INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

A Bell & Howell Information Company 300North Zeeb Road. Ann Arbor.Ml 48106-1346 USA 313/761-4700 800/521-0600 THE JUDICIARY AND THE DUAL EXECUTIVE IN THE AMERICAN STATES: A COMPARATIVE ANALYSIS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

By

James Brent, B.A., M.A.

*****

The Ohio State University 1995

Dissertation Committee: Approved by L . Baum

G. Caldeira Adviser E. Slotnick Department of Political Science UMI Number: 9525998

UMI Microform 9525998 Copyright 1995, by UMI Company. All rights reserved.

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103 VITA

May 21, 1966 Born - Lynwood, California 1988 B.A.,B.F.A., Southern Methodist University, Dallas, Texas

1992 ...... M.A., Ohio State University, Columbus, Ohio

FIELDS OF STUDY

Major Field: Political Science Studies in: Judicial Politics (Laurence Baum) Legislative Politics (Samuel Patterson) Presidential Politics (John Kessel) State and Local Politics (Laurence Baum) Comparative Politics (Goldie Shabad)

ii TABLE OF CONTENTS

VITA ...... ii LIST OF TABLES ...... iv CHAPTER PAGE I Judicial Relations with the Executive Branch ...... 1 II The State Attorney General as Independent Political Actor ...... 34 III The Research Design and S e t t i n g ...... 66 IV Attorney General Agendas and Success Before Courts Collectively ...... 105 V Contours of Support Among Individual J u s t i c e s ...... 137

VI The Governor and Attorney General as Litigants...... 174 VII Conclusion ...... 202 APPENDICES

A. Letter and Questionnaire Sent to State AG's Offices ...... 214 B. Coding Rules ...... 220 BIBLIOGRAPHY ...... 226

iii LIST OF TABLES

TABLE PAGE 1 Dissent Rates on State Supreme Courts .... 67 2 Selection Systems and AG Power in 12 States . .71 3 Number of Amicus Briefs Filed by State AGs . .73 4 Selected Features of State Supreme Courts . . . 97 5 Political Control of Three State Institutions . 98

6 Statistics Describing AG Workloads ...... 99 7 Percentage of Cases in Which AG Took A Conservative Position ...... 108 8 Supreme Court Support for the Attorney G e n e r a l ...... 115 9 AG Win Rates, Criminal vs. Non-Criminal . . . 116 10 AG Advantage Over Other Litigants...... 119 11 AG Win Rates, Petitioner vs. Respondent . . . 123 12 Parameter Estimates for Model of Individual Support for A G ...... 125 13 Parameter Estimates for Model of Indivdual Support for AG (Including California Dummy) 131 14 Parameter Estimates for Model of Individual Support for AG (Including GOVADVAN variable)...... 132 15 Dissent Rate on State Courts of Last Resort . 139 16 Support for the AG, Democrats vs. Republicans...... 143 17 Support for the AG, Democrats vs. Republicans (Criminal and Non-Criminal Cases) ...... 146

iv 18 Support for the AG, Members of AG's (Governor's) Party vs. other justices ...... 149 19 Support for the AG, Appointed vs. Elected J u s t i c e s ...... 152 20 Support for the AG, Appointed Justices minus Other Justices ...... 155 21 Parameter Estimates for Individual Model of Overall AG Success ...... 158 22 Parameter Estimates for Individual Model of Overall AG Success (Includes SAMEGOV v a r i a b l e ...... 162 23 Comparison of Parameter Estimates of Two M o d e l s ...... 163 24 Parameter Estimates for Model of AG Success (Criminal and Non-criminal Cases) ...... 165

25 Correlations Between Ideological Variables . 167

26 Parameter Estimates for Models of AG Success, Controlling on Ideology ...... 168 27 Probit Estimation of Model Explaining AG Success...... 189 28 Probit Estimation of Model Explaining Gubernatorial Success ...... 197

v CHAPTER I JUDICIAL RELATIONS WITH THE EXECUTIVE BRANCH

In a system of separation of powers such as that which exists in the United States, power is distributed in order to ensure that no one governmental institution is able to

dominate the others.1 While this arrangement offers the substantial benefit of helping to prevent tyranny, it can also lead to negative consequences for effective government: stagnation, inefficiency, and gridlock. Although these terms are usually mentioned in reference to executive - legislative relations, the judicial branch of government also plays an

important role in the implementation of public policy. The

U.S. Supreme Court may thwart the will of the executive or the legislature through its use of constitutional and statutory interpretation. A great normative debate continues to rage about the proper role of the judiciary in a democratic system of government.2 But the fact remains that courts have the power

to void legislative and executive actions if they chose to wield it (see Leshy 1988). Naturally, courts do not always — nor perhaps even customarily3 — exercise this power, but the fact that they possess it means that other governmental actors

1 must take the judiciary into account. Various means must be employed to help ensure that the courts will respond positively rather than negatively. These methods may be employed before the adjudicative process — i.e. drafting of legislation or administrative rules in such a form as to pass constitutional muster, selecting judges who have the "correct" ideological or jurisprudential philosophies, etc. They may also be employed during the adjudicative process itself — i.e., through the careful screening of cases and the presentation of compelling legal arguments. They may also be employed after adjudication is complete4 — i.e. through attempts to remove judges, to restrict the jurisdiction of courts, or to redraft legislation or propose constitutional amendments to reverse "incorrect" judicial decisions. Each of these forms of executive - judicial interaction would be suitable for empirical analysis. However, as all research must, this dissertation focuses more narrowly upon

one particular form of interaction between the two branches — the success of the executive branch and two of its most significant agents, the governor and state attorney general, in actual cases decided on the merits by state courts of last resort. The decision to focus on litigation was dictated partially by practical concerns: cases decided on the merits

are discrete events which are reliably recorded, easily ob­ tainable, and fairly easy to quantify. But apart from this, collectively these cases are intrinsically important. A few examples, taken from the more than 8,000 cases coded for this dissertation, may serve to demonstrate the stakes involved when the executive branch appears before a state's highest tribunal. Fiscal Consequences — Appendix A reveals that among the most common types of cases involving the executive branch

resolved by SSCs on the merits are cases involving tax, zoning, and workers compensation regulations and statutes. Eminent domain cases are also quite prevalent. While many of these cases involve private individuals and small businesses, often they involve major corporations and millions of dollars.

Collectively, billions of dollars in governmental revenue are at stake.

Other, less routine decisions can also have substantial fiscal consequences. School financing is a perennial source of contention.6 SSCs are regularly called upon to decide questions involving regulation of business (including environmental policy), labor relations, unemployment compensation, welfare, state pensions, and a host of other

issues and entitlement programs which can affect a state's coffers to a substantial degree.

Social and Moral Consequences — Many questions of public policy have a moral component at their heart. Hence, state supreme courts are required from time to time to get involved

in volatile issues which may produce a substantial social change or have significant moral or religious implications. 4 And the executive branch is often called upon to defend governmental policy in court. While these types of cases do

not make up as large a proportion of the typical SSC docket as they do the docket of the U.S. Supreme Court, they nonetheless occur, and their incidence is probably increasing (i.e. Galie 1982) . The treatment of criminals, including the decision of whether to impose the death penalty, is a highly emotionally-

charged issue which consumes a relatively large proportion of the docket of every state court of last resort. Examples of other potentially contentious thickets in which SSCs included in this study have tread include freedom of religion - church/state issues,5 freedom of information6, nude dancing - freedom of expression,7 AIDS policy and gay rights,8 civil rights generally,9 surrogate motherhood,10 and the right to die11, among others. Political Implications — At times the executive branch will appear in court, whether voluntarily or not, because its very ability to wield power is challenged. These types of cases are by far the rarest of the three types discussed, yet they may be among the cases perceived as most important by the executive branch. Whereas in the previous types of cases the executive may be required to go to bat for a policy with which it disagrees, these cases may involve separation of powers issues or other issues which go to the heart of the executive's ability to govern as it sees fit. A few examples may serve to illustrate the point. In Amoitheater Unified School District #10 v. Harte12. the Arizona attorney general's office sought an expansive interpretation of the office's power to conduct civil rights investigations, and obtained it. However, some of the more interesting cases occur when there is infighting within the executive branch itself. In State ex rel Link v. Olson13, the governor of North Dakota sued the state attorney general and secretary of state over the constitutionality of a partial gubernatorial veto he had wielded, and the constitutionality of a legislative assignment of duties to the lieutenant governor (he lost). A similar conflict occurred between the California governor and attorney general in State ex rel Deukmeiian v. Brown.14 In In re Governorship15. California Governor sued his lieutenant governor Mike Curb in order to rescind a judicial appointment made by Curb in Brown's absence from the state (he won) . In each of these cases, the particular direction of the policy pursued by the executive actors in question was less important than their very ability to act. For these reasons, investigating the correlates of executive success in state courts is a matter of practical significance. As will be discussed below and the following chapter, it is a matter of theoretical significance as well. There has been a rather large literature devoted to documenting the success of the federal executive in court. However, very little of this work has attempted to explain executive success at the federal level. This dissertation will attempt to shed some light on this question by conducting a multi-state, multi-year comparative analysis. A fundamental assumption of this dissertation is the characterization of judges as political actors, and "that actors in political situations are goal oriented; that is, they have certain goals they wish to accomplish in these situations" (Rohde and Spaeth 1976, 70). Similarly, Leonard has written, "Modern jurisprudence contends that American courts are influenced by factors other than legal doctrine and analysis...Courts are generally considered policy-making bodies whose decisions are affected by factors that are exter­ nal to litigation, including the liberal or conservative inclinations of the individuals who sit on the bench" (1989, 935) . However, as Rohde and Spaeth point out, judges act within a framework of rules and norms which can constrain or influence their behavior. Hence, if we are to gain a better understanding of what influences executive success in court, both individual and structural factors must be taken into account. This dissertation proposes to do so. A brief portrait of judges as political actors motivated by legal, ideological, and political considerations will be painted, drawing upon the relevant literature. Then, in the remainder of this chapter and the next, some general comments about the relationship between courts and executives will be made and appropriate hypotheses developed. These hypotheses will then be tested, utilizing a twelve-state comparative analysis. JUDGES AS POLITICAL ACTORS

There is a normative ideal of judges as neutral scholars of the law. This approach, variously referred to as positivist, analytic, or mechanical jurisprudence, views the law as something to be discovered through logical reasoning and scholarship (i.e. Corwin 1948). Described in its most basic form, a decision is reached as follows: "1) judges

observe a similarity between cases; 2) they announce the rule of law inherent in the first case; and, 3) they make that rule applicable to the second case" (Levi 1949). It is clear that fidelity to precedent and statute are well-established and widely accepted components of the judicial role (Flango et al 1975; Murphy 1964). And empirical research strongly suggests that the substance of the law is important in partially explaining judicial behavior (i.e George and Epstein 1992; Hagle 1991; McGuire 1990; Ignagni 1990; Segal 1986, 1984) However, simple common sense and casual observation tell us that such an explanation is only a partial description of the actual decisional process used by judges. Statues are often vague. Precedents often conflict or do not exist (see Pritchett 1969, 31). When such conditions exist, reasonable justices can and do disagree in their reasoning, and this disagreement must have its antecedents in some extra-legal basis. Further, other political actors obviously recognize the inherently political nature of the judiciary. Robert Bork would not have suffered the fate that he did if members of the Senate thought his decisions would have been basically the same as any other

nominee's. At the state level, there has been a trend over the last two decades to institute some kind of merit system (the "Missouri Plan") for selecting justices. This is usually justified as an attempt to take "politics" out of the system (Bunch and Casey 1990; Watson and Downing 1969), strongly implying that judges are perceived as political actors by the non-academic public.

But we need not rely only on common sense and anecdotal observation — rigorous empirical research has fairly convincingly demonstrated that judicial decisions can be explained at least in part in reference to a whole host of non-legal factors. It seems clear that in some circumstances, the backgrounds and demographic characteristics of judges

influence decisional patterns (i.e. Tate 1981; Ulmer 1973). Many judges also have ideological predispositions that may color their decisions (Segal and Spaeth 1994; Segal and Cover 1989; Danelski 1966; Schubert 1960), a tendency whose influence is partially conditioned by the degree to which a judge's own judicial role conception permits such personal influences (Scheb et al 1989). Other influences which operate on judges may include such things as public opinion (Hall 1987; Cook 1977) and small group and collegial influences (Spaeth and Altfeld 1985; Rhode 1972). One line of inquiry has focused upon personal characteristics or attributes of jurists, attempting to identify correlations between personal attributes and patterns of decision making. Many studies have shown correlations between certain personal attributes and case outcomes in certain types of cases (i.e. Welch et al 1988; Tate 1981; Ulmer 1973). The idea is not that the attributes themselves cause particular decisions, but rather that the attributes are objective, measurable indicators of underlying values or attitudes. This being the case, it is not surprising that the one attribute variable which consistently produces robust results is a judge's political party (see Gryski and Main 1986; Beiser and Silberman 1971; Adamany 1969; Nagel 1961). Certain attributes — race, gender, birth order, previous ex­ perience, etc. — may indeed generally be associated with particular political attitudes, but at times the connection is weak. However, one's partisanship is perhaps the best surrogate for political preferences. It is an attribute that is chosen rather than inherited at birth, and the choice is loaded with political meaning. People call themselves Democrats or Republicans because their choices reflect the political values that they hold. Hence, background and attribute analysis indicates that ideological predispositions make a difference. Other techniques have been utilized to document this phenomenon. Scholars have attempted to uncover political attitudes of judges through content analysis of their speeches (Danelski 1966) and newspaper editorials (Segal and Cover 1989). Scalogram (Rohde and Spaeth 1976) and factor analysis (Schubert 1962) have also been employed. The results are clear: on at least some issues, different justices display different voting tendencies, and these tendencies are due in part to differences in judges' policy preferences. Hence, any comprehensive study of judicial decision making ought to take this into account. As a result, in this dissertation, attempts will be made to control for the effect of an individual's ideology, so that we might more clearly see the effects of the other variables under consideration.

This dissertation is about judicial relations with the executive branch. Much has been written about this relationship at the federal level, but little has been written about it at the state level. In fact, we know far more about how state governments fare in front of the United States Supreme Court than we know about how they fare in front of their own state courts of last resort (Kearney and Sheehan 1992; Douglas and Catalano 1988; Epstein and O'Connor 1988; Baker and Asperger 1982; Howard 1982; Morris 1987; Jordan 1985; Jacob 1984, 93). This is unfortunate, since, as Flango and Ducat note, "state courts operate within the context of the American federal system, and share a common language, 11 history, and tradition. Accordingly, unlike cross-national research where controls for diverse customs and laws must be instituted, the American states are an ideal laboratory for non-experimental comparative research" (1979, 25-6). Others

(i.e. Glick 1992; Hall 1992; Tate 1983) have called for more comparative judicial research. This dissertation heeds these admonitions.

In spite of the similarities between federal and state judicial systems, there are three notable ways in which the

relationship between the executive and the judiciary in

individual states may differ from the relationship at the federal level: 1) the role of the chief executive in the judicial selection process, 2) the resources that the executive branch may bring to bear in litigation, and 3) the relationship between the chief executive and his or her

representative in court, the state attorney general. Each of

these differences ought to have certain implications for the interaction between the two branches. For the remainder of this chapter, I would like to focus on #1. I will discuss the conditions under which a state court of last resort might or might not be predisposed to render decisions that are favorable to the executive branch in general. Then in the following chapter, I would like to dis­ cuss the office of state attorney general, and the various implications contained therein. Hypotheses will then be developed and tested. 12 FACTORS CONDITIONING JUDICIAL SUPPORT FOR THE EXECUTIVE I would argue that there are at least five reasons why a justice or court chooses to cast his (its)16 vote in favor of the executive. These are: policy agreement, party loyalty, a shared governmental perspective, gratitude, and deference. Policy Agreement — As discussed above, I believe that

state supreme court justices to some extent seek to achieve

policy goals through their decisions. Hence, I expect that when state supreme court (SSC) justices are presented with a case in which the state or the governor is a litigant, jus­ tices will vote with the executive if doing so fulfills a policy goal. This has previously been fully developed, so

there is no need to duplicate those arguments here. Party Lovaltv — As argued above, one's choice of political party affiliation reflects the political values one holds. Hence, it should not be surprising to find that justice of the governor's political party support the state at higher rates than justices of an opposing party. Much of this support would obviously be due to ideological and policy agreement. But there are likely to be other bonds working as well. Many justices have been party activists, and in many states they must run on a partisan ballot. Such justices may feel a feeling of loyalty toward the party and other officeholders of the same stripe. This loyalty may well make justices more likely to give the executive the “benefit of the doubt", especially when the case is a close one or falls into 13 a justice's "zone of indifference" (Johnson and Canon 1984). The same sense of obligation which causes a Senator to change

his or her vote in order to uphold a presidential veto almost certainly exists in other political relationships as well. Hence, Hypothesis #1 — Justices belonging to the same party as the governor should support the executive branch at higher rates than justices belonging to the opposite party

Shared Governmental Perspective — In his pathbreaking study of executive/judicial relationships, Robert Scigliano (1971) attempts (as have many other scholars) to explain the high success rate of the federal executive branch before the U.S. Supreme Court. Policy agreement is one of the factors that he cites as most important (1971, 185) . Yet he also suggests that it may be due to the fact that the two branches share a "similar governmental perspective, in that the power they exercise is generally concerned with the execution or enforcement of law" (1971, 182). This shared perspective, he suggests, will more often keep the judiciary and the executive on the same side of a question than on opposing sides. While this initially seems compelling, I do not give it as much importance as Scigliano seems to. I would further argue that for analytical purposes, his theory is virtually impossible to verify. Both problems stem from the fact that Scigliano only speaks of this shared perspective in the most general terms. It would seem to me that any institution of government — legislative, executive, or judicial — has a 14 "shared governmental perspective" almost by definition. Since each is a part of the government, the authoritative allocator of values in society, each has a vested interest in ensuring that the polity abides by the dictates of the government.

However, this is so obvious and taken for granted that this "shared perspective" probably rarely ever is a major factor in a political actor's decision-making calculus. Members of all three branches agree that the policies adopted by the government ought to be followed by the public at large. But the agreement often stops there. In spite of this "shared governmental perspective" that governmental policy ought to be obeyed, the three branches can (and often do) have substantial disagreement over what the content of that policy should be. And when it comes time for a supreme court justice (or a member of Congress) to cast a vote, the policy stance of the decision maker is probably going to win out more often than some abstract sense of governmental camaraderie — particularly if the competing policymakers are of different political parties. Furthermore, this is a study about state supreme courts. Scigliano's logic about a shared perspective between the executive and the judiciary presumably holds true for all fifty states. And yet support rates for the executive branch on SSC's vary across states.17 Hence, as an analytic tool to explain differences across states, the "shared governmental perspective" is not particularly useful. However, there is something valuable here. Scigliano is correct when he says that both branches of government are interested in enforcing the law (1971, 3). This essentially means that each branch of government is interested in securing faithful implementation of its policies. Scholars have argued that the U.S. Supreme Court, in spite of its concern for

"rights" and "duties" (Horowitz 1977), shapes its opinions to a greater or lesser degree with this goal in mind. For example, Ulmer (1971) relates the bargaining and negotiating that occurred during the draftingof the opinion in Brown v. Board of Education18. The Court was willing to forego a

stronger statement of principle in order to secure a unanimous opinion, which it perceived would provide potential opponents of the Court's decision fewer opportunities to justify resistance. Similarly, Rohde (1972) finds some empirical support for the notion that the members of the Court are more likely to form larger-than-minimum-winning coalitions when a case involves a threat to the Court's power — those cases in which the Court is in the greatest danger of being ignored or opposed by opponents of the decision. And as a final example, there has only been one U.S. Supreme Court case in history in which each member of the Court individually signed the court's opinion, a case which directly involved blatant disregard of one of the Court's previous desegregation mandates — Cooper v. Aaron.19 16 As these examples suggest, the U.S. Supreme Court is

indeed interested not merely in announcing legal principles, but also in ensuring that these principles are followed in practice. Sometimes it is necessary for the Court to strike a balance between these two goals. Although no empirical research has been conducted, there is no reason to believe that state supreme courts are not characterized by the same motivations. They, like the U.S. Supreme Court, are part of the governmental triumvirate, and I think it is safe to assume that they would like their policy dictates to be faithfully implemented. As Johnson and Canon (1984) and others have noted, a court loses some of its legitimacy when its commands are ignored or disobeyed, and legitimacy is one of the few sources of real judicial power.

And this is where the governor enters the picture. In most instances, a SSC must rely on the executive branch to enforce its commands, just as the U.S. Supreme Court must rely on the president (see Salokar 1992, 177). Executives must take an oath to enforce the law, regardless of its content.

Yet human nature being what it is, the vigorousness of the governor's attempt at enforcement is likely to hinge at least in part on whether or not the governor agrees with the policy (Canon 1991; Johnson and Canon 1984; C. Johnson 1979; Muir 1967; R. Johnson 1967). Thus, if a SSC wants its policies to be implemented, rather than remain mere dicta which are widely 17 ignored, a court would be wise to take the opinion of the governor into account. Hence, justices can be expected to look to the governor for some indication of how much help the executive branch will provide in carrying out the will of the judiciary. Often the justices will not be able to discern a gubernatorial stance — and in many cases, he or she won't even have one, just as presidents don't have a stated opinion on many bills

considered in Congress. However, if the state itself is a

party to a judicial proceeding, the chances that the governor will have a position are greatly increased. The point to be emphasized here is that since justices are interested not only in the articulation but also in the successful implementation of policy, and since the governor is a key actor in aiding or obstructing implementation of that policy, justices may modify their behavior in order to take the governor's views into account. This is most likely to be true in those situations in which the governor's position can be discerned.

Deference to the Executive — A sizable body of scholarship has arisen dedicated to the proposition that judicial behavior can be at least partially explained by jurists' role orientations. Although one of the weaknesses of this work is that the role orientations which are "discovered" can merely reflect the content of the questions that the analyst has asked (i.e. Flango and Wenner 1975), one 18 contrasting pair of role orientations seems to form the core of distinction between jurists: "law makers" vs. "law interpreters" (Wold 1974), or more commonly, "judicial activists" vs. "judicial restraintists" (Scheb et al 1989; Canon 1983; Gibson 1978; Brennan 1985; Meese 1985).

These phrases are loaded with meaning. Bradley Canon (1983) argues that when scholars and lay people speak of "judicial activism", they may be referring to as many as six different dimensions, even though at any given time the speaker is usually only emphasizing one or two of these dimensions. Thus, the analyst who is planning to use judicial

activism or restraint as an analytic tool ought to plainly state which dimension is most salient to the purpose at hand. For the purposes of this dissertation, Canon's dimension of judicial activism entitled "availability of an alternative policymaker" fits best.

