provided a context in which we could measure those influences. We believe that our research reveals important information about the resolution of intercircuit conflicts, opening a window on the influence of such variables as judges' prestige and dissenting behavior. Most fundamentally, we think our results strongly support the view that judges and justices engage in sincere efforts to find solutions that are persuasive according to a commonly held set of criteria.

This is not to deny that the justices' personal values are influential in conflict cases. Nor do we claim to have found direct evidence that the law as an independent entity shapes or constrains the justices' decisions. Our argument and inferences concern only the justices' motivations or mental processes. In fact, even considered only with reference to our argument, our evidence is largely inferential. We initially attempted to develop a valid way of measuring the quality of an opinion's legal arguments, and we hope that we or some other scholars will do so in the future, but we were not able to for this project. Instead, we relied on measures that on their face do not appear to involve the law, arguing that their effects are most plausibly interpreted as arising from jurisprudential considerations.

Ultimately, of course, we recognize that our findings cannot be generalized to Supreme Court decisions that do not involve conflicts. But conflict cases have made up about 30 percent of the Court's docket in recent years—a substantial proportion. Consequently, we think it is fair to characterize our evidence as suggesting that the desire to find legally sound, persuasive solutions to legal questions plays a significant role in Supreme Court decisionmaking. At the least, it provides new support for those scholars who argue that we should look beyond ideology in trying to understand the justices' decisions.

Moreover, this study sheds light on the decisionmaking dynamics within a multitiered judicial hierarchy. Our findings indicate that decisions at individual levels within the federal judicial system may be interdependent. Although we cannot identify precise causal influences, these findings suggest that the justices may consider information associated with decisionmaking processes in lower courts in formulating their perspectives about an appeal. At the very least, they suggest that the justices are influenced by the same factors that affect lower court judges' choices between two competing legal rules. If the justices are influenced by the choices made by other judges, it suggests the importance of viewing judicial decisionmaking not as a solitary activity but rather as one shaped by the judicial system as an institutional unit. Since multitiered court systems are common throughout the individual states and in other nations, this conclusion points to the importance of considering courts as organizations and recognizing the potential impact of organizational structure on the development of legal norms.

Appendix 1: Measuring Judicial Prestige

Constructing a measure of prestige for individual judges presents three major challenges. Below we explain how we dealt with these challenges, after which we describe other steps in the calculation of our measure.

The first challenge we faced was how to count name citations of a judge's concurring or dissenting (separate) opinions. As we have explained, it is unnecessary and rare for circuit judges to refer to each other by name when citing majority opinions. By contrast, convention requires that one identify the author of a concurring or dissenting opinion when citing it. For this reason, it might seem that separate opinions should not be counted at all. But it is possible that judges are selective in citing concurrences and dissents; they might hesitate to cite those from less prestigious judges and welcome the chance to cite well-respected judges. If so, citations to majority opinions should be correlated with citations to separate opinions. In an earlier study involving prestige, Reference Klein and MorrisroeKlein and Morrisroe (1999) found strong evidence that this was so. For the reasons given in that article, we adopt their convention of treating a citation of a concurrence or dissent as 0.27 of a majority opinion citation. We are under no illusion that this approach perfectly reflects reality, but we are certain that it is a better alternative than excluding separate opinions entirely or counting them the same as majority opinions.

The second problem we confronted was how to avoid biases in favor of judges who have been on the bench longer or simply write more opinions. For both types of judge, colleagues have more opportunities to cite their work. Their names might come up in opinions more often than others' simply because their cases do. We decided to take opportunities into account by dividing each judge's citation score by the total number of opinions he or she had written through 1990. This simple adjustment would not be quite correct, though. Explaining why brings us to a discussion of our third problem.

On average, older opinions are less often cited than more recent ones (Reference Landes and PosnerLandes & Posner 1976), and older opinions would form a larger proportion for judges who have been on the bench longer. Simply dividing by the number of opinions would penalize these judges. (For instance, a judge who wrote four hundred opinions in the 1970s would be treated as having as many opportunities as one who wrote four hundred opinions in the 1980s, even though the latter judge's opinions have a much better chance of being cited.) To measure opportunities for citation realistically, we must discount them according to their age.

To determine an appropriate discount, we took a random sample of 30 circuit court cases for each year between 1969 and 1988, used LEXIS to ascertain the number of times each case was cited between 1989 and 1991, and calculated the mean number of times cited for each year. We then ran the following log-linear regression:

lnY=B(1)+B(2)t+u,

where lnY = the natural log of the mean number of citations for each year and t represents time passed between the year at issue and 1988, so that 1988 = 0, 1987 = 1, and so on. The coefficient for t, B(2), yields the rate at which the citation rate decays as one moves back in time from 1988 (Reference GujaratiGujarati 1995:169–70). Since B(2) = −0.147 in this analysis, we can conclude that a case decided in 1969 has only about 0.85 as much of a chance of being cited between 1989 and 1991 as one decided in 1970, which in turn has only 0.85 as much of a chance as one decided in 1971, and so on. (Another way to look at this is that a typical case decided in 1988 is about 20 times as likely to be cited as one decided in 1969.)

