The So-Called Moderate Justices on the Rehnquist Court: The Role of Stare Decisis in Salient and Closely-Divided Cases

Problem statement: A major question that has puzzled political scient ists is what factors influence the decisions of US Supreme Court justice . Despite 20th Century statutory reforms that have led to a fundamental weakening of institutiona l cohesion on the Supreme Court, the norm of stare decisis continues to serve as a constraint to moder ate decision-makers under certain external conditions. Evaluate voting behavior on the Rehnqui st Court to discover which justices are indeed demonstrating moderate behavior. Approach: This research makes a unique contribution by expanding the US Supreme Court Justice-Centered Reh nquist Database (1986-2000) to include two new variables to measure the level of salience in e ach case. Therefore, allowing researchers to better access the impact of issue salience in closely divi ded precedent-setting cases. Both the New York Times and Congressional Quarterly indicators are us ed to gauge case salience. The analysis focuses on the existing academic and law review literature on the role of precedent and issue salience which may place constraints on the Court. The jurisprudential styles of Justices O’Connor, Kennedy, Souter, and White are analyzed to ascertain similar moderate be havior traits. Since the data is binary, the logist ic regression method is applied within the parameters of the moderate judicial model to illuminate the degree of moderate behavior. Results: The findings reveal that Justice Kennedy does not nea ly fit the moderate judicial model. Instead, O’Connor was the only justice that consistently demonstrated moderate voting behavior. Interestingly, only Justi ce White was more likely to maintain precedent in cases that were both salient and closely divided. Conclusion: This work helps close the glaring gap in the prevailing literature by developing a political model which predicts the conditions in which moderate justices are likely to uphold or not uphol d precedent. In addition, it provides a more accura te assessment of the current relevance of the norm of stare decisis to the Legal Model.


INTRODUCTION
Justices might be motivated by their own preferences over what the law should be, but they are constrained in efforts to establish their preferences by a norm favoring respect for stare decisis.
Lawrence Baum: The question of judicial voting alignments has been an interesting and important issue in judicial politics (Hensley et al., 1997). If a majority of justices form a highly cohesive voting bloc, then this majority coalition have the ability to bring about significant policy change on the United States Supreme Court. To analyze judicial alignments, scholars often categorize the justices on the Court into voting blocs on the basis of ideological behavior (Biskupic, 1992;Jost, 1995). Chief Justice William Rehnquist and Justices Antonin Scalia and Clarence Thomas often comprised the right of center bloc. The left of center bloc often includes Justices John Paul Stevens, Ruth Bader Ginsburg and Stephen Breyer. For one of the voting blocs to form a majority, it must obtain at least two votes from members of the moderate center of the Supreme Court. The moderate center often consisted of Justices Sandra Day O'Connor, Anthony M. Kennedy and David Souter (Greenhouse, 1992b). In closely divided cases, the votes cast by the moderate center justices determine the balance of power on the Court. For example, during the 1991 term, the Souter-O'Connor-Kennedy alignment voted together 71% of the time and when they did, they did not lose a case (Biskupic, 1992).
The votes of these three justices have arguably had a significant impact on the interpretation of constitutional law (Phelps and Gates, 1991). For instance, in Planned Parenthood of Southeastern Pennsylvania v. Casey (1992), moderates Kennedy and O'Connor joined with Justices Stevens, Souter and Blackmun to prevent the Roe v. Wade (1973) precedent from being overturned by the doctrinally conservative bloc of justices on the Rehnquist Court. In the joint opinion, O'Connor, Kennedy and Souter emphasized the significance of precedent and the need to protect the integrity of the Court. In addition, the opinion revealed a deep concern about the Court as an institution in American society and resistance to rapid change in the charged political circumstances surrounding abortion.