At the national level, it is often argued that the Court ought to defer whenever possible to the legislature or, more relevant to this dissertation, the executive. This school of jurisprudence is usually justified in terms of legitimacy and democracy: In a democracy, the will of the majority of the citizens ought to be respected unless it blatantly violates the Constitution. And since Congress and the president are popularly elected, and the Supreme Court is not, the two popularly-elected branches can more legitimately claim to represent the popular will. Hence, the Court ought to defer 19 to them in most instances. To fail to do so is a perversion of democracy (see Graglia 1986; Bickell 1962; Eakin v. Raub20) . Naturally, this is both a normative and empirical debate. Whether or not we agree that judges ought to behave with restraint, it is clear that some judges hold this conception of their judicial role, while others consider themselves to be more activist (see Wold 1974) . However, the distribution of activist and restraintist judges should not be uniform among state supreme courts nationwide. As I shall argue below, certain institutional arrangements should have an effect on a justice's judicial role orientation. Gratitude — Finally, a supreme court justice might be willing to support the executive purely out of gratitude. This is most likely to happen when the justice has obtained his or her position by virtue of appointment by the chief executive. Apart from policy agreement and the other factors mentioned above, the justice may simply wish to say "thank you" to his benefactor and may support the executive at higher levels than he or she might otherwise. "Gratitude" is virtually impossible to quantify without conducting interviews, and hence has not really been discussed in past investigations of judicial/executive relations. However, it is one possible explanation for the results obtained by Segal (1989) in his study of the U.S. Supreme Court. Segal hypothesized that justices would support different presidents in the following descending order: the 20 president that appointed them, later presidents of the same party, and later presidents of the opposing party. His

results were not resounding: he found that justices do tend to support the presidents that appointed them at the highest rates, but when it came to later presidents, partisanship made little difference. I would argue that these findings are consistent with gratitude as an explanatory factor. The justices support "their" president at higher rates, and

thereafter feel more free to follow their own jurisprudential thought regardless of the partisan identity of the executive. To summarize, thus far I have offered five reasons why state supreme court justices might be inclined to support a governor in cases involving the state: 1. policy agreement 2. party loyalty

3. the justices are interested in enforcement of their policies, and look to the governor as a gauge of executive branch support

4. a judicial role conception which stresses deference 5. gratitude

The preceding comments were general ones, supposedly

applicable to judicial/executive relations in every state. However, as obliquely suggested above, the influence of these factors ought to be dependent upon the institutional arrangements found in the various states, which themselves vary. Specifically, justices who are appointed by a governor are more likely to be subject to the above influences, and 21 hence, should support the state more often than elected justices. In the following section, this argument will be developed. JUDICIAL SELECTION SYSTEM AND ITS EFFECTS

The research on judicial selection systems is one of the most consistently futile in the whole realm of the judicial subfield. Scholars have repeatedly attempted to link judicial selection systems with other phenomenon and have had little luck (see Baum 1990, 114) . It appears that the selection

system does not have a very strong effect on either the types of people selected or their "quality" (Dubois 1983; Slotnick 1984; Glick and Emmert 1987; Flango and Ducat 1979; Canon 1972; Watson and Downing 1965; Jacob 1964). Nor have scholars been able to find a strong effect linking selection systems with case outcomes (Allen 1991; Atkins and Glick 1974). Despite this failure, I do not believe we should yet

conclude there are no substantive differences that result from the choice of judicial selection system for at least two reasons. First, scholars tend to focus on formal selection systems, leaving aside the fact that in most instances it is the governor who actually appointed many of the justices, even

in states that provide for judicial elections (see Champagne

1986; Dubois 1983; Herndon 1962). Studies ought to take this fact into account. Secondly, the choice of dependent variable has often been insufficiently justified by theory. I see no reason why we should a priori expect one type of selection 22 system to produce more Democrats (Glick and Emmert 1987) or justices with markedly different educational qualifications (Dubois 1983). The linkage between the structural arrangement and the outcome is not well explicated. It has been very close to a fishing expedition, with scholars looking for any difference in outcome, hoping that at least some of them will

pan out. I believe that judicial selection systems probably do make a difference to case outcomes, but obviously not in all cases, and perhaps not even in most cases. What is needed is a clearer conception as to what subset(s) of cases in which selections system might make a difference in the outcomes. I

would argue that one type of litigation in which it might make a difference would be litigation directly involving those institutions and people responsible for selecting judges. For

example, although none of their results were statistically significant, Atkins and Glick found that judicial selection systems were related to the frequency with which the state won in court as an appellant (1974, 441). I am arguing here that the method by which state supreme court justices are selected should have an effect upon the degree to which they will feel inclined to support the executive branch. Specifically, justices who are appointed by a governor should be more likely to support him than would justices that are elected on either a partisan or non-partisan ballot. I would argue that the judicial selection system 23 alters the effects of three of the factors for gubernatorial

support developed previously: policy agreement, deference to

the executive, and gratitude. First let us consider policy agreement. I have accepted as reasonable the proposition that if the governor takes a position that is consistent with the policy preference of a particular justice, that justice will support the governor's position, all else being equal. The frequency with which the

two actors will be in harmony on policy will be dependent in part on how that justice was selected: a justice selected by a governor ought to share that governor's policy preferences more often than an independently elected justice for a number of reasons. Most obviously, during the course of the appointment

process, the governor presumably has an opportunity to screen judicial candidates in order to ensure just such policy agreement. This opportunity for screening does not guarantee that a governor will get someone who shares his ideological predisposition. Governors may not avail themselves of the chance, or if they do, they may not perform a screening which

is sufficiently rigorous. Further, as a cursory examination of behavior on the U.S. Supreme Court demonstrates, justices may experience an evolution of their jurisprudential thought while on the bench (see Baum 1992; Ulmer 1979) . This important fact aside, appointments provide at least the possibility of screening by the governor; elective systems 24 provide no such opportunity. Hence, in the aggregate, gubernatorial appointments should produce justices more sympathetic to the governor than elective systems.

Current popular political discourse and debate acknowledges that such ideological motivations may underlie gubernatorial appointment of SSC justices. Over the last quarter-century, many states have moved to reform and modernize their courts systems, and the trend has been largely toward the merit plan, also known as the Missouri Plan. First

implemented in the Show-Me State in 1940, some variation of the original plan is now in use in 20 states (Book of the States 1992-93) . The Missouri Plan will be discussed in more detail in Chapter 3, but in essence, it authorizes a non­ partisan commission to present a list of potential nominees — chosen on the basis of merit — to the governor, who must then choose one of the names submitted to him or her.

One of the common justifications offered by proponents of merit selection is that it takes the "politics" out of the judicial selection process. Much scholastic energy has been expended judging whether this has, in fact, occurred, and the results have been unanimously negative (Bunch and Casey 1990;

Watson 1986; Watson and Downing 1969). Summarizing their findings of an extensive study of the plan's operation in Missouri, Watson and Downing state, "...it is naive to suggest (as some of the Plan's supporters do) that the Plan takes the 'politics' out of judicial selection. Instead, the Plan is designed to bring to bear on the process of 25 selecting judges a variety of interests that are thought to have a legitimate concern in the matter and at the same time to discourage other interests. It may be assumed that these interests will engage in the 'politics' of judicial selection, that is, they will maneuver to influence who will be chosen as judges....Whether the Plan eliminates politics in judicial selection is a false issue. Instead, the key issue is whether the particular kind of politics that evolved under the Plan adequately represents the legal, judicial, public, and political perspectives thought to be important in determining who will sit on the bench." (1969, 332) The studies find that the governor remains a very important actor in the selection process. Not only does he retain the ultimate right of selection, he is highly influential in the selection of the members of the Merit Commission. Watson and Downing further report: "Laymen on the commission have also served to channel political considerations into the selection process. In some instances, this has come at the specific request of the governor, transmitted directly or through an intermediary to the lay commissioner. More often, however, politically astute laymen have been able to bring gubernatorial attitudes on candidates to bear on the nominating process simply because they have known, without being told, who have been the friends and enemies of particular chief executives over the course of their political careers. Generally, the Missouri governors have viewed selection under than Plan almost entirely from a personal or political standpoint." (1969, 338-9) Hence, it would appear that gubernatorial selection ought to produce justices more sympathetic to the governor on policy issues than other selection systems.

A second factor which is related to the method of judicial selection and may contribute to judicial support for the governor is the justices' role conception. I have argued 26 above that a justice whose jurisprudence emphasizes a less activist role for the judiciary will be more likely to support the governor. Now I would argue in addition that gubernatorial appointments will be more likely to produce deferential jurists than will either partisan or non-partisan elections. This is likely to be so for a number of reasons. First, a governor may look to the candidate's views on judicial activism and restraint as one criteria for appointment. But far more importantly, when a justice is appointed by the governor, the justice has no "constituency" apart from the governor. Elected justices need to worry about pleasing members of the electorate for their continued tenure in office.21

Recall that a major justification for a jurisprudence of judicial restraint at the federal level largely involves the fact that federal judges are not elected. Hence, in order to prevent an abuse of their power, it is argued that they ought to defer whenever possible to the other, popularly-elected branches of government. Judicial restraint is seen as being one way in which the Court can check its own power. However, this argument loses much of its force when judges are independently elected. In theory, justices can no longer subvert the will of the majority, since if they do, that majority can vote them out of office, just as they do other officials. Justices no longer need to rely on looking to the other branches to discern the will of the "law-making 27 majority" (Dahl 1957), since that majority can impose its will directly on the judiciary. Further, simply having an independent power base may embolden members of a court to chart their own course that is in opposition to the will of the other two branches. Two objections might be raised to this line of thinking. First, it might be argued that judicial elections are such low-information contests that they have virtually no policy content, and hence cannot possibly serve as an independent base of power from which the judiciary could challenge the executive on matters of policy. There can be little doubt that judicial elections are indeed of the low-information variety (Baum 1987; Lovrich and Sheldon 1983, 246; Johnson et al 1978; Griffin and Horan 1983; DuBois 1984). However, this does not make them unique. It has been demonstrated repeatedly that voters have very little information when they vote in many types of elections (i.e. Hinckley et al 1974; Shaffer 1982) . Democracy is predicated on the premise that the voters shall have the opportunity to reject incumbents who do not represent their views adequately — it does not require that the voters take advantage of this opportunity in all instances. Furthermore, as Mann (1978) suggests regarding members of Congress, fear of possibly detrimental electoral consequences may be a very potent weapon for ensuring elite conformance with constituents' wishes, even if those fears have very little foundation in reality. The recent works of 28 Hall (1992) and Hall and Brace (1992) provide ample evidence that this occurs in death penalty cases. Thus, even though elected judges are extraordinarily insulated from electoral loss, they probably feel they owe their allegiance more to the electorate than to the governor.

Another objection might be raised. It might be argued that in most states, supreme court justices must face the voters in one form or another: even in most states that provide for gubernatorial selection, justices must face the voters in periodic retention elections. Thus, there is no reason to expect that deference toward the governor would be a more common value in those states than in states which have contested elections. Further, canons of judicial ethics dictate that judicial candidates ought to emphasize such things as skill, integrity, and temperament rather than policy stances. As a result, contested elections ought not be much different from retention elections in tone and style of campaign, and information available to the voter. There may be some truth in this. However, I take the stance that the two are different types of elections that ought to have different effects on judicial role orientations. Contested elections (whether partisan or not) should be more difficult to win. In such races, there exists an opponent who may be motivated to distribute negative information about the incumbent (see Dubois 1983). This makes the election more uncertain for that incumbent than a 29 retention election, in which there is no guarantee of organized opposition, would be. Despite the fact I am arguing that contested elections ought to produce justices less deferential to the executive than gubernatorial appointments followed by a retention election, I think the objections I have mentioned merit considering another possibility. It sounds logical to suppose that gubernatorial appointees will have a tendency to defer to the governor when they initially join the court, but that the tendency may weaken or vanish after their first retention election. At that time, they may perceive themselves as having received an independent mandate from the people which may embolden them to become more independent. In any case, I believe that it is a possibility worth testing, and I intend to do so.

Finally, the system of judicial selection should be important in determining the effect of a third factor mentioned above: gratitude. The hypothesized relationship is straightforward and relatively obvious: there ought to be more gratitude displayed toward the governor by justices who were initially appointed by the governor. Obviously, in elective states, the governor may assist a judicial candidate, but he or she may also provide no aid or even oppose the candidate's selection. Justices who have been appointed by the governor know exactly who to thank for their jobs. 30 In summary, then, I had previously mentioned five reasons why SSC justices might support the governor. I have now suggested that three of those relationships depend at least in part on the system by which those justices were selected. These are: 1. justices selected by the governor should be in policy agreement with that governor more frequently than elected justices 2. appointed justices should be more deferential to the governor than elected justices

3. appointed justices should be more grateful to the governor than elected justices Hence: Hypothesis #2 — Among members of the governor's party, appointed justices should support the executive branch more than elected justices.

In this chapter, I have presented a picture of judges as actors motivated by a combination of legal, ideological, and political motives. I have developed various explanations as to why a judge or a court might support the executive branch in adjudication. I have further speculated on how this relationship ought to be affected by the manner in which the judges are selected.

However, just as the U.S. states differ from the federal government in that many of them elect justices rather than provide for their lifetime appointment, they also differ on another important dimension — at the state level, the judiciary is usually confronted by a dual executive. In most states, the actor that represents the state in court, the state attorney general, is separately elected and relatively independent from the governor. Hence, although the executive branch is ultimately under the control of the governor, it does not always speak with one voice in its legal strategy. Further, some state attorneys general's offices are more powerful than others, possibly affecting their fortunes in litigation before the state's courts. In the next chapter, I would like to focus more closely upon the office of state attorney general, its relationship with the governor, and the implications of these things for executive success in court. 32

ENDNOTES

1. See, for example, Federalist #51. 2. Among the many scholars engaged in this enterprise are Robert Bork, Alexander Bickell, John Hart Ely, William Brennan, Edwin Meese, Richard Epstein, Bernard Siegan, among others. 3. We know that the court declines to declare laws unconstitutional far more frequently than it strikes laws down. This fact has led to an intense debate, initiated by Dahl (1957) on whether or not the court was living up to its role in democratic society. 4. And naturally, the distinction between these phases is not complete. For example, attempts to remove a judges are both after-the-fact, in that they •'punish'1 a judge for a wrong already committed. But they are also attempts to shape future actions of the court by preventing the justice from casting any more votes. Other tactics are similarly multifaceted. 5. I.e. State v. Schmidt 29 Ohio St. 3rd. 32 (1987).

6. Arizona Newspapers Association v. Superior Court 694 P.2d. 1174 (Arizona 1985). See also Swinford 1993. 7. Conwav. Inc. v. Ross 627 P.2d. 1029 (Alaska 1981). 8. Under 21 v. City of New York 65 NY. 2d. 344 (1985) . 9. Stafford v. Muster 582 SW. 2d. 670 (Missouri 1979). 10. In the Matter of Baby M 109 NJ. 396 (1987). 11. In re Grant 109 Wn. 2d. 545 (1987). 12. 624 P.2d. 1281 (Arizona 1981). 13. 286 NW 2d. 262 (North Dakota 1979). 14. 29 Cal 3rd. 150 (1981). 15. 26 Cal. 3rd. 110 (1979). 33 16. While the micro- and macro-level processes are likely to be moved by somewhat different forces, a court's decision is obviously a direct result of the votes cast by individual members. In any case, analyses will be conducted at both the individual level and at the level of the court collectively. 17. See Chapter 4. 18. 349 U.S. 294 (1955). 19. 358 U.S. 1 (1958). 20. 12 S. & R. 330 (Pa. 1825). 21. Obviously, even in most appointive states, judges must face the voters in a retention election relatively soonafter taking the bench, a fact which is addressed below. CHAPTER II THE STATE ATTORNEY GENERAL AS INDEPENDENT POLITICAL ACTOR

Virtually all research concerning the executive branch in court has focused on the office of the Solicitor General of the United States. The Solicitor General (SG) is the third- highest ranking official in the U.S. Department of Justice and oversees all litigation the federal government takes to the

U.S. Supreme Court. The office is "accountable for all

government litigation before the Supreme Court" (Salokar 1992, 3). With few exceptions,1 all agencies and agents of the federal government must obtain the approval of the SG before they can appeal an adverse decision to the Court. In most instances, it is this office that writes the briefs and participates in oral argument. (For a fuller description of

the office's powers and duties, see Salokar 1992, 12-13; Schnapper 1988, 1195). Much of the literature on the Solicitor General's office is qualitative, describing the workings of the office (i.e. Bell 1981, 1978; Note 1977/78; Griswold 1975, 1969; Burt and Schloss 1968/69; Sobeloff 1955; Fahy 1942; Thatcher 1931).

The quantitative literature that exists reaches a consensus on a major point: the Solicitor General's office is extremely

34 successful before the U.S. Supreme Court. It wins the vast

majority of cases in which it participates, at each stage of the adjudicative process: in the certiorari decision (Salokar 1992, 1988, 35; Schnapper 1988, 1211; Chamberlain 1987, 394; Provine 1980; Scigliano 1971, 176; Griswold 1969; Tanenhaus et al 1965), decisions on the merits (Salokar 1992; Sheehan

1992, 37; Sheehan et al 1992; Scigliano 1971, 178; Brigman

1966), and as amicus curiae (Salokar 1992, 29; Segal 1988; Ulmer and Willison 1985; Puro 1971, 100; Scigliano 1971, 193). This success is relatively consistent from administration to administration, although there is variation.2 Few questions in the social sciences are settled, but the literature that attempts to explain the success of the

Solicitor General approaches as close to a perfect consensus as one could possibly hope for. Essentially, scholars have generally argued that the Solicitor General's office possess a number of resources which few, if any, other litigants possess, and these resources give them a relative advantage over other litigants. Most (i.e. Salokar 1988, 6; Provine

1980, 91; Burt and Schloss 1968/69) dub the Solicitor General a "repeat player", utilizing a concept borrowed from Marc Galanter (1974). This line of reasoning will be developed more fully in the following section. For now, the core of the argument is that because the federal government (and hence the

SG's office) appears before the Court so frequently and has such a high amount of discretion in pursuing appeals, it 36 accrues advantages such as expertise; low start-up costs; the ability to pick both those cases with the highest likelihood of victory and those that will have the maximum impact on the

development of the law; and so on. As a result of these advantages, repeat players should "come out ahead". These arguments are compelling — I have little reason to doubt their veracity. Indeed, they form the basis of much of the discussion which comprises the rest of this chapter. Unfortunately, however, for the most part the validity of this

"repeat player" theory of SG success remains empirically untested. The reason for this is simple: there is little or no variation in the independent variable surmised to be important. The federal government, and hence the Solicitor

General, has always been a repeat player. The powers and duties of the Solicitor General's office have not changed

dramatically over time. Thatcher's 1931 description of the office reads as though it could have been lifted from the pages of Salokar's comprehensive 1992 study. Thus, although expertise and other advantages surely contribute to the SG's success, there is no way to verify this if we consider only the federal government.

One of the goals of this dissertation is to test this "repeat player" hypothesis at the state level, focusing on the Solicitor General's closest state-level counterpart, the state attorney general. The analogy between the two offices clearly is not perfect, as discussed below. Many of the essential 37 advantages of the repeat player are possessed by both offices, however — and since these advantages vary between states, the states become an excellent setting in which to study their effects. The importance of conducting comparative judicial research was stressed in the previous chapter.

To my knowledge, only one quantitative study of the state attorney general's office has previously been conducted. Boersema (1987) argues that the formal powers of an attorney general's office, combined with the political ambition of the officeholder, can explain certain outputs from a state

attorney general's office. This study is valuable, for it is the only extensive quantitative study of the office that exists. Its comparability to the present study is limited, however, for two related reasons. First, Boersema is concerned with many different outputs of the AG's office, including things such as advisory opinions, type of caseload,

and cases appealed to the U.S. Supreme Court.3 He spends very

little time (about two pages in an entire dissertation) examining AG success rates before their own supreme courts — although when he does, he finds some support for the proposition that success rates depend in part upon the formal powers of the office.4

The second reason why Boersema's work does not shed much light on the current topic is related to the first. Since Boersema is concerned with many different outputs of the AG's office and not merely success in litigation, his 38

operationalization of the formal powers of attorneys general is comprehensive, but overly inclusive for my purposes. He includes numerous indicators, but many of them simply are not relevant in explaining litigation success.5 As fleshed out below, my operationalization of this variable is simpler, and hopefully more directly pertinent.

Hence, a portion of this dissertation is dedicated to determining to what extent state AGs' formal advantages in litigation contribute to their success in court. The remainder of this chapter includes a description of the office of state attorney general and a fuller discussion of the concept of the repeat player and the advantages which the AG

ought to possess. Appropriate hypotheses will be developed.

[There are obviously variables other than formal advantages enjoyed by the Solicitor General and state attorneys general which might be important to consider when explaining their success. One shortcoming of both Boersema and most of the works on the SG, is that they fail to consider

any of these other explanations (although Segal 1989 is a notable exception). Some of these — ideology, partisanship, judicial selection system — were discussed in the previous chapter.] THE OFFICE OF THE STATE ATTORNEY GENERAL The attorney general is a state's chief legal officer.

The incumbent's duties are numerous, ranging from controlling litigation involving the state to issuing written opinions for 39 the governor to legislative advocacy. The office exists in all fifty states, and it is constitutionally established in 44 of them. It has been said that "there can be no doubt that the attorney general today must be considered one of the two or three most important and influential figures in state government" (Christenson 1970, 338). The attorney general, like the Solicitor General, serves two masters. Scholars have commented about the conflict the attorney general feels: the AG is a member of the executive branch in 48 states5, but considering the heavily legalistic nature of the AG's responsibilities, the office's ties to the judicial branch are strong (Christenson 1970; Krislov 1959). The AG occupies an interesting and perhaps unique position in our system of separation of powers: an officer of both the court and the executive branch. Abraham and Benedetti have described the AG as "the quasi-judicial officer in the ad­ ministration whose job it is to bridge the gap between law and state practice. The attorney general does not fit neatly within the framework described by the doctrine of separation of powers, since he exercises both executive and judicial functions" (Abraham and Benedetti 1969, 797).

Over the last three decades, as part of a national trend to modernize state government generally (Howard 1988; Sabato 1983; Rosenthal 1981), the state attorney general's office has generally increased its powers and responsibilities (Ross 1990, 12; Morris 1987, 299). The AG is frequently called upon 40 to offer legal advice to the governor and to agencies of the state (Heiser 1982; Chanin 1976; Carberry 1975). Many AGs are members of commissions created for legal or administrative reform. Virtually all have the power to investigate allegations of misconduct by public officials and institutions. Many are involved in the drafting of and advocacy for legislation. In 43 states, they have an independent political power base as a result of having been elected in a state-wide election, which they may use to advance the public interest as they perceive it (see Thompson 1974). As a result, AGs have become more aggressive in pursuing investigations and litigation in the areas of consumer protection (Miller 1988; Bernstine 1977; Kazanjian 1973), public health and safety (Webster 1988), antitrust litigation (Brown 1988; Korenberg 1982), environmental protection (Note 1974), civil rights enforcement, insurance and securities regulation, utility ratemaking, Medicaid fraud (Ross 1990), and of course, criminal justice (Van de Kamp 1988; DeLong 1934). Generalizing about the powers of state attorneys general is a task that ought to be performed with some care. As is the case with governors, the range of powers can vary widely from state to state. Referring to the development of the office of attorney general, the National Association of Attorneys General characterize the office as possessing "an astonishing array of mutations which make it altogether 41 impossible to reach any sweeping generalizations..." (NAAG 1977, 15). Nevertheless, certain conclusions can be reached. Lynne Ross (1990, 37-38) lists the following powers that state AGs generally have (among others):

— The Attorney General has the duty to appear for and to defend the state and its agencies — The Attorney General has the right to control litigation and appeals. — The Attorney General has the right to intervene in legal proceedings on behalf of the public interest. — The Attorney General has the authority to prosecute criminal activity, in the absence of express legislative restriction.

The preceding is merely offered to suggest the scope of activities in which state AG's offices engage.7 As mentioned above, I am concerned with explaining the amount of success that the executive branch experiences in court. Hence, the only powers which concern me are those that are directly relevant to the AG's conduct of litigation. Therein lies an important differences between the Solicitor General and the state attorney general: at the federal level, responsibility for litigation-related activities and non-litigation related activities are divided between two individuals — the Solicitor General and the Attorney General respectively. There is no such division of labor at the state level — the attorney general is responsible for it all in most states.8 However, I am only concerned with litigation-related activities. Naturally, the state AG fulfills many of his or her duties without resorting to litigation. For example, a study of the Consumer Protection Division of the Massachusetts' AG's office revealed that the office attempted to resolve as many disputes as possible without going to trial (Silbey 1980/81) . In this sense, the AG's role is little different from Macauley's (1979) description of the role of the private attorney in dissuading clients from seeking a total victory in the courtroom and encouraging them to accept a partial settlement. However, litigation remains an important part of the office's tasks. Most published descriptions of the office emphasize its role in litigation. And as indicated in the previous chapter, the stakes in these cases is often high.