We applied this decay rate to discount the number of opinions written by each judge in each year. We then added up the discounted figures for each judge, to yield that judge's total opportunities. This sum was entered into the denominator, with the judge's total number of name citations going in the numerator. This yielded a citation rate for each judge.

To make this rate more interpretable, we multiplied it by the mean number of opportunities for all judges. The resulting score indicated the number of name citations each judge would be expected to receive, given the same number of opportunities as the average judge. On examining the distribution of scores, we found that it is dominated by low values. Nearly half of the scores are under 1.0. This distribution suggests that the difference between a score of zero and two is considerably greater than, say, the difference between 10 and 12. To reflect this fact, we transformed the scores by taking their natural logs (after adding 1 to each). This was the final step in our calculation of individual judges' scores. Conflict-level prestige scores were derived from these as described in the text.

Appendix 2: Descriptions and Descriptive Statistics for Variables in Model of Justices' Decisions

Dependent Variable

Vote: Did the justice vote in favor of the petitioner's position on the conflict-generating issue or the respondent's position? Pro-petitioner (1): 52.9%. Pro-respondent (0): 47.1%. N = 2988.

Independent Variables

Team Size Differential: Difference between number of circuits adopting petitioner's position and number rejecting it.

Dissent Differential: Difference between number of concurring or dissenting opinions rejecting petitioner's position and number supporting it.

Prestige Differential: Difference in prestige scores between highest-scoring judge authoring opinion supporting petitioner's position and highest-scoring judge authoring opinion rejecting it.

SG for Petitioner: Did the SG submit a brief supporting petitioner's position?

Yes (1): 37.0%. No (0): 63.0%. N = 338.

SG for Respondent: Did the SG submit a brief supporting respondent's position?

Yes (1): 32.8%. No (0): 67.2%. N = 338.

Ideology: Ideological attractiveness of petitioner's position (function of ideological direction of petitioner's claim and ideology of individual justice).

Descriptive Statistics

Footnotes

The authors are grateful to Harold Spaeth, Herbert Kritzer, and the anonymous reviewers for their careful and valuable feedback.

1 Aggregate analysis of the Court's decisions resolving circuit court conflicts between 1985 and 1995 does reveal significant differences between conflict and nonconflict cases. First, conflict cases differ from nonconflict cases in one important dimension: they tend to involve statutory, as opposed to constitutional, claims. As for voting behavior, Justices Scalia, O'Connor, Kennedy, Souter, and Rehnquist were significantly more likely to cast a liberal vote in conflict cases than in nonconflict cases. And while the differences are not statistically significant, Justices Ginsburg, Powell, and Thomas also cast a higher percentage of liberal votes in conflict cases.

2 In the vast majority of intercircuit conflicts, the Court is faced with a dichotomous choice between two alternative approaches adopted by the circuits involved. Thus, for example, in Qualitex Co. v. Jacobson Products Co., Inc., 514 U.S. 159 (1995), the Court resolved the issue of whether the use of a certain color alone could be protected under trademark laws. The circuits had split on the issue, with two (the Respondent's Team of the 9th and the 7th Circuits) establishing a prohibition against such protection, and two (the Respondent's Team of the 8th and the Federal Circuit) ruling that color could be protected under certain circumstances. The Court ruled for the Petitioner's Team.

3 To generate our list of conflict cases in which the Supreme Court noted the existence of a conflict, we used the following values in the Spaeth dataset: analu = 0; dec type = 1, 2, 5, 6, 7; and cert = 1, 2. While this method relied on the Supreme Court itself to identify the presence of conflicts and the circuits involved in them, we believe that this was the most reliable method available to construct our data, especially since venturing outside the opinions could require difficult judgment calls. Of course, we may have missed some conflicts in this way. But since we can think of no reason why the Court would systematically choose to recognize some conflicts but not others, we see no threat to the external validity of our results. And while we noticed instances in which the Court failed to list all of the circuits involved in a conflict, our examination of these cases did not reveal any bias in the cases listed by the Court (in favor of one team or another).

4 The following was typed into the search field (using Judge Friendly as an example): “(judge friendly or friendly, j. or friendly, c.j.) and not court (second circuit) and not judges (friendly).” In cases where one circuit judge shared a last name with another, this search was modified slightly, and we checked each case to determine which judge was cited.