Moderate justices seek to retain institutional legitimacy because they are mindful of the prestige of the Court, adherence to the rule of law and the overall stability of the political system. Since the Court lacks the power to implement its decisions, it is inclined to be mindful of the public's perceptions (Biskupic and Witt, 1996). Precedent is an integral aspect of institutional legitimacy, which becomes particularly significant in salient cases where the prestige of the Court is placed in jeopardy. Some scholars maintain that Court prestige decreases when the Court overturns precedent because of the appearance of the triumph of policy preference over law (Miceli and Cosgel, 1994). Thus, moderate justices are less likely to overturn precedent in salient cases. In addition, they are most likely to adhere to precedent when their vote is pivotal to a minimum winning coalition. Even when controlling for judicial ideology and the ideological direction of the case, the moderate centrist justices will exemplify this peculiar type of judicial behavior.
In this study, I develop a scale of moderate judicial decision-making to develop a more precise definition of a moderate decision-making style. The scale measures moderate decisions through the application of three methods. First, law articles and legal newsletters are analyzed to ascertain whether moderates' lack a structured judicial ideology. Second, the ideological directions of the moderates' vote are examined to discover whether it differs markedly from that of other justices on the Court. Third, the precedential voting records for the entire bench are scrutinized to determine whether the moderates' behavior differs substantially from that of their associates on the Court.
The following three characteristics define moderate justices. They lack a structured judicial ideology, tendency to uphold precedent and provide the pivotal vote in determining the outcome in closely divided cases. Moderate justices adopt an issue-by-issue or case-by-case approach rather than one based on rigid ideological concerns. Moreover, they are more likely than others on the Court to join the majority in 5-4 or other minimum winning decisions. Since moderate justices' lack a firm ideological predisposition, they are more likely to be influenced by external pressures in cases that are salient (external pressures consist of public legitimacy concerns, Congressional statutory action and media attention). Therefore, they are more inclined than their court associates to uphold precedent when a case is salient than when it is not. Pivotal or swing justices on the Supreme Court tend to possess a significant impact on policy outputs, as the power to shape the law through written opinions very often falls on them (Schmidt and Yalof, 2004). Especially in closely divided cases, any opinion writer must be extra attentive to the views of their wavering colleagues (Epstein and Knight, 1998).
Stare decisis translates to "let the decision stand". It is the doctrine that principles of law established in earlier judicial decisions should be accepted as authoritative in similar subsequent cases. I argue that the norm of stare decisis serves as a greater constraint to moderate judicial decision-makers than their more ideological counterparts when cases are both salient and closely divided on the Court. In recent years, there have been highly active academic debates on the relevant influence of stare decisis on the votes of US Supreme Court justices (Brenner and Stier, 1996;Brisbin, 1996;Segal and Spaeth, 1996a;1996b;Songer and Lindquist, 1996;Spaeth and Segal, 1999;Segal, 1995). I hope to contribute to the academic debate by demonstrating that, despite Twentieth-century judicial reforms that have led to a fundamental weakening of institutional cohesion, the norm of stare decisis continues to influence the decision making of justices. My model will predict under certain conditions when moderate justices may respect precedent and when they may not. Since there is a lack of research in the area of moderate judicial behavior, my research helps fill the gap.
Theoretical underpinnings: Many theoretical models of judicial decision-making acknowledge that US Supreme Court justices are goal or policy oriented actors. The Court is subject to several external constraints. To create policy that the other branches will respect, the justices must consider the preferences and expected actions of other government actors. Since the Constitution permits the other elected branches of government to check the actions of the Supreme Court, external institutions serve as a constraint on judicial decision-making.
In many law articles and newsletters, the authors refer to Justices O'Connor, Kennedy and Souter as moderate compared to the ideologically defined justices on the Rehnquist Court (Cannon, 1997;Coyle, 1996;Coyle, 1999a;1999b;Filter, 1998;Greenhouse, 1992a;Kobylka, 1993;Savage, 1992;1996a;1996b;1996c;Smith, 1992;Taylor, 1989;Merrill, 1994). Taylor (1989) described O'Connor as a justice who "puts fairness above ideology and balance above clarity". In addition to O'Connor, Justice Souter has been described as a moderate justice who is not as ideologically driven as Justices Scalia or Thomas (Filter, 1998). Moreover, Justices O'Connor, Kennedy and Souter's jurisprudential style has been described as one that is "moderate in tone, respectful of precedent and leaning in favor of individual liberty" (Savage, 1996a). According to Smith (1992), Justice Kennedy moved toward a moderate position on the court to preserve precedents in salient cases concerning abortion and the Establishment Clause is to maintain the public's perception of the Court's legitimacy.