Like the Solicitor General, the state attorney general is the individual responsible for representing the state in court. In a number of states, the AG has a good deal of discretion over which cases the state shall appeal to the supreme court. In some states, public officials must obtain the approval of the AG's office before they may take an appeal forward (Oberly and Silverman 1988, 94; Haflett 1984, 1055). Often, if two agencies are in conflict, or are arguing contradictory legal principles, it is the job of the attorney general to mediate or arbitrate the dispute to help ensure that the state pursues a coherent legal agenda (Ross 1990, 42). The attorney general and members of his or her staff 43 prepare briefs and amicus briefs, and appear before the court to argue cases orally. As the state's chief legal officer, the AG is unguestionably a repeat player in the sense implied by Galanter (1974). Galanter argued that repeat players have certain advantages which make them "come out ahead” when com­ pared to one-shotters who are not so advantaged. In the following section, I would like to review these advantages as they may apply to the state attorney general. (Since there has not been a great deal of scholarly

investigation of the office of the state attorney general, I will have to draw substantially on the literature concerning the Solicitor General, drawing inferences and contrasts where appropriate. In some instances, quotes and information are derived from a questionnaire administered for this dissertation. See Appendix B.)

THE STATE ATTORNEY GENERAL AS REPEAT PLAYER The notion of the repeat player was given its widest exposition by Marc Galanter. According to Galanter, a repeat player (RP) is "a unit which has had and anticipates repeated litigation, which has low stakes in the outcome of any one case and which has the resources to pursue its long-term

interests” (1974, 98). The RP can be contrasted with the "one-shotter”, who is an individual or organization which does not appear frequently in court. Since repeat players participate in litigation often, Galanter and others (i.e. 44 Atkins 1993) have argued that they ought to enjoy certain advantages not enjoyed by the occasional litigant. A relatively crude measure of whether or not a litigant is indeed a "repeat player" is simply the percentage of a court's docket that consists of cases involving the litigant. By this standard, it is clear that both the U.S. Solicitor

General and state attorneys general are repeat players. Be­ tween 1959 and 1989, the SG appeared in 48.5% of the cases decided on the merits by the U.S. Supreme Court (Salokar 1992, 21) . My own research demonstrates that state attorneys general also appear frequently before their own state courts of last resort. In the court terms included in this study, participation by the AG's office in cases decided with written opinion ranged from a low of 10.1% in North Dakota (1981) to a high of 60.5% in Arizona (1981). In total, the AGs in the twelve states included in this study participated in 33.8% of all cases decided with written opinions. Thus, if we consider simply sheer numbers, both the Solicitor General and the state attorney general are repeat players. Hence, among the advantages they ought to possess are: 1) On-aoinq Relationship with the Court — Galanter argues that RP's "have opportunities to develop facilitative informal relationships with institutional incumbents" (1974,

99) . Much has been written about the informal relationship between the Solicitor General and the U.S. Supreme Court. Al­ though the SG is a formally a member of the executive branch, 45 the office has a perceived institutional duty to the Supreme Court as well (see, for example, Salokar 1988, 8; Wilkins 1988, 1181; Caplan 1987; Lee 1986, 595; Griswold 1969; Burt and Schloss 1968/69). The Court trusts that the SG will ap­ peal only those cases which are truly worthy of the Court's attention. This "gatekeeper" function is a valuable service to the members of the Court, and the relationship between the two sets of actors has been institutionalized as a result (Harmon 1988, 1092; McConnell 1988, 1201; Provine 1980, 87; O'Connor 1982/83; Griswold 1975; Scigliano 1971, 170; Stern

1960, 155) . A symbol of this special relationship is the fact that the SG has an office inside the Supreme Court building itself, the only attorney so honored.

As a result of this intensive interaction, the SG and his or her staff eventually come to know the thinking of the justices both individually and collectively (Schnapper 1988,

1213; Schwartz 1988, 1138; Jacob 1984, 90; Provine 1980, 87; Burt and Schloss 1968/69; Brigman 1966, 67). This knowledge is more complete and more immediate than that obtained by any other set of lawyers in the world. The office is thus in a position to select those arguments it suspects at least a majority of the Court will be sympathetic to. Furthermore, since the SG is so familiar to the Court, justices may be inclined to give him the benefit of the doubt close cases. Hence, the SG certainly does have a "facilitative informal relationship" with members of the Court. 46 Unfortunately, little has been written of the analogous relationship at the state level. It is commonly acknowledged that state AGs are more independent of the governor than the SG is of the president (Ross 1990, 42; Morris 1987; Palinscar 1979, 264; Note 1977/78, 345; Thompson 1974; Christenson 1970; Abraham and Benedetti 1969, 798; Krislov 1959, 82). However, this does not necessarily mean that they feel a corresponding increase in kinship with the state's courts. In fact, Thompson (1974) finds that when interviewed, only a very small percentage of state AGs stated that the most important role of the attorney general is to serve as an officer of the court

(1974, 33).

However, some informal facilitative relationship is likely. As mentioned above, virtually all commentators on the office describe it as bridging the gap between the executive and the judiciary, and in all states the AG is considered an officer of the court. In questionnaires and interviews con­

ducted for this dissertation, while none of the state AGs

questioned stated that their duty to the court was their primary duty, each of them acknowledged a responsibility to it that was taken seriously by the office. Sudnow (1965) quite

vividly describes the symbiotic relationship which arises be­ tween the prosecutor, judges, and public defender in our nation's criminal trial courts. Although the system is designed to be adversarial in nature, the day-to-day interaction among the actors results in the development of 47 courtroom norms that emphasize the expeditious processing of cases rather than providing a truly adversarial proceeding.

I do not know for certain that this phenomenon occurs between state AGs and their courts of last resort.9 But my own questionnaire, the extant descriptions of the AG's office, and extrapolation from the experience of the Solicitor General lead me to expect it. 2) Credibility — A second advantage repeat players enjoy according to Galanter, is the fact that they can (and must) "establish and maintain credibility as a combatant" (1974. 99). Since repeat players know that their relationship with the court is an on-going one, they seek to maintain a reputation for credibility from which they might draw. There is no doubt that U.S. Solicitors General strive toward this goal. Former SGs, reflecting upon their tenures in office, often emphasize the office's reputation for intellectual rigor and candor as one of its greatest assets. Former Reagan SG Rex Lee has written, "There has been built up...a reservoir of credibility on which the incumbent Solicitor General may draw upon to his immediate adversarial advantage" (Lee 1986, 60; see also Harmon 1988, 1092; Provine 1980, 87; Griswold 1975; Scigliano 1971, 183; Stern 1960, 157; Sobeloff 1955; Thatcher 1931).

There has been little written that might indicate whether or not state supreme court justices perceive their attorneys general's offices as being credible litigators. Certainly, individual AG's may been seen as political opportunists or legal hacks. However, the questionnaires administered for this study demonstrate that the staffs of AG's offices profess that they strive to perform their work with a high degree of professionalism and skill. When asked if they felt that justices perceive the AG's office to be more credible than most other parties appearing before the state's high court, one AG responded, "Not only do I think that, the justices have told me personally that this is true." Another indicated that he or she was not sure whether or not the court perceived the office as credible, but "we try very hard to put forth a good product, and I hope that they take notice." These are profes­ sional men and women who obviously take their jobs seriously, and SSC justices probably realize it, on the whole.

3) Expertise — Another advantage cited by Galanter which state AGs ought to enjoy as repeat players is that "RPs develop expertise and have ready access to specialists. They enjoy economies of scale and enjoy low start-up costs for any case" (1974, 98) . Others (i.e. Salokar 1988, 5) have noted that this is true for the Solicitor General. The office can (and does) hire some of the top law school graduates in the nation. Working so consistently before the same court gives the office intimate knowledge of how the justices are likely to vote, both individually and collectively. These statements can largely apply to the state AG's office as well. While most of them cannot hire the best law school graduates in the country, they can make up for it in other ways. In recent years, AGs offices have modernized and gained power (Morris 1987). Many have established "sections that handle general litigation and sections responsible for specialized litigation such as consumer protection, environment, and anti-trust cases" (Ross 1990). Furthermore, the workload of most AG's offices is rather routine, even more so than the workload of the Solicitor General. A large proportion of cases each AG's office argues in front of the SSC generally falls into a relatively small number of categories.

Those individuals interviewed for this dissertation generally reported that their experience has given them an inside advantage. One reports that "from arguing and reading opinions, you feel you get an opinion on what areas are likely to interest what judges, particularly in labor cases", although he also cautions that this knowledge may be erroneous on occasion. Another asserted that "In criminal cases, my staff is extremely knowledgeable about the justices and how they think." However, this opinion is not universally held. While five of the seven AGs offices strongly agreed that they knew the justices' thinking, two were unsure. One stated that, "We have some understanding of the thinking of individual justices, but it is difficult, if not impossible, to predict how the justices will act in particular cases." One office reported that it was "not usually" able to predict 50 justices' votes. But this was a minority view. Hence, it does appear that state AGs and their offices become specialists in litigation concerning the state. 4) Ability to Plav for Long-Term Gain — A fourth advantage enjoyed by repeat players cited by Galanter is the fact that "RPs can play for rules as well as immediate gains...(and they) can also play for rules in litigation itself" (1974, 100). For the Solicitor General, at least, this is an apt description. Numerous scholars (i.e. Schwartz 1988, 1129; Chamberlain 1987, 393; Jacob 1984, 89; O'Connor 1982/83, 259; Provine 1980, 91; Burt and Schloss 1968/69, 1456; Brigman 1966, 68) and former SGs (i.e. Lee 1986; Gris­ wold 1969) have noted that the SG's office can afford to be quite selective when choosing the cases it will appeal to the

Supreme Court. Prominent among the criteria that argue in favor of appeal are two things: a) Immediate Gains — This is, quite simply, the ability to win the present case. Each year the Solicitor General gets hundreds of requests to appeal cases. Since the office has a perceived duty to avoid flooding the Court with appeals, it enjoys a good deal of discretion in deciding which ones to take forward. In such a situation, it would be only rational to choose those which offered the best chance of victory. As Herbert Jacob notes, "the Solicitor General can often choose to delay the appeal of an issue until the right case or the 51 right moment comes along. Most private litigants cannot do this" (1984, 90). b) Long-Term Gains — this is Galanter's "ability to play for rules in litigation itself." The SG tends to refuse to pursue cases which involve issues which are not likely to recur or cases that will not break any new legal ground. Most scholars appear to portray this as the more important of the two criteria, although the two considerations are interrelated: significant cases must be won in order to achieve the administration's goals. The office's resources, while vast, are not unlimited. The winnable but frivolous case isn't worth any more effort than is the case that is important yet impossible to win (see Lee 1986; Jenkins 1983; Puro 1981, 228; Burt and Schloss 1968/69; Stern 1960). "The amount and variety of government litigation are sufficiently great to permit the Solicitor General's selection of test cases to expand, contract, or clarify Supreme Court precedents" (Burt and Schloss 1968/69, 1456). Former SG Erwin Griswold notes that "it is inevitably the Solicitor General's function to consider not only the immediate case, but also the collateral consequences of the position he may take in presenting it", one of which he believes is the development of the law (Griswold 1969, 528, 534). Clearly, the Solicitor General possesses this advantage of the repeat player. Does it also apply to the state attorney general? Unlike the previous advantages, which I argued that state attorneys general indeed should possess, probably no state AG enjoys this advantage to the extent that the SG does. The key to the SG's advantage in this area is the office's great discretion over appeals. Virtually every federal agency which wants the Supreme Court to review a case must receive the blessing of the SG, and while he will try his best to accommodate their requests, the decision to appeal is ultimately his. Even those agencies which are not required by law to gain the SG's approval virtually always seek it nonetheless (see Scigliano 1971, 174; Brigman 1966, 25; Stern 1960, 157).28 Most state attorneys general do not enjoy this degree of independence. As Herb Jacob states, Attorneys general in the states do not usually control appeals in the same manner that the United States Solicitor General does in federal courts....Because most cases originate with local agencies, the attorney general usually does not control access to the state supreme court in the same way that the Justice Department does. Nor does his office exert the same degree of control over the kinds of cases that state appellate courts hear, for those courts are much more generous in granting appeals than is the U.S. Supreme Court. (1984, 94) Hence, the state AG's office faces a double liability: it cannot completely control the flow of cases from the executive branch to the court, and the court itself is more willing to hear these "unauthorized" appeals (see also Morris 1987) . And to make matters even less palatable, in many states the AG's office is bound by law to represent the agencies, even if it disagrees with that agency's position. Ross notes that "the prevailing view is that the Attorney 53 General should represent the agency if it is acting within the arguable range of its legal discretion" (1990, 45).

My questionnaires do reveal that AG's discretion is sometimes limited. While four of the seven offices returning questionnaires reported that they had "total discretion" over the decision to pursue appeals, three reported possessing only "a fair amount" of discretion. In these latter cases, often an AG's office will negotiate with public agents, attempting to use persuasion and pedagogy in order to convince them to appeal or not appeal a case, depending on the AG's preference. But if an agency deeply desires to pursue a case with which the AG disagrees, the AG generally "puts his best lawyer on it and gives it the best shot he can" (Interview). Several AG's assistants noted that, like all lawyers, the interest of the client is supposed to take precedence, and AGs must distinguish between their political desires and their professional responsibilities. A striking example of a state AG doing this has occurred recently here in Ohio. As a state legislator, Democrat Lee Fisher was a consistent advocate of abortion rights. Upon taking office as attorney general, however, Fisher and his office have vigorously defended Ohio's

24-hour waiting period and parental notification laws regarding abortion, in spite of Fisher's personal opposition to these laws (see Johnson 1992). Given these constraints, it appears that some state attorneys general are not able to play for short- or long-term gains as successfully as the Solicitor General can. However, there are two points to be kept in mind. First, while few state AGs have the control over litigation that the SG does, many of them do possess a good deal of discretion nonetheless.

Ross reports that AGs "typically have broad discretion in litigating matters. The AG usually controls decisions regarding initiating, pursuing, and appealing lawsuits" (1990, 85) . As is the case with governors and state legislatures, AGs do not have identical grants of power from state to state. Those AGs which have more discretion over their litigation should have greater success in both the long and short term than AGs with less discretion. A second point is that while there are certain types of cases which AGs must defend in court, the office of the attorney general has undergone a resurgence in the last two decades (Tierney 1992). As a part of an overall trend of professionalizing the institutions of state government (i.e. Sabato 1983; Rosenthal 1981), the AG's office has become a more active participant in state government. As Thomas Morris describes it,

Prior to the 1970's, state attorneys general tended to look upon their role as being merely ministerial functionaries of the state administration....The size and responsibilities of state attorneys general's offices expanded in the 1970's to include public advocacy roles in such areas as consumer protection, antitrust enforcement, utility rate intervention, and environmental protection. Largely passive attorney general's offices were transformed into activist ones. A new breed of state attorney general, younger and better-educated than their 55 predecessors, increasingly exploited the political advantages of their office. (1987, 299). Hence, although state AGs are not as selective as their federal counterpart, in most states they will be able to be more selective than many private litigants. More importantly, the amount of discretion varies from state to state, and this variation should be important in determining how an AG will fare in court. 5) Other Advantages - Galanter mentions other advantages, related to those discussed above, which repeat players possess. While there is really no way to tell whether state AGs actually possess these advantages (since they seem difficult to operationalize), they ought to be mentioned.

Galanter argues that repeat players are better able to discern which rulings are more likely to be substantively important (that will "penetrate") , and which are likely to be mostly symbolic. He says that they "can trade off symbolic defeats for tangible gains" (1974, 103), and focus their resources on achieving these gains. In addition, RPs can "play the odds" over a long series of cases, adopting a minimax strategy. We do not know for sure if the AG, or any repeat player, does these things. All that can be said is that AGs are no less likely to do these things than any other repeat player might. In this chapter, I have attempted to establish that the state attorney general's office is indeed a repeat player before a state's supreme court, and that as such, it enjoys many of the advantages over other litigants that Marc Galanter 56 suggests it ought to. Thus, the expectation that arises from the preceding discussion is that AG's ought to win at high rates. But what constitutes "high”? It is known that judges may respond to many different stimuli when deciding a case. As suggested in the previous chapter, one factor is ideology. Hence, an AG's success over the course of a term should be partially dependent upon the ideological positions the office takes.

Similarly, research focusing on the U.S. Supreme Court has suggested that the Court grants certiorari as an "error- correction" strategy, in order to reverse lower court decisions with which it disagrees (i.e. Brenner and Krol 1989). The evidence regarding a similar phenomenon at the state level is scant, but suggests that respondents tend to win more frequently than petitioners (see Wheeler et al 1987; Mason 1978; Baum 1976). Utilizing an error-correction strategy is only possible when a court has discretion over whether or not to hear a case, and no state supreme court has the amount of discretion bestowed upon the U.S. Supreme Court. But nevertheless, research has suggested that the side a litigant takes is important.

In assessing an AG's success, there must be a benchmark with which to compare. A natural benchmark would be other litigants. Thus: Hypothesis #3 — State attorneys general will win at higher rates than other, similarly-situated litigants. 57 This means that when AGs take a liberal (conservative) position, they ought to win more often than other litigants within the state who take liberal (conservative) positions. Also, when AGs are the petitioner (respondent) , they ought to win more than other petitioners (respondents). The preceding hypothesis should hold true in all states, because all state AGs are repeat players. However, not all state AGs are equal in their "repeat playerness", if you will. Although Galanter mentions numerous specific advantages, it appears to me that, for the purposes of analysis, each advantage can be considered to spring from one of two fundamental sources: 1) the sheer frequency of an AG's appearance in court, and 2) the control over litigation which the AG enjoys. I will briefly discuss each in turn. First, attorneys general's offices acquire expertise and other advantages as a function of the amount of times they appear in court. When one performs a task repeatedly, it becomes routine. As already discussed, the cases represented by state AGs' offices tend to be concentrated into a few areas of law. Among the advantages mentioned by Galanter that this ought to promote is expertise — they will know both the substance of the law and the procedural, technical, and jurisprudential points upon which cases often turn. This should also promote better knowledge of the justices' thinking. Also, the more frequently an AG's office appears in 58 court, the more likely it should be to form a "facilitative

institutional relationship" with the court. Hence: Hypothesis #4 — The higher percentage of a court's docket that consists of cases involving the AG's office, the higher the office's winning percentage ought to be. The second major source of a repeat player's advantages is their ability to select strong cases. An AG's office may appear frequently in litigation, but if it is not able to prevent weak appeals from proceeding, its expertise will be undercut. In their study of litigant success before state supreme courts, Wheeler et al (1987) find that the 'haves' tended to come out ahead, noting that "the advantage of larger government parties stemmed...from their apparently greater selectivity in choosing which cases to appeal" (1987, 419) .

Thus, the variation in the amount of discretion ought to explain, at least in part, variations in the rate of success experienced before a state supreme court. Hence, the hypothesis:

Hypothesis #5 — The more discretion over litigation which an AG's office possesses, the higher its success rate in court will be.

THE DUAL EXECUTIVE The extant research focuses on the office of the Solicitor General even though the concern is usually judicial reaction to the president. Although this may seem indirect, at the federal level it poses few analytical or conceptual problems: when the Solicitor General's office takes a position in court, we can safely assume that it is taking a position which is consistent with the desires of the president. To be sure, the Solicitor General operates with a substantial amount of independence, and only rarely consults with the attorney general or the president in decisions involving legal matters (Salokar 1992; Harmon 1988, 1091; Schwartz 1988, 1139; Wilkins 1988; Scigliano 1971, 163; Burt and Schloss 1968/69, 1444). However, the Solicitor General serves at the pleasure of the president. The incumbent is appointed by the president and will not continue in his position for long if frequently advocating legal positions which conflict with the administration's political agenda (Chamberlain 1987, 413; Witt 1986). Some (i.e. Neuborn 1988; Schwartz 1988; Caplan 1987) have even argued that the Solicitor General's office has recently become too eager to become a mere mouthpiece for a president's political agenda.

Nevertheless, the point remains: at the federal level, the position which a Solicitor General takes in court can adequately be utilized as a surrogate for the president's position. Even though this is the way we customarily think about the relationship between the chief executive and the government's chief legal officer, it is really the exception rather than the rule. In only five of the fifty U.S. states is the attorney general appointed by the governor.10 In 43 of the others, the AG is elected on a partisan ballot.11 This fact ought to have tremendous implications for the way in 60 which justices view the executive branch, because the executive branch may no longer be speaking with one voice. Instead, justices must consider the wishes of two legal actors, both of whom have their own independent bases of power and may even belong to different political parties. There is definitely a consensus among observers that the state AG is typically much more independent of the governor than the SG is of the president, and it seems that this increased independence is due in large part to the manner of selection. "Since the attorney general is usually an elected official, he sometimes is a political rival of the governor in some instances.... Under such circumstances, he may use his powers to argue appeals and bring original suits to a state supreme court to frustrate the governor's program" (Jacob 1984, 94; see also Boersema 1987; Palincsar 1979). Beyle ex­ presses a similar sentiment.

Some of our state governments' most interesting legal and political infighting has been between the governor as the chief executive officer of the state and the attorney general as the chief legal officer. It is clear that these two offices do have the potential for built-in conflicts at several levels, from politics to policy to administration. (Beyle 1983, 192). Not all of the AG's actions are designed to thwart a governor. Much of it is simply a result of differing role orientations. When governors appoint the AG, "the people" are represented in court only indirectly. However, when the AG is elected, the people are directly represented, and this fact seems to make AGs see themselves as an independent "people's advocate” rather than strictly as a legal mouthpiece for the administration. It appears that many AGs seem their most important duty is to be a representative of the public interest as they see it (see Thompson 1974) . Dave Frohnmeyer, former attorney general of Oregon, has written that "the attorney general has a recognized, overriding legal duty to protect the state and public interest. The interests of a specific agency or official may not coincide with the public interest or the interest of the state as a whole (1988, 92; see also NAAG 1974, 30). Most state supreme courts have agreed with this conception of the role of the attorney general, since most have conferred upon the state AG substantial common-law powers to contest governmental actions on behalf of the public. "Courts have expanded the role of the state Attorney General beyond representation of governmental entities to protection of the public's legal interests" (Ross 1990, 39; see also NAAG 1977, 2). There is some empirical evidence to suggest that the role orientation adopted by state AGs depends in part on the manner in which they are selected. William Thompson (1974) conducted a survey of current and former AGs in both elective and appointive states. While the majority of AGs in both types of states identified their most important role as being a repre­ sentative of the state government, elected AGs were more likely than appointed AGs to say that their most important role is serving as the peoples' attorney (26% vs. 17%) . 62 Although the difference is not spectacular, Thompson appropriately concludes:

The elected group is more likely to perceive their role as the people's attorney. Also, they show a greater willingness to supervise programs in the environmental and consumer protection areas. The opinions examined certainly offer support to the notion that the elected attorneys general show more independence in their behavior. (Thompson 1974, 41) . Hence, at the state level, the executive branch does not necessarily speak with one voice. The independent influence of the attorney general must be kept conceptually and analytically distinct. Hence: Hypothesis #6 — Justices who belong to the same party as the attorney general will support the AG more than justices of the opposite party.

Hypothesis #7 — The AG will win more frequently when the AG and governor belong to the same political party.