5 The results for all variables were virtually identical, regardless of the specific measure of ideology we employed.

6 As expected, if the model is estimated using conventional standard errors, significance levels are much lower—specifically, every coefficient in the model has a p value smaller than 0.001. We also ran the analysis clustering on justice instead of case, recognizing that a justice's vote to support the petitioner or the respondent might not be independent across cases. With one exception, all the standard errors are much smaller when we cluster on justice than when we cluster on case. The exception is the standard error for ideology, which is still less than one-sixth the size of its coefficient. These results suggest that there is less independence across justices within cases than across cases within justices. Furthermore, analyses clustering on case provide the more conservative tests of our key hypotheses. For these reasons, we report those analyses in the text.

7 Our prestige variable used information about only the most prestigious judge from each team. As an alternative, we counted the number of judges on each team scoring at or above the 90th percentile on prestige and took the difference between the two teams. If this variable is substituted for our original prestige measure, all three jurisprudential variables perform as well as or better than in the original model, and there are no meaningful changes in the other variables.

8 As these predicted probabilities suggest, jurisprudential factors might have different effects for justices at different points on the ideological spectrum. To the extent that these differences result from justices being more or less sure of their positions before jurisprudential factors are considered, they are already captured in the logit model. But ideology could moderate jurisprudential effects in other ways too: for instance, justices who are more inclined to view cases ideologically may pay less attention to jurisprudential factors. To investigate this possibility, we constructed a new variable by taking the absolute value of each justice's individual ideology score (so that justices at the left and right extremes would score highest and moderates lowest). We then created three interaction terms by multiplying this variable by our jurisprudential variables and separately added each of the interaction terms (along with the new ideological strength variable) to the original equation. Intriguingly, although none of the coefficients are significant at traditional levels (for all three, p>0.06, one-tailed), all are in the predicted direction, suggesting that the relationship between strength of ideological leanings and susceptibility to jurisprudential effects is worthy of further study.

9 In addition to the model shown here, we also ran a model of outcomes at the Supreme Court. We have chosen not to present results from that analysis because we were unable to include an effective control for Court-level ideology. But it is worth noting that the results closely parallel those presented here and tell the same basic story of modest but consequential effects on outcomes.