Although justices may have the ability to vote their personal policy preferences, there are forces that limit this discretion (Segal and Cover, 1989). According to Knight and Epstein (1996), institutional arrangements create roles that determine the appropriate behavior of anyone who might occupy that role. Therefore, institutions define what Talcott Parsons calls "legitimately expected behavior". In Dickerson v. United States (2000), Chief Justice Rehnquist, an avid critic of Miranda v. Arizona (1966), said the following in the majority opinion: "Whether or not we would agree with Miranda's reasoning and its resulting rule, were we addressing the issue in the first instance, the principle of stare decisis weighs heavily against overruling it now".

Strategic model:
The decision making of moderate justices on the court may be best conceptualized by the theoretical approach of the Strategic Model. Epstein and Knight (1998) view justices as strategic decisionmakers that acknowledge that their ability to obtain goals depends on the institutional context in which they act. Justices modify their positions by considering a "normative constraint" in order to render a decision as close as possible to their desired outcome. A norm supporting a respect for precedent can serve as such a constraint. If the Court establishes rules that the people will neither respect nor obey, the efficacy of the Court is undermined. Perhaps, a constraint is apparent to the Court during a crisis of whether to overturn a precedent or not (Howard and Segal, 2001). Moderate justices are confronted with societal pressure or a "crisis" when a case exemplifies issue salience in the media. This external influence in tandem with the pressure of a closely divided case will tend to convince a moderate justice to join a minimum winning coalition of justices to preserve court precedent.
In Planned Parenthood of Southeastern Pennsylvania v. Casey (1992), the justices painstakingly pointed out that they believe that the norm of stare decisis influences relations between them and society and not only the legal community. The overuse of the power to overturn precedent could possibly undermine the Court's authority and legitimacy and therefore erode the impact of its opinions. The Court may also feel constrained to follow precedent so that its decisions are respected by future Courts. According to O'Brien (1996), denied the power of the sword or the purse, the Court must cultivate its institutional prestige. The Court's prestige depends on preserving the public's view that justices derive their decisions on interpretations of the law, rather than on their personal policy preferences. Therefore, the power of the Court "ultimately rests with other political institutions and public opinion" (O'Brien, 1996;Richards and Kritzer, 2002).
According to the strategic model, justices are policy-seekers who use precedent and other legal rules in a strategic way to persuade others to believe in the significance of the tenets of the Legal Model. There are four main parts to the strategic model. First, justices are considered to be primarily followers of legal policy, not unconstrained actors who make decisions based solely on their own ideological attitudes. Second, justices are strategic actors who realize that their ability to reach their goals depends on the knowledge of the preferences of other justices on the Court. Third, the model focuses on the choices the justices expect others to make. Fourth, the institutional contexts in which the justices act are significant to the model. Epstein and Knight (1998) believe that the model stipulates that strategic decision-making is about interdependent choice; an individual's action is a function of her expectations about the actions of others. The Court does not make policy in isolation from the other main actors in government; the justices must moderate their decisions by what they "can do" (Eskridge, 1991). Justices need to consider not only the preferences of their colleagues but also the preferences of other political actors, including Congress, the President and even the public. Constitutional checks and balances compel justices to consider the preferences of other actors. When justices proceed to quickly or too far in their interpretation of the Constitution, as they did in Brown v. Board of Education (1954), the public's acceptance of the Court's legitimacy is placed in jeopardy. As James Gibson sufficiently states, "Judges' decisions are a function of what they prefer to do, tempered by what they think they ought to do, but constrained by what they perceive is feasible to do".
For example, in the Planned Parenthood v. Casey (1992) decision the so-called moderates' O'Connor, Kennedy and Souter considered the reliance people placed in the rule of law in Roe v. Wade and whether overturning the rule of law in Roe would create special hardships. The joint opinion revealed that the moderate justices feared that the Court's legitimacy would be placed in jeopardy by overturning Roe (Kahn, 1999): "To examine under the fire in the absence of the most compelling reason to reexamine a watershed decision would subvert the Court's legitimacy beyond any serious question". According to Kahn (1999), the Casey decision illustrated that, because of the Court's institutional standing in the political system and the justices' conceptions of their personal obligations, a commitment to precedent takes priority over a more instrumental approach to the law.