Finally, there is the case of Tennessee. The link between the state supreme court and the state attorney general ought to be closer in Tennessee than in any other state, because in Tennessee, the AG is selected by the supreme court itself. While no empirical studies have been conducted, this system ought to promote policy agreement between the AG and the court for the same reasons why gubernatorial appointees should be in policy agreement with the governor. The court has the opportunity to screen candidates for their ideological and jurisprudential views. Hence, this "minor" hypothesis is proposed:

Hypothesis #8 — The Tennessee attorney general should enjoy more success than any other state AG. The stage is now set. In these two chapters, I have argued that judges are goal-oriented actors whose behavior toward the executive branch on state supreme courts should be influenced by a number of factors, including ideology, the manner in which they were selected, and the powers brought to bear by the executive. The complications posed by the dual nature of the executive in most states were discussed. In the following chapter, I will turn to the task of describing the data and methodology to be utilized in testing the hypotheses that have been developed. 64

ENDNOTES

1. There are certain regulatory agencies, such as the Interstate Commerce Commission, that are not required by statute to seek the Solicitor General's approval nor utilize the services of the office. However, even these agencies usually desire the SG's support and will thus occasionally compromise their position to obtain it (Stern 1960, 157). 2. Interestingly, even though there is some variation, it is minor. Salokar even focuses on Erwin Griswold, who was the Solicitor General under two presidential administrations, one Democratic (Lyndon Johnson) and the other Republican (Richard Nixon). While the ideological content of the cases pursued by the SG became more conservative after the transition, Griswold's success rate did not fluctuate markedly (Salokar 1992). 3. Boersema (1987) attempts to link both formal AG powers and ambition with things such as the types of litigation pursued, the types of clients represented by the AG, types of advisory opinions issued, and amount of legislative lobbying performed by the AG. 4. Ironically, stronger AGs tended to have less success, primarily because they handled a smaller proportion of criminal cases, which AGs tended to win more than non-criminal cases (a finding confirmed by this research). 5. For example, he includes such factors as tenure potential, and powers to issue advisory opinions. Many of the factors included in his index are, however, relevant to the present study. These include the AG's powers over criminal prosecution, consumer protection powers, salary and staff, and control over administrative agencies. 6. Tennessee and Iowa are the exceptions. 7. For a comprehensive, 50-state survey of the powers, duties, and other aspects of the AG's office, see Ross (1990). 8. But in most states, the attorney general does not participate in oral argument except on rare occasions, whereas the U.S. Solicitor General often does. 65

9. Ideally, this question could be answered through a questionnaire administered to members of the various state courts of last resort, but such an undertaking is beyond the scope of this dissertation. 10. Alaska, Hawaii, New Hampshire, New Jersey, and Wyoming. Three of these states are included in this study. In addition, the AG is appointed by the executive in Washington D.C., American Samoa, Guam, and the Virgin Islands. 11. This leaves two states remaining. As will be discussed below, the Tennessee Supreme Court itself chooses the state AG, and is included in this study. In Maine, the legislature chooses the attorney general. As this is a study of judicial relations with the executive, Maine was not considered for inclusion. CHAPTER III THE RESEARCH DESIGN AND SETTING

As indicated in the last chapter, a multi-state comparative design will be utilized in order to adequately test the hypotheses developed. I decided that states would be selected for analysis non-randomly, in order to select states

that exhibited variation in the characteristics which have been surmised to be important — method of judicial selection, method of attorney general selection, and the amount of discretion and expertise possessed by the attorney general's office.

The states selected may be somewhat unrepresentative of states as a whole, as preference was given to those state courts on which dissent rates are relatively high (see Glick and Vines 1973, 79). This was done in order to maximize the number of non-unanimous decisions. Whereas unanimous decisions are often important and represent the majority of a

state supreme court's docket, there are well-known methodological difficulties in analyzing cases in which there are no differences of opinion expressed for the analyst to witness (Grossman 1966, 1555). However, although states with high dissent rates are somewhat overrepresented, a number of

66 states with low dissent rates have been included (see Table 1) . This consideration has resulted in the fact that "innovative" courts also appear to be overrepresented (Canon and Baum 1981).

TABLE 1 DISSENT RATES ON STATE SUPREME COURTS

1 . California 52.7% 2. Ohio 42.6 3. Missouri 40.2 4. Washington 29.9 5. New Jersey 25.4 6. Wyoming 23.1 7. Illinois 21.3 8. New York 21.1 9. Alaska 20. 3 10. Tennessee 12.7 11. Arizona 9.7 12 . North Dakota 8.6 LL DISSENT RATE 24 .9%

I decided that twelve states would be included for study. This number was arbitrary, but in conjunction with the decision to include five years (see below), this decision produced 60 state-years for analysis, a number sufficiently large to enable meaningful statistical analysis. Based on the preceding criteria, the twelve states which were selected were

Alaska, Arizona, California, Illinois, Missouri, New Jersey, New York, North Dakota, Ohio, Tennessee, Washington and Wyoming.

A more complete description of these states and their legal and political contexts follows later in this chapter. 68 For now, it will suffice to give a brief overview of how these states were classified on the theoretically important dimensions previously developed. As for the selection of the attorney general, states were classified into three types: those in which the AG is selected by the Supreme Court (Tennessee), those in which the AG is selected by the governor (Alaska, New Jersey, Wyoming), and those in which the AG is selected by a state-wide partisan election (all the remaining states).

There are five main methods of selecting judicial candidates in the U.S. states — legislative appointment, gubernatorial appointment, the Missouri or merit plan, partisan election, and non-partisan election. For my purposes, they have been reduced to two — election and gubernatorial appointment. Those states with legislative selection have been excluded from the study. Merit plan states are considered, for the purposes of this study, to be gubernatorial selection states. This is because virtually every study of the Missouri plan in operation has concluded that although the mechanics of the two systems are different, the governor is the main player in the judicial selection process in states utilizing the Missouri Plan (Bunch and Casey 1990; Champagne 1986; Watson 1986; Watson and Downing 1972) .

In each state that provides for "merit" selection, the governor is responsible for naming some of the members of the judicial nominating commission, and the selection of the 69 eventual justice is ultimately his or hers.1 As discussed in Chapter I, observers of the merit plan clearly believe that "merit" is not the only criterion upon which judicial selections hinge. A close observer of the Washington Supreme Court has written, "The governor considers legal and professional matters in making appointments, in much the same way as the Bar. The governor, however, consults with sources outside the profession, weighing factors that are not typically regarded as relevant by lawyers....Partisanship and political experience are recognized by the governor as an influential base within which more detailed professional criteria can be weighed" (Sheldon 1986, 117). If these observers are correct, if the plan has indeed "not...taken the politics out of judicial selection", then merit appointees should be subject to political influences similar to those affecting "straight" gubernatorial appointees.2

Partisan and non-partisan election states have similarly been collapsed into one category. Unquestionably, the dynamics underlying these elections are different (i.e. Squire and Smith 1988) . Both are similar, however, in the fact that they are extremely low-information elections (Dubois 1983) . In addition, the theoretically important dimension of the judicial selection system is the extent to which the governor influences that selection, and thus bifurcating the states along this dimension is appropriate. In elective states, 70 justices owe their jobs to the people; in gubernatorial and merit states, justices owe their jobs to the governor.3 Finally, the states were chosen to vary in the amount of power possessed by the attorney general's office. "Power” was measured by adapting Boersema's (1987) AG power index to fit the purposes of the current project.4 Table 2 summarizes information regarding AG and judicial

selection systems, and attorney general power in these twelve states. For each court, it was decided to code every case decided with a written opinion of one full page or longer handed down during the calendar years of 1979, 1981, 1983, 1985, and 1987. The decision to examine five years (rather than four or six) calendar years5 has no theoretical justification other than the fact that it provides enough cases for meaningful statistical cross-state comparative analysis. Odd-numbered years were chosen for this reason: in some of the intra-state analyses to be performed, changes in judicial behavior are hypothesized to be observed following a change in the cast of political actors — either the governor, the AG, or supreme court justices. Since most state-level elections are still held during even-numbered years, examining consecutive odd- numbered years should allow for a comparison of behavior before and after a substantively important change of incumbent status. 71

TABLE 2 SELECTION SYSTEMS AND AG POWER IN 12 STATES

STATE JUDICIAL AG BOERSEMA'S SELECTION SELECTION AG POWER RATING NJ gubernatorial appointed 95 AK merit appointed 39 WA elected elected 87 IL elected elected 85 OH elected elected 84 CA gubernatorial elected 82 NY elected elected 82 AZ merit elected 81 ND elected elected 79 TN merit Supreme Court 76 MO merit elected 70 WY merit appointed 68

# of states in which justices are appointed = 7 # of states in which justices are elected = 5

# of states in which AG is appointed by governor = # of states in which AG is elected = # of states in which AG is appointed by Court = 72 Only cases with a written opinion of one page or longer were included in this study. Courts will often be presented with a number of cases dealing with the same or similar legal issues. At times, courts will review a number of these cases simultaneously. Typically, they will hand down a full written opinion for the first case, and follow this by a number of per curiam opinions in the subsequent cases directing the reader to refer to the legal principle established in the previous cases.6 The one-page rule was an attempt to exclude these derivative types of cases, so as not to overrepresent one case outcome. The validity of a one-page rule may be questioned, since what constitutes "one" page differs from state to state. Cases for six of these state courts7 were coded directly from the case reporter published by that state government. Cases for the remaining states were coded from West's regional reporters, cross-checked with Shepard's citations. Since each of these sources uses different faces and sizes of type, and each structures their reporters differently, the length of a "one-page" decision may vary. However, very few opinions were as short as this, and in the case of any ambiguity, I was less inclusive rather than more so. Cases in which the state AG's office submitted amicus curiae briefs rather than representing a litigant are not differentiated and are combined with the other cases in this analysis. While there is a compelling argument to be made that amicus briefs and actual cases represent different 73 stimuli for the jurist (i.e. Segal 1989), their aggregation in this study should be unproblematic. First, including amicus cases is desirable, because as Segal notes, positions taken in amicus briefs are more likely to be pure statements of policy preference than positions taken in other cases, since the government does not have a direct interest at stake. Hence, their inclusion will increase the chance of discovering the expected relationships. Second, attorneys general virtually never participate as amici. In most states in most years, there is no amicus participation (see Table 3). The highest rate displayed in a single year in this study was 10 by the New Jersey AG in 1987. Thus, statistical analysis would not be possible considering amicus cases on their own. They were included in the analysis in order to increase the number of cases for those states in which the number of cases was small.

TABLE 3

NUMBER OF AMICUS BRIEFS FILED BY STATE AGs (ENTIRE STUDY PERIOD) New Jersey 37 California 15 Washington 10 Ohio 8 Alaska 7 Missouri 6 New York 4 North Dakota 3 Illinois 3 Arizona 2 Tennessee 1 Wyoming 1 74 The coding of much information from these cases was straightforward and unambiguous. For each case, I noted the issue area8, whether or not the AG's office was involved (either representing a litigant or appearing as a litigant), the decision that was reached, and the identity of any dissenters and non-participants. Further, where possible, I noted whether a case was heard under the court's original, mandatory, or discretionary jurisdiction.9 The complete set of coding rules is found in Appendix C. For now, three aspects of the coding procedure should be highlighted:

a) SIDE — If the AG's office was the litigant seeking high court review of an adverse decision, it was coded as the petitioner/appellant. Conversely, if it had won at the previous level, it was coded as the respondent/appellee. If the ruling at the previous level was mixed and both parties filed appeals, the AG's status was also coded as mixed.10 b) IDEOLOGY — As have many scholars, in order to determine the ideological outcome of a case, I have adopted

(with modifications) a coding scheme similar to that formulated by Stuart Nagel (1961). It is explained in detail in Appendix C. For now, it is sufficient to say that this is for the most part an "upper-dog" vs. "underdog" scheme (i.e. Ulmer 1979), in which the identity of the litigants rather than the issue area involved normally determines the coding decision. Generally, a vote for the litigant with the fewer resources is coded as a "liberal" vote. Examples would 75 include: a vote for the criminal defendant in criminal cases; a vote for the worker in workers compensation cases; a vote for labor in labor-management disputes; and a vote for the tenant in landlord-tenant disputes. While this scheme is relatively straightforward in most cases, I found it particularly vexing in taxation and eminent domain cases. The taxing and takings powers of government are usually viewed as instruments employed by liberals, and Nagel

(1961) coded all votes in support of the government in tax cases as liberal decisions. Keeping with an upperdog-underdog scheme, however, I have modified this. When a government was seeking taxes from a corporation or business entity, a decision for the government was coded as liberal. When a government was seeking taxes from a private individual, a decision for the government was coded as conservative.11 In numerous cases, such as divorce proceedings and will contests, there was no clear disparity in the resources brought to bear by either side. In these instances, ideology was coded as "neither".12 Further, there were many instance of business corporations suing one another. In some of these cases, it was apparent that one company was relatively small (often a sole proprietorship) while the other was relatively large. Rather than attempting to make an arbitrary determination in each case, however, all cases involving conflict between two business entities were coded as having no ideological content. Similarly, when different units or 76 levels of government were opposed in litigation, no ideology was coded. Coding for ideology is a task fraught with danger, including that of being overly simplistic or crude. However, it can yield valuable insights. As regards its crudity, James Leonard has written, tests for liberal or conservative bias presume a judge will favor certain parties over others. For example, a liberal judge should vote for a non­ commercial party against a commercial one in a debt collection dispute. Admittedly, the classifications are surrogates for more useful information. Other details, such as wealth and savvy, might indicate more precisely whether a party would evoke the sympathy of an ideologically minded judge. Likewise, it would be advantageous to know a business's size or an individual's income, social status, race, religion, or whether the litigation was conducted by a third party, such as an insurance company. Because the opinions do not yield this information easily, if at all, one must rely on cruder measures (Leonard 1989, 941).

There were some instances, however, in which the upperdog/underdog scheme was not strictly followed. The most frequent examples were those cases in which a business was challenging a governmental action or regulation. In the strict sense, the government would be classified the upperdog.

However, consistent with Nagel's scheme and the conventional wisdom, in these types of cases a decision for the government was coded as liberal.

Finally, while in most cases it was the identity of the litigants rather than the legal issue(s) raised that was determinative, there were certain instances in which following 77

this logic produced counterintuitive results. Libel cases were often of this type. In most libel cases, a private individual (the underdog) is pitted against a newspaper or other media outlet (the upperdog). However, since a decision in favor of the media outlet usually reguired extending protections of freedom of the press (a position which most

recognize as being "liberal"), such a decision was coded as liberal. Other exceptions are discussed in Appendix B. c) WINNING SIDE — In most instances, determining a winner was not problematic — the lower court decision was either affirmed or overturned in its entirety. Case outcomes were coded as "mixed" in two different situations: first, when both litigants were seeking high court review (implying an ambiguous decision at the lower level) and the court either affirmed or rejected the lower court's decision in its entirety (thus maintaining the ambiguity). Second, cases in which only one party sought review (implying no ambiguity at the lower level) that were "affirmed in part, reversed in part.13 Following the preceding general rules, 8,391 cases were coded, resulting in one of the largest datasets every utilized in a study of comparative judicial behavior. However, as I am attempting to explain variations in support for the AG among courts and justices in terms of the factors mentioned in Chapters I and II, gathering data on the individual justices, governors, AGs, and political structures in the states was 78 necessary. Information regarding structural arrangements, governors, and AGs was obtained from various editions of The Book of the States. Information regarding individual justices (their political party and manner of ascension to the bench) was obtained utilizing a variety of sources. I relied primarily in The American Bench and Who's Who in American Law, although information was also culled from a number of other published sources (.e. articles taken from newspapers, magazines, bar journals, and law reviews) . I was also the beneficiary of the efforts of two intrepid state supreme court law librarians.

The statistical procedures utilized for analyzing the data thus obtained will be discussed in Chapters IV, V, and VI, which are devoted to findings. The remainder of this chapter will be devoted to describing the courts of last resort and attorneys general's offices in the 12 states selected for inclusion in this study, in order that the findings can be placed into their proper contexts. In this section, I will present thumbnail sketches of the courts and executives in these twelve states, highlighting political factors and conflicts which may be salient to understanding the relationship between the three key sets of political actors — governors, AGs and justices.

ALASKA — Supreme Court justices in Alaska are selected by the governor through a variation on the Missouri Plan. Once appointed, they must run in a retention election in the 79 next general election. If retained, their term runs for 10 years, when another retention election must be held. Considering this long term of office, it is little surprise that the Alaska Supreme Court during the 1979-87 period was marked by a remarkable degree of stability. Between 1979 and 1985, there was only one membership change on the court, and two justices served on the court for the entire period under consideration. However, since Democrats held the governorship during the latter part of this period, Democratic representation on the court gradually increased, from 4-1 in 1979 and 1981, to 3-2 between 1983 and 1987. However, the split in partisan orientation did not lead to high levels of dissent (approximately 10% of all cases were decided non- unanimously). In spite of the state's small population and the court's dominance by Republicans, the state's supreme court is generally considered one of the more activist SSCs in the nation, and has been so since statehood. Peter Galie (1982/83) reports that in the area of criminal procedure rights, "The Alaska court adopted its own constitutional standards in the sixties before the retrenchments of the Burger Court in this area....Alaska from its inception as a state has taken its constitution seriously and adopted an activist and innovative stance in interpreting its provisions"

(1982/82, 261). The same observer argues in another article that the Alaska SC is one of the eight SSCs that have been 80 most active in the area of constitutional law (Galie 1986, 732). Perhaps their most (in)famous decision occurred in Ravin v. State14, in which they declared unconstitutional a statute prohibiting the possession of marijuana in the home. The Alaskan attorney general is one of five AGs who is appointed by the governor and confirmed by the state legislature. Hence, the relationship between the two ought to be more analogous to that found at the federal level than in most other states. It is only one of six state AG's offices that is created by simple statute. The AG's ties to the governor are thus stronger than that found in most other states, as evidenced by the fact that Alaska is one of only a half-dozen states in which the attorney general is a member of the cabinet (Ross 1990, 44). Its caseload before the court is relatively diverse, with only about 25% comprised of criminal cases. The Alaskan Supreme Court has broadly construed the power of the AG to defend the interests of the executive branch. In Public Defender Agency v. Superior Court15, the court stated that "When an act is committed to executive discretion, the exercise of that discretion within constitutional bounds is not subject to the control or review of the courts.... Both federal and state courts have...held that the attorney general cannot be controlled in either his decision of whether to proceed or in his disposition of the proceeding."16 81

ARIZONA — As in Alaska, supreme court justices in Arizona are selected by a variant of the Missouri Plan. They must face a retention election after serving on the bench for three years, and every ten years thereafter. In Arizona, the cast of characters under consideration remained remarkably stable during the study period. Three justices served on the court for the entire period. Democrat Bruce Babbitt was in control of the governorship for four of the five years under study, and yet the court remained in Republican control, 3-2, during the entire period. Further, one man, Republican Robert Corbin, was the attorney general for the entire span. An unnamed supreme court justice insists that "although politics plays a key role in the selection process, party affiliation has little influence on actual rulings

(Kleinschmidt and Feldman 1992, 16). Events make it clear that politics are indeed involved in the selection process. In late 1987, a highly visible and public clash occurred between the Republican governor, Evan Mecham, and the state's Democratic chief justice, Frank Gordon Jr. took place over the governor's perceived manipulation of the rules in order to appoint a conservative candidate (Lopez 1987; Asseo 1987). Unlike the unnamed supreme court justice quoted above, others are not so sanguine about the lack of partisanship displayed on the bench itself. Sallen (1987) argues that there are distinct liberal and conservative wings that happen to coin­ 82 cide with partisan divisions on the court. However, this view is not confirmed in the following chapter. The Arizona AG is elected to a four-year term. The independence which this arrangement can give to an attorney general is apparent in the case of Evan Mecham. During that governor's turbulent term, the Arizona Supreme Court authorized the AG to pursue criminal charges against Mecham, finding that the AG does not represent the governor if the latter is not performing his official duties.17 Overall, however, it appears as though the Arizona AG's office does not have a good deal of discretion over its litigation before the Arizona SC. It was the respondent in criminal cases over 90% of the time. It is one of only ten states in which the AG has no common law powers (Ross 1990,

38n). It also does not have the same amount of control over executive agency litigation as the Solicitor General. Ross reports that "the Supreme Court of Arizona acknowledged that the state's statutory scheme allows a client agency authority to decide, in some circumstances, not to accept the services of the Attorney General" (1990, 43).

CALIFORNIA — As one of the nation's most innovative courts in the nation's most populous state, more has been written about the California Supreme Court than another other state supreme court. The controversies involving Chief Justice Rose Bird and her Democratic colleagues are perhaps the starkest illustration of one of the basic premises 83 underlying this dissertation — that politics and partisan considerations may influence judicial behavior toward the executive branch. In California, justices are appointed by the governor and confirmed by a Commission on Judicial Appointments. They then face retention elections. The California SC was marked by a high level of stability during most of this period, dominated by liberal justices appointed by Democratic governor Jerry Brown and led by Chief Justice Bird. "The California justices for decades pursued an activist court boldly and openly" (Culver and Wold 1986). The court attracted a great deal of controversy and public attention by overturning death sentences in a high percentage of the capital cases that came before it. While it was capital punishment that riveted public attention, more attentive critics argued that "the Bird Court's decisions in other areas have shown no less disregard for constitutionalism and the rule of law" (Erler and Vincent 1986) .

Things came to a head in 1986. Brown's eventual successor, Republican , had made the Bird Court one of the cornerstones of his electoral campaign in 1982. Deukmejian's attitudes toward the court were well- informed, considering he was the attorney general at that time, defending a large number of criminal convictions and losing most of them (see Chapter IV). In 1986, Bird and two of her allies, and Joseph Grodin, faced a 84 retention election in which Deukmejian campaigned aggressively for their defeats. They ultimately were defeated and replaced by Republicans of Deukmejian's choosing. The state promptly began reaping the benefits, as documented in the following chapter. The connection between judicial politics and judicial policy is clear in California at least (see Emmert and Traut 1994; Egelko 1987; Craig and Norman 1986).

The California AG is elected. George Deukmejian held the post during the beginning of the period under study. Although the office maintains "complete control of agency litigation" (Ross 1990, 53), less than 10% of the cases which it argues before the California SC are non-criminal. There is clearly a tradition of independence in the California state AG's office, particularly considering that the governor and AG were of different political parties during the entire period under study. This tradition of independence is quite well typified by the following excerpt from an interview with incoming Democratic AG John van de Kamp, who assumed the office on the same day that Republican Deukmejian assumed the governorship:18 Q: Since he is from the opposite party, do you anticipate any problems with Governor Deukmejian?

A: I hope not. The top executives in my Department of Justice will carry on a very professional relationship with the administration....The areas of potential conflict lie in the independent power of the attorney general. If I have the power, I'm going to use the resources that I have to take the consumer unit in one direction, or the environmental unit in another, if it's necessary. The governor and I might handle those 85 resources differently; we might have different areas of priority. [Interview (no author) 1983, 36]

One suspects that consumer or environmental protection would not likely be the top priorities of a Republican governor.

ILLINOIS — The seven members of the Illinois Supreme Court are elected on a partisan ballot. Like the others, it is characterized by a relatively high amount of stability over the period under study: four of the justices served for the entire period. Galie (1982) names it as one of the eight SSCs that have been most active in the area of state constitutional law. Both parties were represented on the court during the period, and perhaps as a result, dissent occurred in approximately one-fourth of the cases decided with written opinion.

The Illinois governor has less influence on the process of judicial selection than governors in most other states. In most states which provide for the popular election of judges, governors have the power to appoint interim justices if a sitting justice dies or retires between elections. This has resulted in governors ultimately appointing many of the justices even in "elective" states (Herndon 1962). However, in Illinois, the power to make interim appointments lies not with the governor, but with the Supreme Court itself. This is one of the features that made this state an attractive candidate for inclusion (see note 1). The attorney general is elected to a four-year term. The caseload of the office is more diverse than most, with criminal cases comprising 57% and non-criminal cases comprising 43%. Furthermore, the Illinois AG appears as the petitioner a larger percentage of time than any other AG

included in this study. This suggests that the Illinois AG has a good deal of discretion in selecting cases to pursue. Further lending credence to this hypothesis is the fact that the Illinois AG is the only AG in the nation to have its supreme court acknowledge that the AG's common law powers cannot be abridged by either the legislature or the

judiciary.19 However, if the AG has utilized this extremely generous grant of power to actively oppose the governor, it has not been widely written about.