References

References

Aldrich, John H., & Nelson, Forrest D. (1984) Linear Probability, Logit and Probit Models. Newbury Park: Sage Publications.CrossRefGoogle Scholar
Baum, Lawrence (1997) The Puzzle of Judicial Behavior. Ann Arbor: Univ. of Michigan Press.CrossRefGoogle Scholar
Bhattacharya, Mita, & Smyth, Russell (2001) “The Determinants of Judicial Prestige and Influence: Some Empirical Evidence from the High Court of Australia,” 30 J. of Legal Studies 223–51.CrossRefGoogle Scholar
Burton, Steven J. (1992) Judging in Good Faith. Cambridge: Cambridge Univ. Press.CrossRefGoogle Scholar
Caldeira, Gregory (1985) “The Transmission of Legal Precedent: A Study of State Supreme Courts,” 79 American Political Science Rev. 178–93.CrossRefGoogle Scholar
Cushman, Barry (1998) Rethinking the New Deal Court: The Structure of a Constitutional Revolution. New York: Oxford Univ. Press.Google Scholar
Edwards, Harry T. (1991) “The Judicial Function and the Elusive Goal of Principled Decisionmaking,” Wisconsin Law Rev. 837–65.Google Scholar
Epstein, Lee, & Kobylka, Joseph F. (1992) The Supreme Court and Legal Change: Abortion and the Death Penalty. Chapel Hill: Univ. of North Carolina Press.Google Scholar
Fish, Stanley (1989) Doing What Comes Naturally: Change, Rhetoric and the Practice of Theory in Literary and Legal Studies. Durham: Duke Univ. Press.Google Scholar
Gillman, Howard (1993) The Constitution Besieged: The Rise and Demise of Lochner Era Police Power Jurisprudence. Durham: Duke Univ. Press.Google Scholar
Gillman, Howard (2001) “What's Law Got to Do with It? Judicial Behavioralists Test the ‘Legal Model’ of Judicial Decision Making,” 26 Law and Social Inquiry 465504.CrossRefGoogle Scholar
Greenawalt, Kent (1992) Law and Objectivity. Oxford: Oxford Univ. Press.Google Scholar
Gujarati, Damodar N. (1995) Basic Econometrics, 3rd ed. New York: McGraw-Hill.Google Scholar
Handberg, Roger, & Hill, Harold F. Jr. (1980) “Court Curbing, Court Reversals, and Judicial Review,” 14 Law & Society Rev. 309–22.CrossRefGoogle Scholar
Hettinger, Virginia, et al. (2003) “Separate Opinion Writing on the United States Courts of Appeals,” 31 American Politics Rev. 215–50.Google Scholar
Howard, J. Woodford (1981) Courts of Appeals in the Federal Judicial System. Princeton: Princeton Univ. Press.CrossRefGoogle Scholar
Howard, Robert, & Segal, Jeffrey (2002) “An Original Look at Originalism,” 36 Law & Society Rev. 113–35.CrossRefGoogle Scholar
Klein, David E. (2002) Making Law in the United States Courts of Appeals. New York: Cambridge Univ. Press.CrossRefGoogle Scholar
Klein, David, & Morrisroe, Darby (1999) “The Prestige and Influence of Individual Judges on the U.S. Courts of Appeals,” 28 J. of Legal Studies 371–91.CrossRefGoogle Scholar
Landes, William M., & Posner, Richard A. (1976) “Legal Precedent: A Theoretical and Empirical Analysis,” 19 J. of Law and Economics 249307.CrossRefGoogle Scholar
Martin, Andrew D., & Quinn, Kevin M. (2002) “Dynamic Ideal Point Estimation via Markov Chain Monte Carlo for the U.S. Supreme Court, 1953-1999,” 10 Political Analysis 134–53.CrossRefGoogle Scholar
McGuire, Kevin T. (1995) “Repeat Players in the Supreme Court: The Role of Experienced Lawyers in Litigation Success,” 57 J. of Politics 187–96.CrossRefGoogle Scholar
McGuire, Kevin T. (1998) “Explaining Executive Success in the U.S. Supreme Court,” 51 Political Research Q. 505–26.Google Scholar
Newman, Jon O. (1984) “Between Legal Realism and Neutral Principles: The Legitimacy of Institutional Values,” 72 California Law Rev. 200–16.CrossRefGoogle Scholar
Perry, H. W. Jr. (1991) Deciding to Decide: Agenda Setting in the United States Supreme Court. Cambridge: Harvard Univ. Press.Google Scholar
Poole, Keith T., & McCarty, Nolan M. (1995) “Veto Power and Legislation: An Empirical Analysis of Executive and Legislative Bargaining from 1961–1986,” 11 J. of Law, Economics, and Organization 282312.Google Scholar
Posner, Richard A. (1995) What Do Judges Maximize? The Same Things Everyone Else Does in Overcoming Law. Cambridge: Harvard Univ Press.Google Scholar
Richards, Mark J., & Kritzer, Herbert M. (2002) “Jurisprudential Regimes in Supreme Court Decision Making,” 96 American Political Science Rev. 305–20.CrossRefGoogle Scholar
Sarat, Austin (1977) “Judging in Trial Courts: An Exploratory Study,” 39 J. of Politics 368–98.CrossRefGoogle Scholar
Segal, Jeffrey, & Spaeth, Harold (2002) The Supreme Court and the Attitudinal Model Revisited. Cambridge: Cambridge Univ. Press.CrossRefGoogle Scholar
Shapiro, Sidney A., & Levy, Richard E. (1995) “Judicial Incentives and Indeterminacy in Substantive Review of Administrative Decisions,” 44 Duke Law J. 1051–80.CrossRefGoogle Scholar
Sheehan, Reginald S., et al. (1992) “Ideology, Status, and the Differential Success of Direct Parties Before the Supreme Court,” 86 American Political Science Rev. 464–71.CrossRefGoogle Scholar
Smith, Rogers M. (1994) “Comments on The Supreme Court and the Attitudinal Model,” 4 Law and Courts 89.Google Scholar
Spaeth, Harold, & Segal, Jeffrey (1999) Majority Rule or Minority Will. Cambridge: Cambridge Univ. Press.CrossRefGoogle Scholar
Spaeth, Harold, & Teger, Stuart (1982) “Activism and Restraint: A Cloak for the Justices' Policy Preferences,” in Halpern, S. C. & Lamb, C. M., eds. Supreme Court Activism and Restraint. Lexington, MA: Lexington Press.Google Scholar
Tomz, Michael, et al. (1998) CLARIFY: Software for Interpreting and Presenting Statistical Results. Version 1.2. Harvard University, September 16. http://gking.harvard.edu/.Google Scholar
Wahlbeck, Paul (1997) “The Life of the Law: Politics and Legal Change,” 59 J. of Politics 778802.CrossRefGoogle Scholar

Cases Cited

Grayned v. Rockford, 408 U.S. 104 (1972).CrossRefGoogle Scholar
Qualitex Co. v. Jacobson Products Co., Inc., 514 U.S. 159 (1995).Google Scholar
Figure 0

Table 1. Logit of Individual Justices' Support for the Petitioner's Team on Jurisprudential and Attitudinal Variables

Figure 1

Table 2. Estimated Change in Probability That a Justice Will Support the Petitioner's Team, Given Specified Change in Variable