Role of case salience: Scholars have given considerable attention to the issue of measuring the importance of Supreme Court cases, but no clear consensus has emerged. Thomas Hensley and Jarrod Tudor classified case salience as either of major or minor significance. They utilized the following three sources which identify the Court's most important cases each term: (1) the New York Times, (2) United States Supreme Court Reports: Lawyer's Edition and (3) United States Law Week: Supreme Court Section. To be classified as a major case, all three sources must identify a case as important. The requirement that a case make all three lists provides substantial confidence that the case is of major importance. Epstein and Segal (2000) have also analyzed the variable of issue or case salience as an important influence on the Court. They were able to demonstrate by comparative analysis that the New York Times (NY Times) measure was the best contemporary indicator of issue salience. On the basis of its national orientation, the NY Times was not as susceptible to a region bias as other metropolitan newspapers. Furthermore, the NY Times indicator gauges salience at the time the justices were deciding the case rather than years after the case was resolved. In contrast, Congressional Quarterly (CQ) focuses more on whether a case has "withstood the test of time and less toward whether it was salient at the time the Court was deciding it". Jeffrey Segal found that precedential voting exists in cases of the lowest salience: (Ordinary cases compared to landmark cases) and among ordinary cases, (statutory cases over constitutional cases) and (modern economic cases over modern civil liberties cases). Building on Segal's work, I will apply two independent measures to assess issue salience in all cases presented before the Burger and Rehnquist Courts. The NY Times and CQ indicators have been well tested for reliability in previous research (Epstein and Segal, 2000). I hypothesize that the moderate justices will, as a result of external influences, vote differently than their more ideological associates on the Court.

MATERIALS AND METHODS
Model building and research methodology: Spaeth and Segal (1999) have launched a prolific empirical assault on the relevance of stare decisis to decisionmaking on the nation's highest court. Epstein and Knight (1998) have responded to this assault on the Legal Model's most widely respected rule by analytically defending precedent as a normative constraint on justices' voting their personal preferences. I add to their defense of stare decisis with the following model: MJ In other words, moderate justices' decision to uphold precedent is a function of case salience, minimum winning coalitions, the ideological direction of the case and individual ideology or lack thereof. As previously discussed, the moderate judicial scale will now be examined. While the law literature for the most part demonstrates that the moderate justices' decisionmaking styles are quite unique compared to the styles of the other justices on the Court, the overall direction of the justices' vote during the Rehnquist Court does not indicate a moderate voting effect. According to the data in Table 1, Justices O'Connor and Kennedy are twice as likely to vote in a conservative direction as in a liberal one. In addition, Justice White was 29% more likely to vote in a conservative than liberal direction on the Court.
White's conservative voting record stands in stark contrast to Justice Souter, whose voting record was much more ideologically evenly split than that of his contemporaries. Souter was only slightly more likely to vote conservative than liberal (8.6%). O'Connor and Kennedy's move to the conservative right is very apparent in Table 1.
Therefore, the data from Table 1 indicate that among the so-called moderate justices only Justices Souter's descriptive voting record exemplifies a degree of moderate behavior.
In Table 2, each justice's vote to uphold precedent is illustrated for comparison. In contrast to the Original US Supreme Court Database, the Rehnquist justicecentered database documents when justices individually deviate from the behavior of the majority or plurality. As a result, the Rehnquist Court Database will provide a more complete account of the precedential behavior of the individual justices on the Court.  In Table 2, Justice O'Connor ranked second but closely behind Justices Rehnquist and Stevens with the greatest percentage of votes for upholding precedent. Kennedy and White were not too far behind, ranking fourth and sixth, respectively. According to the descriptive analysis in Table 2, the so-called moderates are no more likely to uphold precedent then their more ideologically set colleagues on the Court.