MISSOURI — Missouri was the first state to institute merit selection of its judges in 1940. As such, its experiences with the system have been studied more than other states. As indicated above and in Chapter I, the consensus has been that this innovation has not removed politics from the selection process. Selection battles have tended to be relatively controversial, and always dominated by the governor. According to popular accounts, this partisanship has translated into behavior on the bench. In 1982, the justices "began quarreling rancorously in public. Norms of judicial conduct which ordinarily restrain hostility among judicial colleagues have broken down....The court seems to split Republicans from Democrats, or more aptly, appointees of Republican administrations from appointees of Democratic 87 administrations"20 (Bunch and Casey 1990, 6). Perhaps evidence of this can be found in the fact that the court displayed one of the highest dissent rates of any court included in this study. The court has been very unstable,

with high membership turnover during the study period, and partisan control of the court has shifted hands. The Missouri attorney general is elected to a four-year term. Republicans, in the persons of John Ashcroft and William Webster, controlled the office during the entire period under study. Popular accounts reveal no major clashes between the AG and governor during these years, perhaps because Republicans also had a lock on the governor's mansion for the whole time. Criminal cases comprised approximately 75% of the cases handled by the AG's office before the court. The office was also the respondent in approximately 80% of all cases in which

it participated, suggesting a certain lack of workload control. NEW JERSEY — The New Jersey SC is consistently cited as one of the most innovative and activist in the nation. "Since World War II the New Jersey Court has assumed a role of leadership in the development of legal doctrine, thereby earning for itself a national reputation for activism and liberal reform" (Tarr and Porter 1988, 184). New Jersey is among Galie's (1982) eight most innovative state supreme courts (see also Fino 1987; Caldeira 1983; Canon and Baum 88 1981). This popular conception is borne out by information collected for this study — the court decided in a liberal direction in over half of all the cases it decided in these five years, the highest rate of any court with the exception of the Bird Court in California.

Justices are appointed by the governor with the advice and consent of the state senate. Democrats dominated the court for the entire period under study, In a display of bipartisanship similar to that sometimes shown in Missouri, Democratic Governor Brendon Byrne nominated two Republicans

during his tenure in office. Dissent rates, at about 20%, are somewhat higher than the average SSC, which would be unexpected due to the relatively homogeneous nature of the court. However the New Jersey SC has a high amount of discretion in accepting cases for review, suggesting that they have a greater ability to select "tough” cases. Turnover on the court was high during the period.

As in Alaska, the New Jersey AG is selected by the governor and is a member of the cabinet. The workload of the office is varied, with criminal cases comprising about 65% of the cases in which the office participated. The AG appears to realize the activist nature of the court he or she must confront — from term to term, the AG took a liberal stance in half of all non-criminal cases.

NEW YORK — As might be expected, there are numerous similarities in the political and structural contexts of the 89 New York Court of Appeals and the New Jersey Supreme Court. As in New Jersey, members of the NY Court of Appeals are appointed by the governor with the advice and consent of the state senate. Both courts are known as judicial innovators. Democrats controlled the governorship, the attorney general's office, and the court during the entire study period. Their domination of the court was thorough — Republican held only one seat on the court during the period,21 with the exception of 1983 during which they held two. As in New Jersey, this Democratic dominance has led to a relatively high rate of liberal decisions (about 45%) . As in New Jersey, partisan homogeneity does not necessarily lead to unanimous decisions. The court's relatively high level of dissent may be due to the largely discretionary nature of its docket. Unlike New Jersey, the attorney general is elected rather than appointed to a four-year term. Democrat Robert Abrams was the incumbent for the entire period. The political nature of the office was evident when Abrams used it as a power base upon which he built his 1992 campaign for the U.S. Senate. Despite this, Abrams mounted no major challenges to the state's governors during this period, a peace partially fueled by partisan compatibility between the two offices. Fewer than 20% of the office's caseload was comprised of criminal cases. As a result, Abrams defended the liberal position in roughly half of the cases in which the office participated. 90

NORTH DAKOTA — The five members of the North Dakota Supreme Court are elected in non-partisan elections. Given that virtually all incumbents win non-partisan judicial elections, it is not surprising that the court's membership was relatively stable during the period. In spite of the non­ partisan nature of the elections, I was able to determine the partisanship of all justices. The court has been closely divided along party lines, with Republicans dominating 3:2 until 1985, when Democrats (or appointees of Democratic administrations) regained control by an equal margin. The competitive division of the court appears not to have been reflected in its collective voting behavior — the dissent rate is the lowest of any court in this study. This paucity of dissent can be partially explained by the state's lack of an intermediate court of appeals and the court's largely mandatory docket. The North Dakota AG is elected to a four-year term. Criminal cases make up about half of the office's caseload, a caseload which is relatively light. The North Dakota AG's office appeared in a smaller percentage of cases decided with opinion by its state court of last resort than any other AG's office in this study. As its caseload is comprised largely of criminal cases and cases in which the AG represents the respondent, the office's control over its caseload appears small. 91 Neither the North Dakota Supreme Court nor the state itself is generally considered an innovator (i.e. Caldeira 1983) , and hence very little has been written about its politics in published accounts. However, in order to obtain a true picture of the relationship between executives and judiciaries in all its facets, states like these must be considered. OHIO — While not as nationally visible as the battle over the Bird Court in California, the Ohio Supreme Court had its own ideological struggle in which Republicans successfully wrested control of the judiciary from Democrats in 1986. There were, naturally, significant differences between the two events. First, the issues which motivated the public were vastly different. Whereas in California it was criminal justice cases that produced the most controversy, questions regarding the judicial temperament of Chief Justice Frank Celebrezze and generous judicial support for labor interests were the key topics in Ohio (see Tarr and Porter 1988) . Secondly, Governor Celeste of Ohio was not nearly as involved in the volatile judicial election campaign as was his West Coast counterpart.

Ohio's unique judicial selection system provides for a partisan primary election followed by a non-partisan general election. However, as suggested in the previous paragraph, this does not prevent partisan considerations from being salient features of judicial campaigns. During the study 92 period, the court was marked by extreme volatility of membership. In 1979, Democrats, led by Chief Justice Frank Celebrezze, dominated the court, 6-1. By 1987, virtually every justice had been replaced and Republicans enjoyed a

narrow partisan advantage. Conflicts on the court have been sustained and sometimes physical.22 With a dissent rate of over 40%, it is one of the most conflictual courts included in this study. However, observers tend to agree that the divisions do not run strictly along party lines. Leonard's (1989) statistical study of voting patterns on the court

attributed very little effect to partisanship (1989, 962). The attorney general of Ohio is elected for a four-year term. Democrat Anthony Celebrezze held the office for the entire study period. The links that can be established

between an AG's office and the state judiciary are particularly evident in this instance — AG Celebrezze is a cousin of Chief Justice Celebrezze, and for a brief period, the AG had a second cousin (James Celebrezze) sitting on the court as an associate justice as well. The vast majority of cases in which the Ohio AG participates are non-criminal, with the emphasis being on workers compensation cases. Ohio is one of the AG's office

that most consistently takes a liberal stance. Although the AG's office is intimately involved with precisely those cases identified by Tarr and Porter (1988) and others as causing the most friction on the court, few reports of conflicts between 93 the AG and the governor have appeared in published accounts. TENNESSEE — Tennessee Supreme Court justices are elected on a partisan ballot. In spite of this, the Tennessee SC displayed the lowest turnover rate of all states included in this study. Four members of the five-member court served for the entire period. This is because of Tennessee's rather unusual tradition of judicial campaigns. In Tennessee, Democratic justices run together, as a slate of candidates in opposition to the Republican slate. Recent Republican electoral gains in the South have not yet been reflected in balloting for the Tennessee Supreme Court, since the

Democratic slate has repeatedly won. In reality, it may be simply a severe case of the incumbency effect, since the sole incumbent Republican on the court (who was initially an interim appointment of a Republican governor) has consistently won re-election. However, the natural expectation that this brand of straight-party voting would exacerbate divisions on the court is not evident by the level of dissent.

As indicated in the previous chapter, Tennessee's AG is unique in that it is the only one in the United States selected by members of the state supreme court. "No other Attorney General is so closely related to the state supreme court" (Ross 1990, 57) . Little has been written about the

consequences of this unusual relationship on the AG's relations with the court, but one incumbent argues "this method of selection and relatively long term [8 years] tend to 94 depoliticize the office" (Cody 1987). I have argued that it ought to promote a congruence of interests and thus result in a high rate of support for the AG in court.

WASHINGTON — The nine justices on the Washington Supreme Court are elected on a non-partisan ballot for a six-year term. The court is generally considered activist (i.e. Galie

1982), despite its domination by Republicans during the period. Regarding partisanship, a very unusual pattern emerged — during the eight years of this study, with only one exception out of 16 justices, ALL of the Republicans were initially appointed to the bench, whereas ALL of the Democrats were initially elected. Republicans controlled the court during the entire period, although their control gradually increased from 5-4 in 1979, to 7-2 in 1987 through the appointments of Republican Governor John Spellman. Perhaps partially as a result of this partisan split, the dissent rate on the court was relatively high during this period. Schmidt (1990) examined voting patterns on the court, and although he

concluded that voting patterns tend to change drastically when the composition of the court changes (990, 13), he does not link that change to political party. The Washington AG's office is stronger than most. It is one of the few state AG's offices that "provide consolidated, centralized legal services to all or some executive agencies"

(Ross 1990, 52) . Reflecting the large amount of docket control possessed by the SSC, the AG's office represented the 95 petitioner in a fairly large proportion of cases (approximately 40%) . This power is enhanced by the fact of the office's electoral independence from the governor. Whereas the Washington governorship switched from Democratic to Republican control, and back to Democratic control, Republicans controlled the AG's office for the entire period. Very little has been written about it, however, so it is

assumed that conflicts between the AG and governor have not reached critical mass.

WYOMING — Members of the Wyoming Supreme Court are selected through a variation of the Missouri Plan and must face a retention election the following year. Judging by my research, judges in Wyoming do a very good job of concealing

their partisan affiliations. If, however, we use the party of the appointing governor as a surrogate, the court was dominated by Democrats during the entire study period, increasing their initial 3-2 advantage to 4-1 in 1987. Like its other rural counterpart, North Dakota, the Wyoming SC is not generally seen as innovative or highly regarded. In fact, Caldeira (1983) lists the Wyoming Supreme Court as having the lowest reputational score of all fifty state courts of last resort. A consequence of this lack of respect is that published accounts of the court's operations and jurisprudence are few. And while we might assume many similarities between the Wyoming and North Dakota courts, they

Wyoming court has a dissent rate that is substantially higher. 96 Like Alaska and New Jersey, Wyoming provides for gubernatorial appointment for the AG. Thus, Democrats controlled the office of AG for the entire period as a result of their control of the governorship during that time. There are indicators, however, that the Wyoming AG's office is not particularly strong. Unlike in those two other states, the Wyoming AG is not a member of the cabinet. Further, it is only one of six AGs offices that is created by statute rather than in the state constitution (Ross 1990, 40) . Criminal cases make up the vast majority of the AG's caseload, suggesting that the office has little opportunity to make policy before the state's highest court. Table 4, 5, and 6 present much of the preceding information in summary form. The desire to include courts with relatively high rates of dissent has resulted in an overrepresentation of courts with discretionary dockets, and states that are activist and respected — although the two are interrelated. I do not consider this to be a significant problem for two reasons. First, activist courts have historically attracted the attention of both scholars and observers. They simply provide more to observe and study. Second, these states are clearly not homogeneous. The states that have been selected are diverse on a number of salient dimensions. Kagen et al (1978) classify state supreme courts 97

TABLE 4 SELECTED FEATURES OF STATE SUPREME COURTS

STATE % OF DOCKET REPUTATIONAL DISCRETIONARY1 RANKING2 Alaska 56% 36 Arizona 97 19 California 97 1 Illinois 93 7 Missouri 97 23 New Jersey 95 3 New York 94 2 North Dakota 0 45 Ohio 96 21 Tennessee 91 37 Washington 96 8 Wyoming 0 50

1. Statistics taken from the National Center for State Courts' Annual Report on Caseload Statistics for 1987. 2. Ranking ranges from one to fifty, taken from Caldeira (1983) TABLE 5

POLITICAL CONTROL OF THREE STATE INSTITUTIONS

STATE COURT CONTROLLED PARTY OF PARTY OF CONFLICTS BY? AG GOVERNOR BETWEEN AG AND GOV?

Alaska Republicans D (1983-87) D (1983-87) no R (1979-81) R (1979-81)

Arizona Republicans Democrat R (1987) yes D (1979-85)

California R (1987) D (1983-87) R (1983-87) yes D (1979-85) R (1979-81) D (1979-81)

Illinois Democrats D (1983-87) Republican no R (1979-81)

Missouri Democrats Republican Republican no

New Jersey Democrats R (1983-87) R (1983-87) no D (1979-81) D (1979-81)

New York Democrats Democrat Democrat no

North Dakota Democrats (some data D (1985-87) no missing) R (1983-81) D (1979)

Ohio R (1987) Democrat D (1983-87) no D (1979-85) R (1979-81)

Tennessee Democrats Democrat D (1987) no R (1979-85)

Washington Republicans Republican D (1985-87) no

R (1981-83) D (1979)

Wyoming Democrats Democrat Democrat no TABLE 6A CRIMINAL CASES AS A PERCENTAGE OF ALL CASES HANDLED A STATE ATTORNEY GENERAL'S OFFICE

1 . Arizona 85.7% 2. California 76.8 3. Illinois 72.4 4. Wyoming 70.9 5. Missouri 65.7 6. Tennessee 55.0 7. Alaska 52. 1 8. New Jersey 48.5 9. North Dakota 15.2 10. New York 13.5 11. Washington 12.1 12. Ohio 4.4 OVERALL 55.6%

TABLE 6B PERCENTAGE OF CASES IN WHICH AG DEFENDED THE CONSERVATIVE POSITION

1 . Arizona 90.2 2. California 87.1 3 . Wyoming 85.2 4. Illinois 80. 3 5. Alaska 73.6 6. Missouri 71.8 7. New Jersey 67.7 8. Tennessee 65.1 9. North Dakota 54.4 10. Washington 51.0 11. New York 46. 6 12. Ohio 42.9

OVERALL 68.0% 100

TABLE 6C PERCENTAGE OF CASES IN WHICH AG REPRESENTED THE RESPONDENT

1 . Wyoming 90.1 2. Arizona 87.6 3. California 83.1 4. Missouri 82 .4 5. Alaska 76.3 6. North Dakota 72.8 7. Ohio 70.2 8. Tennessee 59.8 9. Washington 54.1 10. New York 51. 1 11. New Jersey 47.7 12. Illinois 45.1

OVERALL 67.8% 101 into three types,23 and this study includes examples of all three types.

It is time now to turn our attention to empirically testing the hypotheses developed. The next chapter represents the beginning of that task. 102

ENDNOTES

1. But, as Herndon (1962) and others have noted, in most elective states, governors are given the power to appoint interim justices when a sitting justice vacates the bench for some reason. 2. Clearly, the two types of appointment plans are more similar to each other than they are to either type of electoral plan, and vice versa. This decision to bifurcate the four systems is not compelled by theory, but it is justifiable, it will serve to reduce confusion, and will make for more manageable and parsimonious statistical analysis.

3. Supra note 2. However, in all honesty, the distinction between the types of system is quite blurry. In all those states in which the governor appoints justices, the justice must face the voters in a retention election within a year. In practice, these systems all work rather similarly, which may in large part explain the lack of success researchers have had in linking judicial selection system to other phenomena. 4. Boersema's power index included seven different dimensions. However, for the purposes of this dissertation, only three are directly relevant — 1) control over administrative agencies (which concerns discretion over litigation) 2) control over criminal prosecutions, and 3) staff and office resources (which concerns capability and expertise). In the multivariate analyses that follow in Chapters 4 and 5, the following two independent variables are utilized: a) POWER — this is #1 plus #3. This variable is utilized when attempting to explain OVERALL AG success and AG success in NON-CRIMEinal cases. b) CPOWER — this is #2. This variable is utilized in those models attempting to explain AG success in CRIME cases only. 5. Calendar years were utilized because judicial terms (i.e. when a justice takes the bench) usually coincides with the beginning of a calendar year — usually after a November election or when a justice has retired effective the end of the previous year. 103

6. Often two cases would be consolidated into a single opinion with no other opinions. These were also coded as only one case.

7. California, Illinois, New Jersey, New York, Washington, Ohio. 8. As this was not central to the work at hand, no formal categorization of cases was used. Appendices A and C contain more information about the various categories utilized. 9. This was done because I have argued that a litigant's success will depend in part upon the side (petitioner or respondent) the litigant takes, but the form of that influence will depend upon whether the case was mandatory or discretionary. Respondents should be more favored in mandatory cases than in discretionary ones (Leonard 1989) . However, in the analysis that follows, not much is made of this distinction. First of all, in the vast majority of states, the jurisdiction of the court is NOT revealed in the opinion, and attempt to assign each case a value would be prone to a good deal of error. Secondly, as Baum (1976) , Kagan et al (1987), and others have found, even on state courts of last resort that have more discretion, the respondent tends to win in a majority of cases anyway. 10. If two cases involving the attorney general resulting from the same fact situation were decided consecutively with opinions of longer than a page each, two cases were coded. 11. See caveat on the next page. 12. Another point of departure between Nagel's scheme and mine is the coding of divorce cases. His scheme called for coding as liberal all decisions in favor of the wife. I feel that social circumstances have changed sufficiently as to call the validity of this characterization into question, and hence I have coded cases involving family law — i.e. divorce, wills, estates, custody disputes — as inherently neutral in ideological content. In a study of voting on the Ohio Supreme Court, James Leonard (1989) makes the same choice. In fact, Leonard's scheme is quite similar to mine, although there are some variations. 13. In a few instances, however, cases that were formally "affirmed in part, reversed in part" were not coded as mixed. This occurred in complex cases deciding numerous legal issues. If it could be determined that one party prevailed on over 3/4 of the legal issues that were raised, they were coded as being the winner, most frequently this occurred in cases in which one side won the legal battle, but did not win the right to recover court costs and attorneys fees from the other 104 litigant. In cases in which there was any doubt about the "magnitude" of the relative losses of the two sides, the case was coded as "mixed". 14. 537 P.2d. 494 (Alaska 1975). 15. 534 P.2d. 947 (Alaska 1975). 16. Ibid at 950. 17. Mecham v. Superior Court of the State of Arizona ex rel Corbin No. CV87-0410-SA.

18. Van de Kamp specifically requested to be sworn in by Rose Bird, hoping to symbolize an end to the adversarial relationship that had arisen between the AG's office and the California Supreme Court (Interview 1983). This subtly suggests some of the dynamics which I discuss in previous chapters. 19. Fergus v. Russell 270 111. 304 (1915). Newbera. Inc. v. Illinois State Toll Highway Authority 98 111. 2d. 58 (1983) . 20. Republican Governor Kit Bond appointed two Democrats (Andrew Higgins and William Billings) to the court during his term.

21. And even this Republican, Sol Wachtler, was nominated by Chief Justiceship by Democratic Governor Mario Cuomo. 22. In November 1991, Justice Craig Wright of the Ohio Supreme Court attacked fellow Justice Andy Douglas in chambers, resulting in Douglas suffering two broken ribs. Wright had accused Douglas' secretary of tapping into his computer and leaking embarrassing information about Douglas to the press (Myers 1991).

23. Type I — small states with Supreme Courts having no discretion, no intermediate court, and relatively light caseloads. Type II — medium and large states with little or no discretion and heavy caseloads. Type III - medium or large states with a good deal of discretion and light caseloads. CHAPTER IV ATTORNEY GENERAL AGENDAS AND SUCCESS BEFORE COURTS COLLECTIVELY

In previous chapters, I suggested that state attorneys general, by virtue of their independent electoral power base and their formal powers to protect the public interest, may seek to implement their policy goals through litigation.1

Writings on the attorney general's office tend to highlight its political nature and its tendency to be a stepping stone to higher statewide office (Morris 1987, 299; Jordan 1985; Abraham and Benedetti 1969; Krislov 1959). One question that is not generally addressed, however, is whether or not AGs

actually do use the resources of the office to pursue a political agenda before their state's courts. If we are to assume that justices are, in part, political creatures responding to political stimuli, it would be useful to know whether or not an AG's legal agenda is motivated by politics. In other words, do the types of cases brought by one attorney general resemble those brought by other attorneys general in the same state, or does the agenda of the office change according to the identity of the officeholder? Motivation is a difficult thing for the analyst to establish. However, as a test of political motivation, it

105 106 would be logical to expect that a greater proportion of a Democratic attorney general's caseload would consist of liberal cases than would the caseload of a Republican AG from the same state. There is some evidence that this occurs at the federal level. Looking only at individual rights cases,

Salokar found that between 1961 and 1986, Democratic presidential administrations were significantly more likely to submit "liberal" amicus briefs and to argue liberal positions than were Republican administrations (1992, 167, 162). Erwin Griswold, the one Solicitor General during this period who served both a Democratic and a Republican administration, took

liberal positions more frequently under Johnson than under Nixon (Salokar 1992, 171). More recently, the Solicitor General's office has come under some criticism for too actively using its resources to pursue the Reagan and Bush conservative political agendas (Neuborn 1988; Schwartz 1988;

Caplan 1987) — although others (i.e. Chamberlain 1987) dispute whether the office can ever become "too" political. Whatever the outcome of the normative debate, the notion that SGs pursue political agendas through litigation has received substantial support. We do not know if state AGs' offices similarly pursue caseloads whose characteristics reflect the policy preferences of the officeholder. If they do, the expectation, supported by Salokar and the common wisdom, is that Democratic AGs ought to pursue more liberal cases than Republican AGs. Making comparisons across courts 107 would be a misleading indicator, since the various AGs have substantially different areas of responsibility. Hence, to test for policy change, attorneys general of different parties will be compared within single states. Only 4 of the 12 states in this study can be included in this analysis, however, because in the other states the partisan affiliation

of the attorney general remained constant throughout the study period. Table 7 does not provide a good deal of support for the notion that a change in the partisanship of the attorney general produces changes in the ideological content of the office's caseload. In three of these four states, the

Democratic AGs pursued more conservative agendas than their Republican counterparts. This is contrary to expectations, although none of these differences is statistically significant. The one significant difference which does emerge (in Alaska) is in the predicted direction. However, in order to more accurately gauge an AG's policy agenda, we perhaps ought to exclude criminal cases. The AG represents the respondent in the vast majority ofthose cases, and hence the AG is not able to proactively pursue an agenda of any sort in those cases. Further, in criminal cases, the state inherently represents the litigant taking the conservative position. Hence, there is a certain lack of choice in criminal cases which make them a less-than-ideal 108

TABLE 7 a PERCENTAGE OF CASES IN WHICH AG TOOK A CONSERVATIVE POSITION (ALL CASES)

STATE AG'S PARTY % CONSERVi

ALASKA Democrat 72.7% Republican 86.3% (x2 = 8.14, p < .01 with one d.f.)

CALIFORNIA Democrat 93.6% Republican 92.3% (x2 = 0.16)

ILLINOIS Democrat 85.9% Republican 79. 5% (x2 = 2.18)

NEW JERSEY Democrat 76.2% Republican 73.6% (x2 = 0.18)

TABLE 7b PERCANTAGE OF CASES IN WHICH AG TOOK A CONSERVATIVE POSITION (NON-CRIMINAL ONLY)

ALASKA Democrat 61.4% Republican 48.1% (x2 = 2.16)

CALIFORNIA Democrat 64.0% Republican 62 .5% (x2 = 0 .0 1 ) ILLINOIS Democrat 34 .4% Republican 31.9% (x2 = 0.05)

NEW JERSEY Democrat 50.0% Republican 40.4% (x2 = 1.22) 109 vehicle for political litigation. The bottom of Table 7 presents similar data for non-criminal cases only. Once again, there are no statistically significant differences. Alaska is again the state in which a difference is most pronounced, but this time the relationship is not as expected — the Democratic AG was more conservative than the Republican AG.

Attorneys general simply appear to have fewer opportunities than their federal level counterparts to

undertake a widespread, systematic change in their state's legal agenda. First, they are hampered by the fact that state courts of last resort are required to hear some or all of the cases which seek their review (National Center for State Courts 1987). This means that the state attorney general's

office must utilize a greater proportion of its resources

defending the state against these types of lawsuits, and fewer of its resources aggressively pursuing a legal agenda. Second, as discussed in the previous chapter, state AGs does not possess as much discretion regarding the decision to appeal a case as their federal counterpart (Jordan 1985). In numerous instances, they are constitutionally or statutorily required to press forward with a case and thus cannot pick and choose those that best fulfill any personal agenda the AG might have.

Third, it is possible that the nature of judicial federalism makes it more difficult for the state AG to be as 110 pure a political actor as the SG when engaged in litigation. As discussed previously, many observers feel that state courts are becoming more activist and formulating judicial policies in novel areas of law based upon innovative readings of state constitutions (i.e. Glick 1993; Fino 1987; Friedelbaum 1982). However, one need only glance at the reporters of various SSCs to discover that the cases on their dockets are not as varied or controversial as those of the U.S. Supreme Court. On these courts during the period of the study, pathbreaking cases were indeed decided, but they were clearly the exception rather than the rule. This era saw the transition from the Burger Court to the Rehnquist Court, yet for much of the period there were many areas of controversial law (i.e. abortion, school prayer) which were virtually pre-empted by the national court. Thus, state courts were a less inviting forum for this type of litigation.

Those points having been made, it cannot be fairly concluded that the identity of the incumbent AG is of no significance for the content of state litigation. As a practical matter, the number of observations upon which this generalization is made is far too small to close the book on the subject. Further, all cases are not fungible — while changes in the aggregate may appear slight, a change in the stance of the AG on a few issues of major importance may signal to the justices (and others) that a new regime is in town. Finally, changes that occur between AGs may be more Ill subtle. For example, even if AGs face essentially similar caseloads, those cases may be more or less aggressively and competently prepared and argued depending upon the incumbent. Naturally, the last two points are merely speculative. However, even if we assume that most AGs do not actively pursue an agenda through high court litigation, it does not negate the central arguments behind this dissertation. The argument regarding the repeat player status of the AG does not depend on the ideological content of the litigation involved - - the greater the resources of the office, the more it ought to win, regardless of positions taken. Likewise, the other substantively important variables (AG and judicial selection systems) are expected to influence judicial behavior regardless of the ideological position taken by the AG. For example, it was argued that appointed justices ought to support the executive branch more than elected justices due to factors such as differing role orientations. These arguments remain in force even if there are not major aggregate shifts in the stances taken by AGs from year to year. However, political party is a consideration which looms large in conditioning many of the hypotheses developed. As an example, I hypothesized that justices should support an AG of their own party at a higher rate than an AG of the opposite party. Should this expectation hold true if the litigation conducted by state AGs does not vary significant from incumbent to incumbent, if AGs do not appear to be overly 112 partisan in the conduct of their duties? Indeed, it may not. One of the goals of this dissertation is to determine whether or not these sorts of relationships exist at the state level, and a negative finding would still be a finding nonetheless. However, I will let the expectation stand. First, as indicated, the evidence presented here is far too sketchy to assert conclusively that AGs offices do not pursue distinct

legal agendas that change in some way from incumbent to incumbent. More importantly, partisan identification has often been described as a perceptual filter through which individuals view the political world (i.e. Campbell et al 1960). If this is true, justices will view an AG of their own party more favorably than one of the opposite party, even if they are bringing similar cases and arguments before the court.

Finally, although documenting partisan distinctiveness (or a lack thereof) will be an important contribution of the project, it is not its main purpose. The decision to include partisanship as a variable was essentially an attempt to condition the effects of the other structural variables with which I am most interested — judicial and AG selection system. I had hypothesized, for example, that appointed justices would be more supportive of the executive branch, and that even among appointed justices, partisanship should matter. If the first expectation turns out to receive support 113 and the latter does not, I will be satisfied because the former is more central to my thesis than the latter. THE STATE ATTORNEY GENERAL AS SUCCESSFUL LITIGANT In any case, let us now turn to an initial look at the data regarding AG win rates. When considering decisions on the merits, the least demanding criterion we could apply when defining a "successful" litigant would be whether or not that litigant had a "winning season" — that is, whether it won more than half of the cases in which it participated. Table

8 reveals that by this standard, state attorneys general are winners. In 53 of these 60 state-years, litigants represented by the attorney general's office won at least half of their cases.2

Overall, these state AGs offices won 67.3% of all cases in which they participated. Perhaps it is only coincidence, but Salokar reports that the U.S. Solicitor General won almost exactly the same percentage — 67.6% of decisions on the merits between 1959 and 1989 (1992, 29). When we look at the cumulative totals, we note that

Hypothesis #8 does not receive immediate support — the Tennessee attorney general appears to be slightly less successful than average. The variance in the overall totals is small, as most states cluster near the mean. Wyoming and California clearly stand out as the deviant high and low cases, respectively. However, the relative uniformity of the aggregate totals obscures the often dramatic variations 114 between states and within states across time. Hence, in the bi- and multivariate analyses to follow, individual scores (n

= 60) rather than cumulative scores (n = 1 2 ) will be used.3 As discussed in Chapter I, judicial decisions may be

made on different dimensions depending on the issue(s) involved in the case. In each state, there are too few cases of any one type to conceive of issue dimensions that would be both precise and lend themselves to meaningful quantitative analysis. Furthermore, comparing across states would be difficult because the caseloads of the AGs offices vary widely. However, in a relatively blunt attempt to control for issue dimension, I have bifurcated the cases into criminal and non-criminal for separate and comparative analysis. Table 9 reveals that the overall patterns are indeed different for the two types of cases. (In this instance, the

overall patterns may be more revealing, since in more than half of these states — i.e. Arizona, California, Missouri, New York, Ohio, Washington, and Wyoming — the AG does not participate in enough of one type of case or the other to make comparisons in individual years meaningful.) First, in 9 of the 12 states, the attorney general won more criminal than non-criminal cases. However, the differences are only statistically significant (at the p < .10 level) in three of those states. Overall, AGs won 64.2% of their criminal cases, and 57.8% of their non-criminal cases. 115

TABLE 8 SUPREME COURT SUPPORT FOR THE ATTORNEY GENERAL (PERCENTAGES)

1987 1985 1983 1981 1979

Alaska 58.6% 47.4% 65.7% 64. 3% 76. 1% Arizona 74.3 79.4 56.1 62.5 6 6 . 0 California 55.4 45.0 41.0 56.2 45.8 Illinois 63 .4 61.0 61.9 55.4 67.1 Missouri 79.5 69.8 70.7 81.9 59. 6 New Jersey 52.6 69.2 75.0 71.3 71.2 New York 58.3 81.5 41.1 72.4 58. 1 North Dakota 62.5 65.5 69.4 75. 0 45.2 Ohio 53.1 67.2 68.4 66.2 78.4 Tennessee 77.6 66.7 67.8 61. 5 50.0 Washington 70.0 32.7 63.1 62.9 63.6 Wyoming 73.6 8 8 . 5 83.3 88.0 75. 6

TOTALS Wyoming 81.3% Missouri 71.6 Ohio 67.1 New Jersey 67.0 Alaska 66.8 Arizona 66.5 Tennessee 64.8 North Dakota 64.3 New York 62.4 Illinois 61.4 Washington 59.2 California 47.1 ALL STATES 65.6% 116

TABLE 9 AG WIN RATES, CRIMINAL VS. NON-CRIMINAL (PERCENTAGES)

CRIMINAL WIN RATE NON-CRIMINAL WIN RATE 1 ) OHIO 92.3 NEW JERSEY 67.8 2 ) WYOMING 86.0 TENNESSEE 67.1 3) NORTH DAKOTA 83 . 3 WYOMING 63.2 4) MISSSOURI 73 .9 OHIO 62.6 5) ALASKA 67.2 MISSOURI 61.9 6 ) ARIZONA 61.9 N. DAKOTA 61.6 7) ILLINOIS 61.2 NEW YORK 57.4 8 ) NEW YORK 61.1 WASHINGTON 57.2 WASHINGTON 61. 1 ALASKA 53.8 1 0 ) NEW JERSEY 58.8 ILLINOIS 42.9 1 1 ) TENNESSEE 53.3 CALIFORNIA 39.7 1 2 ) CALIFORNIA 39.2 ARIZONA 33 . 3

DIFFERENCE BETWEEN CRIMINAL AND NON-CRIMINAL WIN RATES

OHIO +29.7 ARIZONA +28.6** WYOMING +2 2 .8 *** NORTH DAKOTA +21.7 ILLINOIS +18.3* ALASKA +13.4 MISSOURI + 12.0 WASHINGTON + 3.9 NEW YORK + 3.7 CALIFORNIA - 0.5 NEW JERSEY -11.0 TENNESSEE -13.8

OVERALL AG CRIMINAL WIN RATE 64.2% OVERALL AG NON-CRIMINAL WIN RATE 57.8% DIFFERENCE 6.4%* (X2 = 3.16 with 1 d.f.)

*** p < .01 ** p < .05 * p < .10 117 Second, the correlation between criminal and non-criminal win rates is very low (r = .120). Success in one type of case does not initially appear to be strongly associated with success in the other. The greatest differentials in rank are New Jersey and Tennessee, who are among the most successful in criminal litigation and least successful in non-criminal litigation. The situation is reversed for the Arizona and

Wyoming AGs. Regarding the California AG's office, we can largely attribute its abysmal overall showing to the fact that it handles far more criminal than non-criminal appeals, and that most of this study took place during the era of Rose Bird. Interestingly enough, however, the California Supreme

Court was no more predisposed to support the AG in these less visible, non-criminal cases. The AG's frustration in court was across the board. However, these findings may be misleading if taken out of context: an AG who frequently takes liberal positions would have less success before a conservative court than before a liberal court. Thus, in order to accurately gauge whether an

AG has won at a "high” rate, we must have a benchmark with which to compare. As discussed in Chapter II, that benchmark is the success of other, similarly-situated litigants. I have argued that, due to the office's status as a repeat player and the resources it may bring to bear as a result thereof, an AG should win more than other litigants taking a similar ideological stance or being in the same adjudicative position 118 (petitioner or respondent). The argument is that, even if justices are not predisposed toward normally supporting a liberal argument or appeals from petitioners, the familiarity of the AG's office and the expertise and skills the office possesses should change minds in some of these cases. Table 10 addresses this hypothesis. The numbers in the table are the differences in winning percentages between the AG and all other litigants in that year who fell into that particular category. For example, the number +22.9 in the first cell is derived by taking the percentage of cases that the Alaskan AG's office won in 1987 when it took a conservative position (68.4%) and subtracting the percentage of cases won by all other conservative litigants in cases in which the AG was not involved (45.5%). (68.4 minus 45.5 = 22.9). Positive numbers thus signify an advantage for the attorney general.4 From this table, it does appear that the AG is more successful than other litigants in court. Of the 240 cells in this table, 202 (84.2%) are in the predicted direction. This number alone provides strong support for the hypothesis.

However, in many of these years the AG did not participate in enough cases of one type or another to make comparisons meaningful. Thus, we ought to look more closely at those relationships which are statistically significant. Here, too, the results give strong support to the hypothesis: Fifty nine 119

TABLE 10 AG ADVANTAGE OVER OTHER LITIGANTS

STATE YEAR CONSERV LIBERPETIT RESPON

Alaska 1987 +22.9 +5.5 + 1.8 +14.4 1985 -35.8** +42.6* + 16.4 -12 .4 1983 +2 1 .0 * +7.2 +6.7 +2 0 .6 * 1981 +23.6** +12.3 +20.2 + 10.2 1979 +36.3****+30.4* +20.4 +27.5****

Arizona 1987 +31.2** +22.4 +22.3 +23.2* 1985 +51.4**** N/A -46.8 +59.8 **** 1983 +30.3** +25.0 +5.8 +13.0** 1981 +24.0* -25.0 -31.9 +40.3*** 1979 +5.2 +36.5 +13.6 +37 . 6 ****

California 1987 -30.7 +50.0 +11.9 +18.8 1985 +10.3 +6.8 -18.6 +3.9 1983 + 15.1 +23.7 +12.9 +4.0 1981 +31.5** +24.1 -5.0 +2.15* 1979 +25.3*** -28.9 +3.7 +4.5

Illinois 1987 + 8.0 +18.1 +20.4* +7.4 1985 +23.9* -4.5 +27.5*** +8.0 1983 +7.1 +9.1 +38.1**** -10.9 1981 + 1.0 +5.0 +28.7**** -23.2** 1979 +28.0**** -11.7 +24.6*** +5.1

Missouri 1987 +28.9** +12.2 +29. 2 +34.9*** 1985 +35.0*** +4.9 +6.0 +24.6** 1983 +41.1**** even -3.8 +26.7*** 1981 +31.7** +36.2 +22.6 +38.6*** 1979 +6.5 -6.5 +23.4 +2.3 New Jersey 1987 +3.1 +0.7 -6.5 +12.9 1985 +5.4 +30.3** +22.0 +24.7* 1983 +52.1**** +11.3 +28.2* +25.8* 1981 +27.6* +13.3 +30.6** +27.9 1979 +33.4*** +25.0 +26.3*** +18.1

New York 1987 -23.1 +31.7 +7.4 +9.7 1985 +4 4 .7 *** +24.1** +37.1*** +23.7 1983 -1.5 -15.2 +37.3* +23.7 1981 +2 1 .8 * +26.0* +22.6 +18.9 1979 + 1.1 +9.9 + 1.1 +15.6

North Dakota 1987 +7.2 +23.1 +17.5 + 10.2 1985 + 10.6 +11.8 +24.3 -4.6 1983 + 11.0 +53.6 * * * * + 2 1 .3 +4.9 120

Table 10 - cont'd

1981 +22.9 +42.8** +17.7 + 10.0 1979 -44.6** +19.1 -23.3 -12.4 Ohio 1987 -19.5* +22.9* -9.8 +8.4 1985 +2.4 +24.9** +9.1 +19.4** 1983 +29.9 +8.7 +18.3 +16.3 1981 +30.4*** +15.5 -14.2 +30.2*** 1979 +12.9 +43.2****+33.9*** +19.2*** Tennessee 1987 +14.6 +52.4****+27.2* +30.1*** 1985 +16.2 +3.8 +24.2 +11.3 1983 +6.2 +45.8*** +13.2 +26.8 1981 +17.5 +1.3 even +26.5* 1979 -4.5 +14.0 +3.9 -9.4

Washington 1987 - 10.1 +67.2**** +3.2 +32.5** 1985 -13.7 -31.8** -6.7 -30.3** 1983 -0.5 +36.2** +15.7 +11.5 1981 -18.3 +48.7*** +6.5 +15.7 1979 +28.2* +2.0 +7.3 +22.3

Wyoming 1987 +17.7** +14.3 +13.4 + 11.2 1985 +36.9**** +32.5** +36.9 +21.5**** 1983 +19.7* +42.7* +18.9 + 11.1 1981 +35.9**** +42.0 +41.7 +16.1** 1979 +27.4*** +11.7 +50.6*** +11.9

**** = p < .001 *** p < .01 ** p < .05 * p < . 10 (Significance levels derived utlizing a chi-square test of significance with one degree of freedom) 121 (92.2%) of the 64 cells that reach the conventional significance level of p < .05 are in the predicted direction. The striking result about these findings is that they appear to a greater or lesser degree in every state. As discussed in Chapter III, these states vary in many respects, including the partisan composition of their benches and the types of litigation handled by the AG, and yet despite these differences, the AGs share a common advantage. Even the California AG had an advantage over other conservative litigants in most of these years — an astonishing fact when we consider that the office's win rate in criminal (and hence conservative) cases was abysmally low during this period. There are other ways that we may view the data. As discussed previously, the attorney general's office has little or no input in the decision to litigate when representing the respondent. If an AG is to pursue a distinct legal agenda, it will have to be accomplished primarily through those cases over which the office has discretion. Thus, it may be appropriate to look at these cases as purer expressions of policy preference, both in the stances taken by AGs and the votes cast by justices. If the AG's office wins a large percentage of the cases in which it is representing the appellee, it is likely due to a large number of non- meritorious appeals taken by opponents, particularly before state courts with mostly mandatory jurisdictions (see Boersema 1987). 122 Again, there is a very low correlation between success rates in the two categories of cases (r = -0.036). As we might expect, AGs tend to win more as appellees than as appellants. However, this is not true for California, Illinois, or New Jersey (which are all courts with substantial discretionary jurisdiction). Indeed, whereas the California AG's office has scored low on every measure up to this point, it is rather successful when initiating appeals. If one accepts the argument that such cases are a better indicator of a deliberate legal agenda, the California AG's office might be gratified by its success. Unfortunately, such cases make up an extremely small percentage of its caseload, and thus its overall success rate remains dismal. Research conducted on the U.S. Supreme Court suggests that the court often pursues an "error-correction" strategy, i.e. the court is more likely to grant certiorari to cases in which it disagreed with the outcome in the lower court (i.e. Armstrong and Johnson 1982) . If this holds true at the state level, the AG ought to win more often as a petitioner in those courts with a larger discretionary docket. However, support for this notion is not found in Table 11. Courts with high levels of docket control — California, New Jersey, Illinois, Washington, Ohio — give both relatively high and low levels of support for the AG as petitioner.

Liberal vs. Conservative — Yet another way to view the data is to compare how AGs fare when they take liberal stances 123 TABLE 11 AG WIN RATES, PETITIONER VS. RESPONDENT (PERCENTAGES)

AG PETITIONER WIN RATES AG RESPONDENT RATES

1) NEW JERSEY 71.9% WYOMING 83 2 ) ILLINOIS 68.3 OHIO 74 ,2 3) CALIFORNIA 63.3 MISSOURI 73 ,9 ALASKA 63.3 NORTH DAKOTA 71 ,7 5) WYOMING 63.2 ALASKA 68 ,7 6 ) NEW YORK 61.6 NEW YORK 67 , 6 7) TENNESSEE 61.2 TENNESSEE 67 ,3 8 ) MISSOURI 61.0 ARIZONA 66 .7 9) ARIZONA 56.9 WASHINGTON 65 3 1 0 ) WASHINGTON 52.1 NEW JERSEY 65 ,2 1 1 )OHIO 50.6 ILLINOIS 54 4 1 2 ) NORTH DAKOTA 43 .8 CALIFORNIA 43

RESPONDENTADVANTAGE

NORTH DAKOTA +27.9*** OHIO +23.6*** WYOMING +2 0 .2 *** WASHINGTON +13.2* MISSOURI +12.9 ARIZONA + 9.8 TENNESSEE + 6.1 NEW YORK + 6.0 ALASKA + 5.4 NEW JERSEY - 6.7 ILLINOIS -13.9** CALIFORNIA -19.5***

Overall AG win rate as Respondent 67.9 Overall AG win rate as Petitioner 61.7 Difference 6 .2 *** (x2 = 10.36)

*** p < .01 ** p < .05 * p < .10 124 as opposed to conservative stances. Naturally, there is no a priori expectation which would generalize from state to state, since some state courts are dominated by Democrats and others by Republicans. Table 12 reveals that the general pattern is for AGs to win more frequently with conservative positions than with liberal positions. However, once again the differences in most states are not statistically significant, nor overall. Further, there is a good deal of overlap between success as conservative and success in criminal cases.

The patterns are not easy to explain at first glance. Aggregate partisanship on a court does not seem to do it. We might expect courts controlled by Democrats to give more support to liberal positions. But while some courts dominated by Democrats during this period — New Jersey, Tennessee, and California — fulfill this expectation, others — New York, Illinois — did not. Conversely, the Washington Supreme Court was controlled by Republicans during the entire period, and yet was more receptive to liberal appeals than conservative ones. Nor can the general Mactivism" of a court seemingly explain support for liberal AG positions. Current expressions of judicial activism on state high courts are usually associated with policies in accordance with liberal ideology (see Fino 1987) . It is true that three of the four most liberal courts in this study are among those that Galie (1982) identifies as among the most activist. But three other courts he also rates highly on this dimension — Illinois, New York, 125

TABLE 12 PARAMETERESTIMATES FOR MODEL OF INDIVIDUALSUPPORT FOR AO

DV =OVERALL DV = CRIME DV=•NONCRIME

b t b t b t

Constant 55.88 2.96** 84.24 2.59** 113.74 2.70** RESPOND 0.07 0.58 0.05 0.28 -0.35 -1.65 PARTICl 0.37 2.13* 0.41 1.84* 0.35 1.32 AGADVAN 1.68 1.54 1.49 1.09 1.72 1.09 POWER -0.13 -0.26 -0.61 -1.89 -0.40 -0.96 CONSERV -0.19 -1.17 (n. a • ) -0.18 -0.75 AGGOV 7.61 1.93* 14.11 2.68** 8.46 1.45 1987 -0.39 -0.08 3.75 0.63 -6.69 -0.92 1985 -1.49 -0.29 5.36 0.82 -9.56 -1.23 1983 -3.65 -0.72 4.86 0.74 -6.68 -0.84 1981 0.30 0.06 7.84 1.20 -13.13 -1.63

r2 = .208 r2 = .323 r2 = .214 n = 60 n = 44 n = 57

*** p < . 001 ** p < . 01 * p < .05 and Washington — are among the least liberal courts included in this study. However, we must keep in mind that there are other factors which have not yet been taken into account.

In the overwhelming majority of these cases, the attorney general's office is merely providing legal counsel to other agents of state government. However, in a few cases, the incumbent AG is actually named as a party to the litigation, as is the governor. This occurs infrequently enough that meaningful comparison is not possible in individual states or across states within individual years. But data tabulated by the author reveals that of the 40 cases in which a state AG was named as a litigant, they won 26, lost 10, with 4 cases being mixed, for a winning percentage of 70%. (As a point of comparison, governors won 60% of the cases in which they were 126 named as litigants.) Thus, AGs win cases in which they are personally involved at higher rates than they win cases for other clients. This difference is statistically significant, but this is chiefly due to the large sample size. I will return to separately analyze these cases in Chapter VI.