To assess contemporary issue salience in US Supreme Court cases, I applied a combination of the New York Times and CQ's Major Cases as indicators. Epstein and Segal (2000) relied on the Index to the New York Times and LEXIS to create the NYT measure. A salient case (1) led to a story on the front page of the Times on the day after the Court handed it down, (2) was the lead or "headline" case in the story and (3) was orally argued and decided with an opinion (p.73). According to Cook (1993), a noted researcher on the subject, a minimum of two authorities must be utilized in order for a case to be considered salient "since to accept a single authority would introduce idiosyncratic standards". Moreover, Cook (1993) found that the list compiled by Congressional Quarterly was a concise but a reliable authority for research on contemporary decisions. Although Spaeth and Segal (1999) has utilized the Lawyers Edition to identify significant nonconstitutional cases, Cook (1993) claims that the Lawyers Edition is a questionable source due to the lack of identifiable scholars who take responsibility for the cases selected.
In contrast to Epstein and Segal (2000), I classified cases as "salient, if they appeared in the New York Times and are listed in Congressional Quarterly's list of major cases. The CQ Press each term selects the major cases for the Supreme Court's term. The selection is based on such factors as the rulings' practical impact; their significance as legal precedent; the degree of division on the Court and the level of attention among interest groups, experts and news media. Judicial ideology is analyzed by the following Supreme Court Database variables: Direction of decision, direction of the individual justices' votes. The direction of decision variable specifies for each issue in each case whether there was a liberal or conservative outcome. The individual justices' votes variable determines whether the individual justice voted in a liberal or conservative direction. The minimum winning coalition variable indicates whether a vote in a case was decided by a margin of one vote. It includes those cases decided 5-4 and 4-3, or by a 5-3 or 4-2 vote that reverses the decision of the lower court. I created two new variables by interacting case salience and minimum winning coalition. The first variable pertains only to minimum winning coalitions formed during a conservative Court decision. The second variable occurs when a minimum winning coalition coalesces during a liberal Court decision.
In this study, I focused only on the part of the alternation of precedent variable that indicates whether there was no formal alteration of precedent in the case outcome. My model is primarily concerned with the conditions in which moderate justices uphold precedent. On the basis of the aforementioned theoretical discussion, I propose the following research questions: • Does the institutional norm of stare decisis constrain the decision-making of the • Moderates more than other justices on the Court in salient rather than non-salient cases? • Does the norm of stare decisis constrain the decision-making of the moderates more than other justices on the Court in minimum winning coalitions?
The answer to these two questions will not only lead to a more accurate assessment of the norm's relevance to the Legal Model but also contribute to the diverse academic debate on the subject. Furthermore, the research questions focuses the study away from consideration solely of vote counts in closely divided decision to a more substantive analysis of the external factors that contribute to the relevance of precedent to the so-called moderates on the Rehnquist Court. Mishler and Sheehan (1996) found that the impact of public opinion is greatest among the moderate justices who are likely to hold swing positions on the Court. The so-called moderate justice is more likely to change their opinions in response to public opinion. According to their data results, O'Connor and Kennedy's increasing liberalism during the late 1980s and early 1990s was more consistently responsive to fluctuations in the public mood than was the stance of either liberal or conservative justices (Mishler and Sheehan, 1996). This finding led me to propose the following hypotheses: Hypothesis 1: The norm of stare decisis is more likely to act as a constraint to the moderate justices' decisionmaking in salient rather than non-salient cases.

Hypothesis 2:
The institutional norm of stare decisis acts as a greater constraint to moderate justices than to other justices on the Rehnquist Courts.
Justices O'Connor, Kennedy and White were more inclined to join the majority in 5-4 decisions than were the other justices on the Court. In regards to voting in closely divided 5-4 decisions, Justice Kennedy joined the majority more frequently than any other justice on the Rehnquist Court (data are derived from the Harvard Law Review statistical tables on 5-4 decisions from November 1982). Kennedy voted with the majority seventy-eight percent between the 1987 and 2000 terms. Justice White ranked second with seventy percent between 1981 until the 1992 term. Justice O'Connor ranked third with sixty-seven percent between the times of her appointment in 1981 until the 2000 term.
Since O'Connor left the bench in 2005, Kennedy has written more majority opinions in landmark cases than any other justice on the Court (Justice Kennedy wrote 27% of the majority opinions in CQ's list of major cases from 2005-2006 until 2007-2008 terms. Each term Congressional Quarterly selects the major cases for the Supreme Court's term. The selection is based on such factors as the rulings' practical impact; their significance as legal precedent; the degree of division on the Court and the level of interest among interest groups, experts and news media). According to Colucci (2009) over the previous twenty years Justice Anthony Kennedy has voted with the majority more than any of his colleagues. Kennedy's vote was the determining factor in several politically salient cases such as Ricci v. DeStafano, (2009); Bush v. Gore (2000); Lawrence v. Texas (2003) and Hamdan v. Rumsfeld (2006). During the 2007 Court term, Justice Kennedy's cast the decisive vote in eight out of 12 (67%) cases that were 5-4 decisions. In 2006, that happened 19 out of 24 times (79%). Overall, the most substantive difference from the 2006 Court term is that Justice Kennedy agreed most frequently with some of the liberal members of the Court and less frequently with the conservatives. During the 2008-2009 Court Term, Kennedy joined the majority in slightly over 2/3rds of the criminal and related 5-4 cases. This particular judicial behavior is a hallmark characteristic of moderate decision-making.
In a "crisis" or external pressure on the Court, Justice Kennedy tends to uphold precedent in order to protect the prestige of the Court. Anthony Kennedy, more than any other justice, changed his decisions and contradicted his previously stated positions to preserve precedent in cases concerning abortion and the establishment clause. In Planned Parenthood of Southeastern Pennsylvania vs. Casey (1992), Kennedy fearing the loss of the court's legitimacy joined a moderate bloc of justices to uphold the central tenets of Roe v. Wade (1973) precedent from reversal in a closely divided decision during a presidential election year. Therefore, on whether to uphold precedent or not in a case the dual effect of issue salience and minimum winning coalitions are key factors weighing on the decision-making of the so-called moderate justices' on the Court.
The aforementioned case examples and statistics led me to form the following hypotheses: Hypothesis 3: The moderate justices are more likely to follow precedent when a case is closely divided in a minimum winning coalition than when it is not.
Hypothesis 4: When a case is minimum winning, moderate justices are likely to respect precedent by voting in either a conservative or liberal Court outcome.
Hypothesis 5: Moderate justices tend to join with the majority to uphold precedent in cases that are both salient and closely divided in a minimum winning coalition than when it is not.

RESULTS
The Rehnquist Database contains only a portion of the cases that Justices White and Powell deliberated in during their tenures on the Court. I primarily used the Rehnquist Court Judicial Database to analyze the socalled moderate justices on the Court. This database differs from the Original US Supreme Court Database because its unit of analysis is justice-centered rather than case-centered. The justice-centered data maintain the advantage of citing when justices individually deviate from the behavior of the majority or plurality. The data covers the Court terms from1986-2000.
The data in Table 3 offered only mixed support to the first hypothesis. Justices O'Connor, Kennedy and Powell tend to uphold precedent more in non-salient than salient cases. In contrast, Justices Souter and White uphold precedent slightly more in salient than non-salient cases. In order to analysis the data in greater detail, I conducted a logistic regression of each moderate justice on the Rehnquist Court. Since the variables are binary, logistic regression was the most appropriate method of inferential statistics. My model included the following independent variables: Direction of individual justice's vote (Vt_Dir), direction of decision (Dir_DUM), Minimum Winning Coalition (MWC) and Case Salience (CS). The dependent variable is whether the justice voted to uphold precedent or not.      According to the data in Table 4a, Justices O'Connor, Powell and White are more likely to adhere to precedent when a case is salient than when it is not. Among the three aforementioned justices, Powell's case salience is at the greatest level of significance. Powell and White are the only two justices that tend to uphold precedent when cases are decided by one vote in a minimum winning coalition. In addition, Powell and White are more likely to adhere to precedent and join a minimum winning coalition when the ideological direction of the decision is liberal than when it is conservative. Alternatively, Justices' Kennedy, O'Connor and Souter are more likely to uphold precedent when the case has a liberal outcome but not a minimum winning coalition. Therefore, all of the socalled moderates tend to respect precedent when the ideological direction of the Court decision is liberal rather than when it is conservative. Despite their conservative voting credentials, Kennedy and O'Connor demonstrated moderate behavior by upholding precedent in liberal outcomes of cases. However, contrary to the moderate model of decisionmaking, both justices did not join the majority or plurality in closely divided cases. Only Powell and White joined the majority or plurality in minimum winning decisions. The data in Table 4a demonstrates that O'Connor, Powell and White are more likely to adhere to precedent when cases are considered salient than when they are not lends credence to my hypothesis.