I have suggested that the advantages possessed by repeat players are generally derived from two sources. The first source is expertise — by merely appearing repeatedly in similar types of litigation, the AG's office gains expertise in certain fields of law and becomes knowledgeable of the thinking of individual justices. Expertise is obviously an ill-defined concept, and thus as an objective indicator I have proposed using an Ag's rate of participation before a court, suggesting that increased participation leads to increased expertise (as well as credibility and a number of other advantages of the repeat player). Thus, I hypothesized that the level of AG participation before a court ought to be positively related correlated with AG success rates — the larger percentage of a court's docket in which the AG's office participates, the greater success it ought to have in court. The correlation between overall win rates and rate of participation is relatively low (r = .241). California again appears to be particularly anomalous, scoring very high on participation and yet having a low overall score. Removing California from the analysis improves the correlation somewhat (r = .344), but it is still not particularly robust. However, 127 the relationship is in the expected direction, and the correlation is high enough to suggest that it ought to be tested as a contributing factor to AG success. Success in Court: A Multivariate Model — To this point, we have discovered that AGs win at high rates, both considered in isolation and in comparison with other litigants. They also tend to win more in criminal cases, and in cases in which they are representing the respondent. There also appears to be a small correlation between AG participation and AG win rates. However, to this point we have considered these dimensions of AG success in isolation. At this point, it is appropriate to test these factors simultaneously in a multivariate model. As a linear relationship is expected between dependent and independent variables, and the dependent variable is continuous, ordinary least squares regression will be utilized. As there was evidence to suggest that different dynamics are involved in criminal and non-criminal cases, three different dependent variables will be utilized in the following analyses: OVERALL = percentage of all cases won by an AG in a calendar year CRIME = percentage of criminal cases won by an AG NONCRIME = percentage of non-criminal cases won by an AG 128 The model to be estimated initially is: OVERALL CRIME = b + b RESPOND + b PARTICI + b AGADVAN + + b POWER + to CONSERV + b AGGOV + e

where RESPOND = the percentage of cases in which the AG represented the respondent in a calendar year

PARTICI = the percentage of the court's cases in which the AG participated in a calendar year AGADVAN = the number of justices above or below a court majority belonging to the same party as the AG

POWER = modified version of Boersema's (1987) index of AG power CONSERV = the percentage of cases in which the AG represented the conservative position5 in a calendar year AGGOV = 1 if the AG and governor belong to the same party 0 if otherwise

Each of these variables is expected to be positively associated with AG success. Most follow naturally from the previous comments, but a few words should be said about a couple of them. First, regarding POWER, I had argued that an AG office's success should be partially determined by its ability to present strong cases, which itself is a function of the office's discretion over appeals and the resources which it brings to the case. To capture this dimension of AG power, I am using a re-calculation of an index created by Boersema (1987), as described in Chapter III, note 4. The AGADVAN variable was calculated thus: the number of justices of the AG's party minus the number of justices of the 129

other party. Thus, if the AG's party held 6 seats on a nine- person court, AGADVAN would equal 3. This is an attempt to control for the size of the court in question, and theoretically ranges from +9 to -9. Finally, as this is pooled, cross-sectional data, there is the danger of autocorrelation. When running the models, the Durbin-Watson d statistic was inconclusive regarding the presence of autocorrelation in the models which had OVERALL and NONCRIME as the dependent variables, but the statistic did

indicate first order autocorrelation in the CRIME model. To err on the safe side, the models estimated in this chapter will also include four time-point dummy variables. Table 12 presents the parameter estimates for this model as it concerns all three dependent variables. As the

direction of the relationships has been specified, we may utilize a one-tailed test of significance. One of the first things that becomes apparent is that the fit of the model is not particularly good concerning any of the three dependent variables; the best model explains less than a third of AG support in criminal cases. Among the substantive variables, only PARTICI and AGGOV give any cause for gratification. It seems at this point that AGs win more frequently when they and the governor belong to the same party, and their winning rate increases as their participation rate increases. (And the POWER variable is strongly but negatively associated with AG success, contrary to expectations). 130 Due to the mediocre performance of this model, I decided to include a dummy variable for California (CALIF). It is clear, both from the data presented thus far and from the conventional wisdom, that the California Supreme Court is an

unusual state court of last resort in many respects. In virtually every bivariate relationship discussed thus far, California appeared to be a deviant case. It possesses the

characteristics which seem to be correlated with high AG success (i.e. appearing freguently, appearing overwhelmingly in criminal cases, and mostly representing the respondent), and yet it had low overall success, for reasons discussed

earlier. The parameter estimates for the original model, now including a CALIFornia dummy variable, are presented in Table 14.

Table 13 reveals that the addition of the California variable improves the performance of the models, to a rather

substantial degree in the OVERALL and CRIME models, but not in the NONCRIME model. However, the California variable itself is strongly and negatively associated with AG success in all three models — and this will continue to be the case in virtually analysis to come. The California AG begins with a deficit on the range of 15 percentage points when compared with other AGs. Interestingly enough, by including CALIF, the AGGOV variable no longer appears significant in any of the models. During all five years of this study, the California AG and governor belonged to different political parties, and 131

TABLE 13 PARAMETER ESTIMATES FOR MODEL OF INDIVIDUAL SUPPORT FOR AO (INCLUDING CALIFORNIA DUMMY)

DV = OVERALL DV = CRIME DV;=NONCRIME

b t b t b t

Constant 63.63 3.73** 37.24 2.36* 96.08 3.21** RESPOND 0.12 1.10 0.21 1.56 -0.23 -1.16 PARTICI 0.39 2.50** 0.37 2.10* 0.30 1.18 AGADVAN 0.90 0.90 0.20 0.15 1.35 0.82 POWER -0.45 -1.01 -0.83 -0.99 -0.69 -0.91 CONSERV -0.18 -1.22 (n.a.) -0.17 -0.71 AGGOV 3.67 0.99 6.67 1.36 5.49 0.86 CALIF -18.48 -3.28*** -16.14 -2.19* -11.10 -1.13 1987 0.45 0.10 5.53 1.02 -5.61 -0.78 1985 -0.49 -0.11 5.35 0.91 -8.51 -1.09 1983 -2.27 -0.50 5.69 0.95 -5.14 -0.65 1981 1.92 0.40 9.09 1.52 -11.93 -1.48

r2 = .383 r2 = .431 r2 = .235 n = 60 n = 44 n = 57

*** p < .001 ** p < .01 * p < .05 the AG fared quite poorly. This is NOT said to imply causation, but it may explain why the AGGOV variable previously seemed important.

Of the other substantive variables, only PARTICI remains significant. I am particularly disappointed with the performance of the AGADVAN variable. I had hypothesized that the more members of an AG's party on the court, the more the AG ought to win. This does not appear to be the case. However, as suggested previously, although the attorney general is the individual responsible for state litigation, it is the governor who is ultimately the head of the executive branch. The attorney general will often be called upon to defend actions undertaken by the governor or the governor's agents, and thus I had hypothesized that justices of the 132 governor's party should give greater support to the AG than other justices. To test this hypothesis, I have rerun the original model (the CALIF variable will continue to be included), but substituting the following variable for AGADVAN: GOVADVAN = the number of justices above or below a court majority belonging to the same party as the governor

TABLE 14 PARAMETER ESTIMATES FOR MODEL OF INDIVIDUAL SUPPORT FOR AO (INCLUDING GOVADVAN VARIABLE)

DV = OVERALL DV =CRIME DV= NONCRIME

b t b t b t

Constant 50.17 2.18* 43.77 3.06** 76.83 2.88** RESPOND 0.15 1.39 0.21 1.83* -0.14 -0.78 PARTICI 0.31 2.06* 0.22 1.52 0.26 1.07 GOVADVAN 1.06 2.54** 1.83 3.54*** 0.89 1.30 POWER -0.28 -0.66 -0.65 -0.93 -0.45 -0.66 CONSERV -0.04 -0.28 (n. a. ) -0.06 -0.27 AGGOV 1.62 0.53 2.85 0.83 2.16 0.43 CALIF -22.70 -4.33*** -19.61 -3.32*** -17.90 -2.06* 1987 3.19 0.78 5.87 1.25 -1.03 -0.15 1985 3.52 0.84 7.64 1.54 -8.51 -0.20 1983 1.74 0.41 7.99 1.60 1.55 0.22 1981 3.59 0.83 7.12 1.45 -5.60 -0.80

r2 = .478 r2 = .624 r2 = .191 n = 60 n = 44 n = 57

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

The results are printed in Table 14. This model is a substantial improvement over the original. It explains almost half the variation in OVERALL support (r2 = .478), and over 60% of the variation in CRIME support (r2 = .624).6 However, like the original model, it is not very successful at explaining variation in non-criminal cases. 133 Regarding specific variables, participation is again positively and significantly associated with AG success, while the CALIFornia AG continues to be disadvantaged. It also appears that the decision to substitute GOVADVAN for AGADVAn was a wise one, since the former not only greatly improves the fit of the model but is also associated with AG success in a statistically significant manner. However, the practical significance of this should not be overstated, because for each additional member of the governor's party on a court, the AG's success only increases by about one percentage point. This means that even in the most extreme circumstances — a 9- person court swinging from complete domination by one party to complete domination by the other — would only result in an 18 percentage point increase in support for the AG .7 This is very unlikely to happen over the term of a single governor. With these exceptions, the other variables wash out. The POWER variable performs poorly in all models, and it appears to make little difference whether the AG and the governor belong to the same political party, once we control for California. The rate at which the AG represents the respondent in criminal cases IS associated with success in those types of cases, but the same does not hold true for non­ criminal cases and cases overall.

CONCLUSION In summary, regarding attorney general success among courts collectively, we may make the following observations: 134 1) AGs win at rates that are comparable to the win rates of the U.S. Solicitor General, and win more than other, similarly-situated litigants. 2) AGs win more in criminal cases than non-criminal cases.

3) The larger percentage of a court's docket an AG's office consumes, the higher percentage of cases it wins.

4) The more members of a governor's party on a court, the higher the win rate the AG will have. The same does not hold true regarding members of the AG's party.

5) California is clearly a deviant case, with the AG losing both criminal and non-criminal cases at an alarming rate.

To this point, we have only considered courts as collectivities. However, the decisions reached by courts collectively are dependent upon the choices made by individual justices. In the following chapter, I will begin to focus more specifically focus on those factors which influence judicial behavior on the individual level, including the effect of ideology and judicial appointment. 135

ENDNOTES

1. Particularly since the term "public interest" is such a amorphous one that it gives the incumbent a good deal of latitude in defining it. 2. In calculating winning percentages, all "mixed" cases were apportioned as one-half point to both sides in the litigation. For example, if an AG had won 50 cases, lost 40, and 10 were mixed, 5 points were given to both sides, for an overall win rate of 55%. The formula would be:

number of wins + 1/2 number of mixed total number of cases

3. Obviously, this was also necessary due to concerns of sample size. 4. Significance levels were derived utilizing the original frequencies. 5. Since the AG inherently takes the conservative position in criminal cases, there was a high degree of correlation (r = .843) between the percentage of times that an AG took a conservative position (CONSERV) and the percentage of an AG's caseload that was made up of criminal cases (CRMPART). Either one could have been utilized in the analysis, but CONSERV performed better and has more substantive content.

6 . Since r2 necessarily increases as variables are added, I reran the models without the four time-point dummy variables. The proportion of variance explained does not decrease dramatically OVERALL = .452 CRIME = .583 NONCRIME =.168 7. In an attempt to test for alternative explanations, I substituted another variable, REPADVAN (the number of Republican justices above a majority) for AGADVAN and GOVADVAN in the model. Essentially, would simply knowing the number of Republicans on a court be more useful than knowing the number of members of the AG's or governor's party on the court? The model including REPADVAN had a higher r2 (.385) than the model including AGADVAN, but lower than the model including 136 GOVADVAN. However, the variable itself was not significant, even in criminal cases. CHAPTER V CONTOURS OF SUPPORT AMONG INDIVIDUAL JUSTICES

The model developed in the previous chapter was relatively successful in explaining variations in AG success among courts collectively, at least regarding overall success

and success in criminal cases. Focusing on courts as collective institutions is appropriate, because the outcome of the case is the matter of substantive importance to the

litigants involved. Most litigants probably care only about whether they win or lose, regardless of the margin of the

victory.1 However, the student of judicial politics is not content to stop there, and thus much effort has been expended in attempting to discover determinants of the behavior of individual jurists. Hence, in this chapter, I now turn to explaining variations in support for the attorney general among individual state supreme court justices. In Chapter IV, the independent variables utilized were, for the most part, structured in such a way as to have a uniform impact on all justices. For instance, the rate at which the attorney general participates in litigation was hypothesized to increase support on the court for the AG's office. This increased support was hypothesized to apply to

137 138 all justices, regardless of their partisan affiliation or method of ascension to the bench. Similarly, the percentage of cases in which an AG represented the appellee was hypothesized to decrease support for the AG among all justices, and not merely those of a particular party. One of variables which was not expected to have a uniform effect (CONSERV) did not perform very well in the model. Two party-

based variables (AGADVAN and GOVADVAN) were included, but this was to take into account the fact that the attorney general and governor are separate institutional entities and hence need to be analytically differentiated. In this chapter, many of the same factors will be

included in an attempt to explain individual judicial

behavior. In addition, I will turn to two dimensions of AG support which have not yet been adequately considered. First, I will attempt to explain AG support while controlling on a justice's ideology as measured by his or her votes in other cases. Second, the various hypotheses regarding the effect of

judicial selection system will be considered. At the conclusion of the chapter, these factors will be tested in multivariate models of individual-level judicial behavior. INDIVIDUAL SUPPORT FOR THE AG: A FIRST CUT A Note on Dissent Rates — Many scholars (i.e. Grossman 1966) have noted that, although the vast majority of judicial decisions are handed down unanimously, focusing on unanimous cases poses certain problems for the analyst. This problem is 139 TABLE 15 DISSENT RATES ON STATE COURTS OF LAST RESORT

STATE DISSENT RATE DISSENT RATE (ALL CASES) (AG CASES) Alaska 20.3% 23.4 Arizona 9.7 8.3 California 52.7 58.5 Illinois 21.3 21.2 Missouri 40.2 42.0 New Jersey 25.4 28.5 New York 21.1 18.0 North Dakota 8.6 14.6 Ohio 42.6 41.6 Tennessee 12.7 22 .2 Washingto 29.9 27.4 Wyoming 23.1 23.9

particularly acute for scholars studying courts other than the

U.S. Supreme Court, where the norm of dissent seems to run stronger. Naturally, the degree to which this was true of the twelve states included in this study varied. Table 5.1

shows the percentage of cases in which at least one justice dissented at least in part, both in all cases generally and in cases involving the attorney general particularly. By design, I selected many states in which the dissent rate should have been high enough to allow for meaningful analysis. However, for both practical and methodological

reasons, I decided to also select a number of states in which the dissent rate is relatively low. Practically, the number

of state supreme courts with "high" dissent rates is limited.2 Theoretically, choosing only these states would result in a wholly unrepresentative sample. Certainly, analyzing courts that tackle controversial issues and display divisiveness is 140 more intriguing for both the analyst and the reader. However, in order to get a truly accurate, comprehensive, comparative idea of the nature of the relationship between AGs and state courts, we must examine the relationship in its many facets. Hence, dissent rates in the states included in this study range from a high of 58.5% in California to a low of 8.3% in Arizona. In the aggregate, dissent occurred more frequently in cases involving the AG than in cases not involving the AG, but this pattern held true in only 7 states and the differences are generally not significant. Despite this, there are still a number of states which, quite frankly, lend little excitement to the analysis. For example, there were no statistically significant differences between the voting patterns of any two justices within a single year on the Arizona Supreme Court during any year in this study. This was primarily due to the very low rate of dissent on that court. In other states, such as New York, North Dakota, and Washington, few statistically significant differences between justices emerged simply because the AG did not appear in a sufficiently large number of cases. However, even in some states with relatively high participation and dissent rates, such as New Jersey, the pattern of dissent was seemingly random enough as to minimize differences in the overall patterns of individuals.3 In other states, the court's level of dissent is primarily the handiwork of an individual justice. For example, much of the dissent on the 141 Illinois Supreme Court during this period came from Justice Seymour Simon, who was a consistent opponent of the death penalty. As a result of the minimal differences between individual justices on a single court, in the first set of analyses in this chapter, AG support scores will be aggregated for all members of a particular class within a state. For example, when comparing appointed vs. elected justices, the support scores of all appointed justices within a particular state will be aggregated,4 and then compared. However, regarding the multivariate analysis later in the chapter, each justice- year is included as an individual case. Partisan Differences in AG Support — One of the traditional ways of viewing differences between jurists is their affiliation with a political party (i.e. Carp and

Rowland 1983; Nagel 1961). The selection of a partisan affiliation is perceived as one of the best indicators of an individual's underlying value preferences. It would be interesting to see if there are indeed differing voting patterns in cases involving the attorney general among members of the two political parties, and if these differences are consistent from state to state. Table 16 reveals that, in most states, members of the two political parties do not differ significantly in their support for the attorney general. In only three of the 53 state-years does the difference between Democrats and Republicans reach TABLE 16 SUPPORT FOR THE AG, DEMOCRATS VS. REPUBLICANS

STATE YEAR DEMS REPS DIFFERENCE

AK 1987 58.6 58.6 0.0 1985 44.2 47.9 -3.7 1983 63.9 67.8 -3.9 1981 59.3 67.1 -7.8 1979 73.6 77.9 -4.3 AZ 1987 74.3 73.7 0.6 1985 83.5 79.7 3.8 1983 56.5 56.2 0.3

1981 61.8 62.6 - 0.8 CA 1987 45.5 57.1 - 11.6 1985 38.2 65.8 -27.6** 1983 32.8 62.3 -29.5** 1981 49.7 66.3 -16.6* 1979 36.9 65.4 -28.5** IL 1987 60.4 64 .2 -3.8 1985 55.2 61.4 - 6.2 1983 54.7 70.4 -15.7* 1981 55.3 58.2 -2.9 1979 65.7 64.5 1.2 MO 1987 71.8 80.5 -8.7* 1985 63.2 67.6 -4.4 1983 63.5 73 .8 -10.3* 1981 72.2 81.9 -9.7 1979 52.7 66.0 -13.3* NJ 1987 51.6 50.6 1.0 1985 74 . 3 68.9 5.4 1983 72.3 73 .7 -1.4 1981 70.0 70.0 0.0 1979 66.5 68.0 -1.5 NY 1987 54.3 58.3 -4.0 1985 80.0 83 .0 -3.0 1983 44.5 42.2 2.3 1981 73.3 69.7 3.6 1979 54.8 55.3 -0.5

ND 1987 64.2 66.3 - 2.1 1985 64.9 67.9 -3 . 0 OH 1987 47.8 58.4 - 1 0 .6* 1985 63.2 72.1 -8.9*

1983 65.6 66.6 - 1.0 1981 56.1 61.8 -5.7 1979 77.4 79.3 -1.9 TN 1987 76.8 77.0 - 0.2 1985 65.0 60. 0 5.0 1983 66. 1 67.8 -1.7 1981 56.6 62.2 -5.6 1979 48.0 51.4 -3.4 143 Table 16 - cont'd

1987 61.2 64.4 -3.2 1985 51.9 39.9 12.0 1983 58.3 61.9 -3.6 1981 59.0 54.2 4.8 1979 6 6 . 3 63.7 2 . 6 1987 69.9 75.1 -5.1 (seven state-years excluded due to lack of data) **p < .05 * p < .10 144 the conventional significance level of .05, and each of these occurs in California. Even if we relax the desired level of significance to p < .1 0 , there are statistically significant differences in only seven state-years. As mentioned above, in many states this lack of significant differences results from the low level or dissent or an insufficiently large number of cases in which the AG's office took part. In spite of this, certain generalizations can be made. First, the California, Missouri, and to a lesser degree, Ohio supreme courts stand out as the most partisan courts in regard to the AG's office. The partisanship on the California court clearly reflects the bipolar nature of support for the death penalty cases handled by the AG's office. As we might expect, the only year in which Democrats were not significantly less likely to uphold the death penalty was the one year (1987) included in this study after the electoral defeat of Rose Bird and her more liberal colleagues. Regarding Ohio, Tarr and Porter (1988) identify workers compensation as being the issue generating the most division on that state's Supreme Court. Workers' comp cases are handled by the Ohio attorney general's office, and this may partially account for that court's partisanship. This depiction of partisan conflict mirrors Leonard's (1989) findings that the Ohio court was more ideologically divided in the 1980s than in the 1970s. The partisanship displayed by the Missouri court is somewhat ironic, considering the claim of proponents of Missouri's 145

merit selection plan that it will somehow "remove" politics from the selection procedure (i.e. Bunch and Casey 1990) — and hence (presumably) also remove politics from the subsequent behavior of justices on the bench. Second, overall Republicans are somewhat more likely to support the attorney general than are Democrats. While we might suspect that this is due in part to a greater Republican propensity to vote for the state in criminal cases, we also find this to be the case in states such as Ohio, in which the AG's workload is composed almost exclusively of civil cases. Table 17 shows Democratic vs. Republican win rates in criminal and non-criminal cases separately (four states are excluded

for a lack of a sufficient number of cases) . As before, Republicans are generally more supportive of the attorney general than are Democrats. This appears to hold true in non­ criminal as well as criminal cases. This is somewhat surprising, because in non-criminal cases, the attorney

general is more likely to represent a liberal position. Partisan differences are most pronounced in California and Missouri in criminal cases, and in Missouri, Ohio, and Washington in non-criminal cases (at least in particular years). In most instances, then, it would appear that

partisanship is not an enlightening indicator to the votes of a state supreme court justice in cases involving the attorney general. However, I had hypothesized that individual judicial 146

TABLE 17a SUPPORT FOR THE AG, DEMOCRATS VS. REPUBLICANS (CRIMINAL CASES ONLY) STATE YEAR DEMS REPS DIFFERENCE

AK 1987 50.0 50.0 0.0 1985 55.6 57.4 -1.8 1983 63.6 80.1 -16. 5 1981 62.5 74.3 -12.2 CA 1987 38.2 55.3 -17.1* 1985 39.3 68.0 -28.7** 1983 33.1 64.3 -31.2** 1981 56.7 73.2 -16.5** 1979 37.8 69.3 -31.5*** IL 1987 64.3 72.3 -8.0 1983 56.2 65.4 -9.2 1981 66.0 66.9 -0.9 1979 76.8 72.8 4.0 MO 1987 74.6 83.2 -8.6 1985 65.2 73.6 -8.4 1983 68.4 75.4 -7.0 1981 73.7 92.3 -18.6** 1979 52. 1 71.2 -19.1** NJ 1987 46.2 45.8 0.4 1985 62. 5 76.5 -14. 0 1983 69. 2 73.7 -4.5 1981 68.4 63.2 5.2 1979 58. 6 58.6 0.0 TN 1987 71. 6 70.6 1.0 1985 69.1 65.8 3.3 1983 63.6 65.2 -1.6 1981 49.7 52.6 -2.9 WY 1987 73. 1 82.8 —9.7*** (certain state-years excluded due to lack of data and/or sufficient number of cases)

* p < .10 ** p < .05 *** p < .01 147

TABLE 17b SUPPORT FOR THE AG, DEMOCRATS VS. REPUBLICANS (NON-CRIMINAL CASES ONLY)

STATE YEAR DEMS REPS DIFFERENCE

AK 1987 61.5 60.2 1.3 1985 40.0 45.0 -5.0 1983 64.0 59.4 4.6 1981 56.6 61.1 -4.5 IL 1987 45.0 48.8 -3.8 1983 61.3 64.2 -2.9 1981 47.4 49.2 -1.8 1979 42.9 44.8 -1.9 MO 1987 62.5 75.0 -12.5 1985 60.7 57.2 3.5 1983 58.3 72.2 -13.9 1981 70.0 45.0 25.0 1979 52.1 71.2 -19.1** NJ 1987 61.0 57.1 3.9 1985 76.9 70.5 6.4 1983 66.5 76.2 -9.7 1981 75.5 76.9 -1.4 1979 71.9 72.2 -0.3 NY 1987 51.7 56.7 -5.0 1985 82.5 85.7 -3.2 1983 42.2 39.6 2.6 1981 70.2 66.2 4.0 1979 56.5 59.4 -2.9 ND 1987 60.7 59.7 1.0 1985 49.3 51.2 -1.9 1983 60.8 66.4 -5.6 OH 1987 47.8 58.4 -10.6* 1985 60.2 69.2 -9.0* 1983 52.3 57.4 -5.1 1981 61.0 62.1 -1.1 1979 77.4 79.3 -1.9 TN 1987 82.5 82.5 0.0 1985 58.0 50.0 8.0 1983 68.8 70.5 -1.7 1981 69.1 72.2 -3.1 1979 75.0 75.0 0.0 * p < .10 WA 1987 78.5 67.1 11.4* ** p < .05 1985 56.3 39.2 17.1* 1983 45.5 60.5 -15.0* 1981 57.8 53.1 4.7 1979 66.1 63.5 2.6 WY 1987 73.1 82.8 -9.7 (certain state-years excluded due to lack of data and insufficient number of cases) support should be dependent upon whether that justice belonged to the same party as the attorney general and/or governor. Table 18 re-presents the data from Table 16, showing differences in AG support between members of the governor's (and attorney general's) party and members of the opposite party. Since this is merely a re-presentation of Table 16, again there are not many statistically significant differences. Looking at individual states tends to lend support to the impression that it is the partisanship of the justice rather than the partisanship of the AG or governor that seems to matter. For example, Democratic justices in California were simply less inclined to support the attorney general, whether that AG was a Democrat (John van de Kamp) or a Republican (George Deukmeijian) , or whether the governor was a Democrat (Jerry Brown) or a Republican (George Deukmeijian) . It appears from this examination that it is the issue rather than the identity of other institutional incumbents that counts.