In Table 4b, Rehnquist was the only justice more likely to uphold precedent when a case is salient than when it is not. However, this finding was not statistically significant. Interestingly, not one conservative justice adhered to precedent when the case was a minimum winning coalition. This result lends some support to the third hypothesis, which claims that the moderates are more likely than other justices to uphold the norm of stare decisis when a case is minimum winning than when it is not.
In Table 4c, Justices Blackmun, Marshall, Brennan and Stevens are more likely to uphold precedent in salient than in non-salient cases. Their voting behavior supports the null hypothesis that there is no difference between the moderates and other justices on the Court. Although several justices in the liberal bloc uphold precedent in salient cases, they do not join minimum winning coalitions to preserve it. Justice Breyer was the only exception to this in the analysis of the data. While Breyer had a tendency to uphold precedent in minimum winning coalitions, he did so only in non-salient cases. The voting behavior of the Left Bloc differs from the so-called moderates' in the aspect that both Powell and White adhered to precedent when cases were both salient and minimum winning coalitions.
The chi-square statistic for all of the justices is very high. Since the chi-square statistic is large and the observed significance level is small, I can reject the null hypothesis that there is no relationship between the dependent and independent variables. Additionally, the likelihood ratio statistic shows that the model is significantly different from the null hypothesis.
In regards to the third hypothesis, I proposed that the moderate justices are more constrained than other justices on the Court by precedent when a case is closely divided, indicating that there may be a threat to the public legitimacy of the Court by overturning precedent. In addition, I also proposed in this study that the political pressure generated by the media in salient cases influences the decision-making of moderate justices more in closely divided cases. Therefore, my model includes a new variable based on the interaction of the direction of the Court decision and minimum winning coalition variables. The results of the data analysis are reported in Table 5a-5c. According to Table 5a, Justices O'Connor, Powell and White are more likely to uphold precedent in salient rather than non-salient cases. In addition, O'Connor tends to respect precedent more in salient cases that are decided in a conservative ideological direction and minimum winning than in a liberal outcome. In contrast, Justices Powell and White are more likely to adhere to precedent when the case is not only salient but also decided in either a liberal or conservative Court direction. Justices Kennedy and Souter tend to uphold precedent more in non-salient than salient cases. Despite Kennedy's solid conservative voting record, he is more likely to respect precedent when the direction of the Court decision is liberal than conservative. Kennedy's judicial behavior is quite different from that of his moderate colleagues who are likely to join a minimum winning coalition in conservative decisions. Justices' White and Powell are the only two so-called moderates to meet the parameters of the model. Souter was the only justice on the Court that is not likely to join a closely divided case in either a conservative or liberal outcome.
In Table 5b, as expected Justices Scalia and Thomas were more likely to join a minimum winning coalition when the outcome of the case was conservative than liberal. Overall, the results are mixed at best. Thus far, Rehnquist was the only ideologue that upheld precedent in salient cases that are closely divided, no matter the ideological direction of the Court decision. Former Chief Justice Rehnquist's judicial behavior is very similar to that of so-called moderates White and Powell.
According to Table 5c, Justices Brennan, Marshall, Blackmun and Stevens are more likely to uphold precedent in salient than non-salient cases. Despite their strong liberal voting records, Brennan and Marshall are more likely to adhere to precedent in cases that are decided in a conservative manner and that are closely divided in a minimum winning coalition. In contrast, Blackmun, Stevens and Justice Ginsburg do not tend to join the minimum winning coalition that decide cases in either a conservative or liberal direction. Only Justice Breyer demonstrated moderate behavior when joining the majority in closely divided cases no matter the ideological direction of the outcome. The data analysis provides a degree of support to the fourth hypothesis that contends that the moderate justices are more likely than their more ideological counterparts to join minimum winning coalitions, no matter the ideological outcome of the case. Both White and Powell also exemplified this type of judicial behavior on the Rehnquist Court. Surprisingly among the two ideological blocs of justices, only Rehnquist and Breyer exemplified characteristics of moderate judicial behavior.

DISCUSSION
The data results in this study do not support the moderate decision-making model proposed in the first hypothesis. The data presented in the third table are based on cross-tabulations from both databases between precedent and case salience. The results demonstrated that Justices O'Connor, Kennedy and Powell were more likely to respect precedent in non-salient than salient cases. Moreover, the logistic regressions in Table 4 and  Table 5, support the null hypothesis by illustrating that case salience had no substantial effect on any of the justices' behavior. Thus far, the results have run counter to the predictions in both the first and second hypotheses and confirm Spaeth and Segal (1999) finding that justices are more likely to defer to precedent in cases that are less important.
The findings in Table 4a provide evidence for the first hypothesis. Except for Kennedy and Souter, the other so-called moderate justices O'Connor, White and Powell were more inclined to adhere to precedent in salient rather than non-salient cases. Among these three moderates, the case salience variable was statistically significant for only Justice Powell. According to the data on the Right Bloc in Table 4b, only Justice Rehnquist upholds precedent more in salient that nonsalient cases. However, the case salience variable was not significant for Rehnquist. In contrast to the conservatives, the Left Bloc was much more likely to respect precedent in salient rather than non-salient cases. The variable for case salience was significant for only Justices Brennan and Marshall. However, the data reported in Tables 4b and 4c does not provide support for the validity of the second hypothesis. Justices Rehnquist, Brennan, Marshall, Blackmun and Stevens are as likely to uphold precedent in salient cases as their so-called moderate colleagues. In regards to the minimum winning coalition variable, the data listed in Table 4a through 4c demonstrate that Powell and White were more likely than the other justices on the Court to cast their vote in a minimum winning coalition. Within the Left Bloc, Justice Breyer was the sole justice that supports precedent no matter the ideological direction of the Court's decision. In analyzing Table 5a, 5b, and 5c there tends to be support for the third hypothesis. While two justices (White and Powell) on the moderate-centrist bloc were more likely in closely divided cases to respect precedent by voting in either a conservative or liberal court decision, only one justice (Breyer) from both the Right and Left Blocs exemplified similar behavior. This finding supports the fourth hypothesis contention that when a case is closely divided in a minimum winning coalition that moderate justices are more likely to uphold precedent by voting in either a conservative or liberal court outcome than other types of justices on the Court.
In regards to the fifth hypothesis tested in Table 4a, only Justices White and Powell tended to join with the majority to uphold precedent in cases that were both salient and closely divided in a minimum winning coalition than when not. These two aforementioned justices were the only of the so-called moderates to meet the conditions of the hypothesis.
The overall effect of precedent in this model is not very compelling. However, students and scholars of judicial politics will as a result of this work gain a better understanding of moderate judicial behavior in both salient and minimum winning cases. Since there was a lack of research in this area, my findings will help fill the gap for future scholars. Finally, it addresses a research question that has been largely neglected by the existing literature.
Codebook: Rehnquist court database: The dependent variable is the individual justice's vote to uphold precedent. The variable is coded one when the justice upholds precedent and zero otherwise. The independent variables are coded in the following way: Case Salience (CS) was coded "0" if it was not salient and "1" otherwise. Minimum Winning Coalition (MWC) has the value of one when the number of justices in the majority voting coalition of the precedent exceeded those in the minority by only one and zero otherwise. The Direction of Individual Justice's Vote (VT_DIR) was coded "0" for conservative and "1" for liberal. The Direction of Supreme Court Decision (DIR_DUM) was coded exactly the same as VT_DIR. A conservative Supreme Court decision in a Minimum Winning Coalition (CMWC) was labeled "1" when the decision is in a conservative ideological direction and zero otherwise. A liberal Supreme Court Decision in an MWC (LMWC) was coded one when the Court's decision is in a liberal rather than a conservative direction and zero otherwise.