The multivariate analysis in the previous chapter suggested that the number of justices of a governor's party on a court was positively related to AG support. Although one has to look for it, Table 18 may provide further support. It is true that individual differences in individual years are not stark. But if we look at the overall patterns, members of the governor's party gave the AG an advantage in 29 state- years, whereas members of the AG's party gave the AG an 149

TABLE 18a TABLE 18b Support by members of AG's Support by members of party minus support by other governor party minus justices support by other justices

AK 1987 0.0 0.0 1985 -3.7 -3.7 1983 -3.9 -3.9 1981 7.8 7.8 1979 4.3 4.3 AZ 1987 -0.6 - 0.6 1985 -3.8 3.8 1983 -0.3 0.3 1981 0.8 - 0.8 1979 1.7 -1.7 CA 1987 -11.6 11.6 1985 -27.6** 27.6** 1983 -29.5** 29.5** 1981 16.6* -16.6* 1979 28.5** -28.5** IL 1987 -3.8 3 . 8 1985 -6.2 6.2 1983 -15.7* 15.7* 1981 2.9 2.9 1979 -1.2 - 1.2 MO 1987 8.7* 8.7* 1985 4.4 4.4 1983 10.3* 10. 3* 1981 9.7 9.7 1979 13.3* 13.3* NJ 1987 -1.0 - 1.0 1985 -5.4 -5.4 1983 1.4 1.4 1981 0.0 0.0 1979 -1.5 -1.5 NY 1987 -4.0 -4.0 1985 -3.0 -3.0 1983 2.3 2.3 1981 3.6 3.6 1979 -0.5 -0.5 ND 1987 -2.1 2.1 OH 1987 -1 0 .6* 1 0 . 6* 1985 -8.9* 8.9* 1983 -1.0 1.0 1981 -5.7* -5.7* 1979 -1.9 -1.9 TN 1987 -0.2 0.2 1985 5.0 5.0 1983 -1.7 -1.7 1981 -5.6 -5.6 150

Table 18 - cont'd 1979 -3.4 -3.4 WA 1987 3.2 -3.2 1985 -12.0 12.0 1983 3.6 3.6 1981 -4.8 -4.8 1979 -2.6 -2.6 WY 1987 -5.2 5.2

* p < .10 ** p < .05 151 advantage in only 18 state-years. Since we previously found that Republicans are more likely to support the attorney general, we might attribute this phenomenon that there were more Republican governors than Republican attorneys general in this sample. However, in fact, the exact opposite is true. Republicans held the office of AG in 42.6% of the state-years, but held the governorship in only 38% of the state-years. Hence, support for the AG is greatest from Republicans and members of the governor's party, but Republicans are much less likely to be governors than are Democrats. This anomaly leads me to include these variables in the multivariate analyses to follow. Judicial Selection System — in Chapter I, I argued that appointed justices ought to support the executive branch (and thus the AG) more than elected justices. However, as I also feel that there should be a partisan component (see Emmert and Traut 1994), I have also adopted Jeffrey Segal's (1990) hypothesis that justices should give greatest support to the governor that appointed them, second highest support to later governors of a different party, and least support for a governor of a different party.

Table 19 shows the differences in rates of AG support between appointed and elected justices in the six states in which such a comparison is possible. Those states which have thus far shown the most interesting features (California and Missouri) are states in which all justices are initially ap- 152

TABLE 19 SUPPORT FOR THE AG, APPOINTED VS. ELECTED JUSTICES STATE YEAR APPOINTED ELECTED DIFFERENCE

AZ 1987 74.3 77.0 -2.7 1985 77.7 79.4 -1.7 1983 55.9 57.0 -1.1 1981 62.9 61.9 1.0 1979 63 .1 63.7 -0.6 NY 1987 55.1 58.3 -3.2 1985 81.0 81.5 -0.5 1983 44.4 42.8 1.6 1981 67.2 71.3 -4.1 ND 1987 64.2 6 6 . 3 -2.1 1985 64.9 67.9 -3.0 1983 68.9 64.4 4.5 1981 74.9 75.0 -0.1 1979 46.9 47.4 -0.5 TN 1987 77.0 76.8 0.2 1985 60.0 65.0 -5.0 1983 67.8 66.1 1.7 1981 62.2 56.6 5.6 WA 1987 64.4 68.2 -3.8 1985 39.9 51.9 -1 2 .0* 1983 61.9 58.3 3.6 1981 54.2 59.0 -4.8 1979 64.5 64.8 -0.3

*p < . 10 153 pointed, and thus they are not included. As a result, only one statistically significant difference appears, and the differences in most state-years are minimal indeed. If we look at overall patterns, the tendency is, contrary to expectations, for appointed justices to be somewhat less supportive of the AG than elected justices. Reflecting Brace and Hall's (1992) finding that voting behavior in death- penalty cases is conditioned by the electoral incentive, it may be perhaps that elected justices are more sensitive to the need to uphold the state in criminal cases and are hence more likely to support the AG. This musing is obviously speculative, however, and in any case, the differences between appointed and elected judges are too slight to linger on.

As indicated earlier, the executive should not receive uniform support from all judicial appointees. Rather, appointees should be more responsive to governors of their own party, and particularly those governors who appointed them. Table 20a compares support for the AG between justices ap­ pointed by the current governor, and those who reached the bench through other means (either through election or appointment by a different governor). Generally, justices do not tend to give greater support to the executive branch when it is controlled by the governor that appointed them. Overall, there are more cells in the predicted direction than otherwise, but the differences in magnitude in most instances are so weak as to dampen any enthusiasm that might generate. 154

TABLE 20a TABLE 20b Support by justices appointed Support by justices ap­ by the current governor minus pointed by a governor of support by other justices the party of the current governor minus support by other justices

AK 1987 n.a. 0.0 1985 -1.6 3.7 1983 n.a. -3.9 1981 4.9 7.8 1979 3.4 4.3 AZ 1987 n.a. n. a. 1985 -3.2 - 2.8 1983 -2.2 0.3 1981 n.a. 0.9 1979 n.a. -4.0 CA 1987 11.6 11.6 1985 27.6** 27.6** 1983 n.a. 29.5** 1981 -8.3 -16.6** 1979 -15.2** -28.5** MO 1987 15.0** 8.7* 1985 n.a. 4.4 1983 1 1 .2 ** 10.3** 1981 1 0 .6** 9.7 1979 13.3** 13.3**

NJ 1987 2.0 - 1.0 1985 -3.6 -5.4 1983 -2.1 1.4 1981 0.5 0.0 1979 5.6 -1.5 NY 1987 2.5 2.5 1985 -3.3 -1.5 1983 n.a. 1.6 1981 -3.7 -3.7 1979 n.a. n.a. ND 1987 -2.5 - 2.1 1985 -1.2 - 1.2 1983 n.a. n.a. 1981 n.a. n.a. 1979 -0.5 -0.5 TN 1987 n.a. n.a. 1985 -5.0 -5.0 1983 1.7 1.7 1981 5.6 5.6 1979 3.4 3.4 * p < 0.10 WA 1987 n.a. n.a. ** p < .05 1985 n.a. n.a. 1983 0.9 3 .6

1981 1.9 - 2.6 1979 n.a. n.a. 155 The lack of strong results is particularly disappointing in Alaska, New Jersey, and Wyoming, the three states in which the AG is appointed by the governor. The exceptions to the general trend are again California and Missouri. The fact that Jerry Brown's appointees were less likely to support the AG can again be largely attributed to the heavy load of criminal cases handled by the California AG's office (i.e. Eiler and Vincent 1986).

Virtually the same generalizations can be made regarding Table 20b, which expands the scope of the investigation a bit. Since gubernatorial appointments presumably give the executive an opportunity to screen candidates for their ideological views, the appointees of Republican governors should act different than the appointees of Democratic governors (Emmert and Traut 1994; George and Epstein 1992). Table 20b compares AG support rates between those justices appointed by a governor of the same party as the current governor, and all other justices (either elected or appointed by a governor of the opposite party). Again, the differences between the two groups of jurists is minimal with the exceptions of California and Missouri. In a table not presented, I compare only judicial appointees — those appointed by a member of the current governor's party, and those appointed by a past governor of the opposite party. Yet again, a majority of state-years is in the predicted direction, but the differences 156 between the two sets of justices in most states are extremely

modest and not statistically significant.5 Thus, with the exceptions of California, Missouri, and Ohio, I have found few statistically significant variations in

intra-court support for state attorney general that can be attributed to either partisanship or judicial selection system. However, I turn now to testing a multivariate model of individual judicial behavior in cases involving the attorney general. While the lack of significant findings in this chapter is largely due to the low levels of dissent on many of these courts, in some cases (such as Illinois and New Jersey) seemingly significant variations on the part of one or more individual justices have been subsumed in the aggregate analysis thus far conducted. A multivariate analysis will allow us to more completely confirm or refute these hypotheses. INDIVIDUAL SUPPORT FOR THE AG: A MULTIVARIATE ANALYSIS The models to be estimated in this chapter utilize a number of variables similar to those included in the models estimated in Chapter IV, adapted to the individual level where necessary. Since these variables were hypothesized to have a uniform effect on all justices, it is appropriate (indeed, essential) to test them on the individual level. For example, an AG's repeat player status was argued to have a positive effect on all of the justices on a court, and hence on the court as a whole. However, these models also feature the 157 inclusion of variables related to the method of judicial selection. In addition, an attempt to control for the ideology of justices will be made at the end of the chapter. As before, it seems logical to begin by focusing on overall support for the attorney general. Findings previously presented suggest that separate analyses ought to be conducted on criminal and non-criminal cases as well. Hence, the dependent variables to be regressed in this first section are:

OVERALL = the percentage of cases in which a justice voted for the litigant represented by the attorney general in a calendar year CRIME = the percentage of cases in which a justice voted for the AG when prosecuting a criminal defendant NONCRIME = the percentage of cases in which a justice voted for the AG in non-criminal cases

Initially, I will only test for the effect of judicial appointments generally. Following this initial analysis, other dimensions of judicial appointments will be considered. Thus, the model to be estimated is:

y = b0 + ^RESPOND + b2PARTICI + b3SAMEASAG + b4POWER + b5APPOINT + b6AGGOV + B?CALIF + eL where

RESPOND = the percentage of cases in which the AG represented the respondent PARTICI = the number of cases in which the AG participated as a percentage of all cases decided by the court SAMEASAG = 1 if the justice belongs to the AG's party 0 if otherwise POWER = the overall power rating given by Boersema (1987) given to that state's attorney general APPOINT = 1 if the justice was initially appointed 0 if otherwise 158 AGGOV = 1 if the AG and governor belong to the same party 0 if otherwise CALIF = l if the state is California 0 if otherwise6

Pooled regression poses certain problems because the number of cases and the identities of justices change from year to year. Hence, ordinary least squares regression was performed five times, once for each calendar year. Table 21 presents the parameter estimates for the model which included OVERALL success as the dependent variable.

TABLE 21 PARAMETER ESTIMATES FOR INDIVIDUAL MODEL OF OVERALL AG SUCCESS

YEAR = 1987 YEAR = 1985 YEAR = 1983

b t b t b t

Constant 111.09 5.33*** 77.43 2.54** 76.02 4.97*** RESPOND 0.21 1.61 0.92 3.09** -0.P1 -0.93 PARTICI 0.03 0.37 -0.46 -1.55 0.54 3.28*** SAMEASAG 0.05 0.03 -1.43 -0.47 2.37 1.06 POWER -0.69 -3.79 -0.62 -2.11 -0.41 -2.48 APPOINT -2.05 -0.80 -11.06 -3.09 2.60 0.93 AGGOV -4.89 -1.88 -3.24 -0.59 9.06 2.54** CALIF -20.03 -3.72*** -37.08 -4.35*** -26.13 -5.97***

r2 = .517 r2 = .641 r2 = .628 n = 69 n = 71 n = 69

YEAR = 1981 YEAR = 1979

b t b t

Constant 73.15 5.06*** -30.96 -0.91 RESPOND 0.41 4.24*** 0.35 2.13* PARTICI -0.16 -1.36 0.18 0.88 SAMEASAG 1.07 0.46 -3.46 -1.23 POWER -0.42 -2.77 0.81 2.38* APPOINT -2.26 -0.73 2.08 0.54 AGGOV 7.17 2.37* -4.85 -1.26 CALIF -7.44 -1.60 -29.22 -5.15***

r2 = .467 r2 = .496 n = 72 n = 72 159 The fit of the model is relatively good and consistent

from year to year, with r2 ranging from .491 to .628. Unfortunately, the two variables which perform most consistently are both disappointing. One (CALIF) was not previously predicted by theory and was simply included as a dummy variable to account for an apparently anomalous case.

The second (POWER) is strongly associated with AG success, but in the wrong direction in 4 of the five state-years. Actually, this lends support to Boersema's findings that strong AGs were less likely to win, simply because of the nature of their caseload (1987, 124). He attributes this to the fact that stronger AGs are less likely to be handling criminal cases, which AGs win more than non-criminal cases. In order to determine whether this accurate, below I will test his finding that strong AGs win more civil cases, and weak AGs win more criminal cases. Other findings from Chapter IV are further supported here. That chapter had indicated that increasing the number of members of an attorney general's party on a court did not result in a corresponding increase in support for the AG. Similarly, the SAMEASAG variable does not perform very well in this model. In no year does it approach conventional levels of statistical significance, and in each year the parameter estimates are in the opposite direction as was predicted. Thus, it again appears that attorneys general receive little special consideration from their fellow partisans. The patterns of the other variables are less clear, because in some years they are significant and in other years they are not. In three of the five years, the percentage of cases in which the AG represented the appellee is positively and significantly associated with AG success. This is as hypothesized, but no similar relationship was discovered on the collective level. Further, in the last chapter, an AG's success was found to be positively associated with his or her rate of participation. In this individual level analysis, this relationship only holds true in one of the five state- years. At the collective level, obviously, only the stances of the justices in the majority are captured by the AG success variable. It may be the added influence of the dissenting justices that muddies the relationship at the individual level. The AGGOV variable also performs as expected in two of the five state-years. And finally, the APPOINTment variable washes out — in this model, appointed justices appear no more likely to support the AG than elected justices. In the last chapter, I found evidence to support the proposition that the number of justices sitting on a court of a governor's party was positively associated with an attorney general's success in court. A notion logically consistent with that finding is that individual members of the governor's party should give greater support to the AG than other justices. Thus, the previous model was re-estimated, substituting the following variable for SAMEASAG: 161 SAMEGOV = 1 if the justice belongs to the same party as the governor 0 if otherwise The results of this analysis are contained in Table 22. Whereas in the previous chapter, a similar substitution

(GOVADVAN for AGADVAN) produced a model that was significantly more robust, the differences here are not as striking. Substituting SAMEGOV for SAMEASAG increases the proportion of variance explained in only three of the five state-years, and in no case is the increase substantial. The performance of most of the independent variables is not significantly affected, so I will not linger on them. However, I would argue that the second model is somewhat preferable. The new variable leads to a better fit than the old, as in the previous chapter. Whereas SAMEASAG was insignificant in all years, SAMEGOV reaches the significance level of p < .05 in three of the five state-years. In each case the parameter estimate is in the predicted direction, lending corroboration to the findings of Chapter IV. Although there are certain dangers of aggregating the data because the identity and number of the justices included in the analysis changes slightly from year to year (Hanushek and Jackson 1977, 101), in order to more clearly test the comparative utility of the two models, it would be useful to pool the data. I have no a priori notion of what, if anything, might be distinctive about these years, and thus I will simply include a four dummy time point variables in the TABLE 22 162

PARAMETER ESTIMATES FOR INDIVIDUAL MODEL OF OVERALL AG SUCCESS (REVISED TO INCLUDE SAMEGOV VARIABLE)

YEAR = 1987 YEAR = 1985 YEAR = 1983

b t b t b t

Constant 107.91 5.11*** 82.07 2.60** 76.02 4.97*** RESPOND 0.22 1.69* 0.08 0.46 -0.11 -0.93 PARTICI 0.02 0.33 0.44 2.93*** 0.54 3.28*** SAMEGOV 1.47 0.75 5.14 1.73* 2.37 1.06 POWER -0.66 -3.63 -0.53 -1.75 -0.41 -2.48 APPOINT -1.71 •0.67 -7.01 -2.00 2.60 0.93 AGGOV -5.21 2.00 10.87 3.10*** 9.06 2.54** CALIF -20.45 •3.79*** -18.09 -2.60** -26.13 -5.97*** r2 = .522 r2 = .572 r2 = .632 n = 69 n = 71 n = 69

YEAR = 1981 YEAR == 1979

b t b t

Constant 72.14 5.43*** -37.45 -1.15 RESPOND 0.45 4.65*** 0.38 2.39* PARTICI -0.15 -1.35 -0.01 -0.03 SAMEGOV 4.75 1.78* 5.69 2.09* POWER -0.50 -3.33 0.87 2.67** APPOINT -5.51 -1.54 1.77 0.51 AGGOV 7.48 2.67** -0.54 -0.17 CALIF -4.73 0.99 -23.32 -4.68***

r2 = .517 r2 = .489 n = 72 n = 72

* p < .05 ** p < .01 *** p < .001 equation, as in the previous chapter. The parameter estimates for both models is presented in Table 23. The model containing the governor variable is more successful in terms of variance explained, but the difference between the two is negligible (r2 = .408 vs .400). And although the overall proportion of variance explained is lower than in the models estimated in the individual state-years, the performances of many of the independent variables is 163 TABLE 23 COMPARISON OF PARAMETER ESTIMATES OF THE TWO MODELS — ONE INCLUDING SAMEASAG, THE OTHER INCLUDING SAMEGOV (ALL CASES AGGREGATED) SAMEASAG SAMEGOV

b t b t

Constant 61.99 6.60*** 60.63 6.61*** RESPOND 0.22 4.13*** 0.19 3.74*** PARTICI 0.17 3.13** 0.19 3.78*** SAMEASAG/GOV -0.79 -0.65 3.98 3.42*** POWER -0.24 -2.62 -0.24 -2.65 APPOINT -1.19 -0.85 -1.35 -0.99 AGGOV 2.80 1.94* 3.03 2.21* CALIF -22.62 -9.32*** -21.75 -9.24*** 1987 0.25 0.37 0.20 1985 0.51 0.34 0.18 1983 -0.46 -0.87 -0.48 1981 0.28 0.71 0.39

r2 = .400 r2 = .408 n = 353 n = 353

* p < .05 ** p < .01 *** p < .001 improved (as might be expected with larger sample sizes). In both models, RESPOND, PARTICI, and AGGOV are positively and significantly related to AG success. For approximately each increase in AG participation or rate of representing the respondent of about 5%, the AG's success on an individual level increases about 1%. And although in this analysis it does appear that the AG wins more frequently when the AG and the governor belong to the same party, it results in only about a three-point increase. Also in both models, Boersema's POWER variable and the CALIF dummy variable are negatively 164 related to AG success, while the APPOINTment variable again washes out. The most striking difference between the two models concerns, not surprisingly, the two variables which are being tested against each other. In this pooled model, all else being equal, members of the governor's party support the attorney general at a rate that is about 4 percentage points higher than other justices. No similar relationship is discovered regarding members of the attorney general's party. This parallels the conclusions reached in the previous chapter. Although it appears that the partisan compatibility of the judiciary and the chief executive is an important element in the relationship, perhaps my biggest disappointment results from the parameter estimates for the judicial appointment

variable. It was hypothesized that, all else being equal, ap­ pointed justices would support the AG at higher rates than elected justices. However, the data suggest that this was not true, even when controlling for partisanship. However, it may be that the appointment variable by itself is too blunt. Thus, I re-ran the model again, substituting the following variable for the independent variables SAMEGOV and APPOINT:

CURRGOV = 1 if justice was appointed by the current governor 0 if otherwise In the interests of not complicating the analysis any further, I do not present the entire results of this model. Suffice it 165 to say that the performance of the other independent variables was not significantly affected, and while CURRGOV almost reached conventional significance levels, it fell somewhat short (t = 1.51) and thus was not as satisfactory as the model including SAMEGOV. Table 24 presents the parameter estimates for the second model run twice more, utilizing CRIME and NONCRIME as the dependent variables. For the remainder of this chapter, again in the interest of simplicity, I will only present parameter estimates for the aggregated dataset. I will not linger on

TABLE 24 PARAMETER ESTIMATES FOR MODEL OF AG SUCCESS (CRIMINAL AND NON-CRIMINAL CASES) CRIMINALCASES NON-CRIMINAL CASES

b t b t Constant 59.30 4.93*** 74.07 5.30*** RESPOND 0.24 3.46*** -0.07 -0.88 PARTICI 0.25 3.26*** 0.10 1. 39 SAMEGOV 5.62 3.41*** 3.46 2.05* POWER -0.42 -3.70 -0.17 -1.31 APPOINT -5.78 -2.58 2.62 1.33 AGGOV 12.22 5.75*** 0.56 0.26 CALIF -13.51 -4.36*** -19.14 -5.35*** 1987 5.41 2.16* -2.74 -1.05 1985 9.57 3.68*** -3.21 -1.21 1983 8.22 3.10** -1.42 -0.53 1981 8.19 3.30*** -4.11 -1.58

r2 = .473 r2 = .173 n = 249 n = 290

* p < .05 ** p < .01 *** p < .001 166

Table 24, but rather will highlight the important features. First, the fit of the model is once again better for criminal cases than non-criminal cases, both in terms of a higher proportion of variance explained and the number of variables showing consistent and robust effects. Beyond that, essentially the findings that have reported earlier are applicable to AG success in criminal cases, but not to AG

success in non-criminal cases. The only two variables which achieve significance in both models are CALIF and SAMEGOV, both in the predicted direction. One exception to the general pattern is that appointed justices seem LESS likely to support the AG in criminal cases. This may be related to Hall's (1992) finding that elected justices react to electoral incentives in death penalty cases, which may apply to criminal cases more generally.7 Ideology — For this dissertation, I coded all cases decided by these state courts of last resort rather than coding only those cases in which the AG's office participated. This was done so that I would be able to test for the effects of the variables in which I am interested while controlling on a justice's ideology, as measured by that justice's voting behavior in other cases. This greatly increased the time needed for data collection, and as Table 25 reveals, it appears as though that time was not particularly well spent.

Table 25 reveals the correlations between three variables — support rates for the AG as conservative (SUPCONS), support 167 TABLE 25 CORRELATIONS BETWEEN IDEOLOGICAL VARIABLES

SUPCONS SUPLIB SUPLIB 0.023 OTHERLIB -0.248 0.165

rates for the AG as liberal (SUPLIB), and the percentage of times the justice voted for the liberal position in cases not involving the AG (OTHERLIB). The correlations are low. It

appears as though justices use separate issue dimensions to evaluate the two groups of cases, although discovering these dimensions through factor analysis or some other technique is beyond the scope of this dissertation. However, since the effort has been invested, I will attempt to test the effects of the other variables while

controlling on ideology in a multivariate context. Again, the model to be estimated is similar to those above. However, here two new dependent variables are to be regressed: SUPLIB = Percentage of cases in which a justice supported the AG's office when it represented the liberal position SUPCONS = Percentage of cases in which a justice supported the AG's office when it represented the conservative position

The right side of the original model is kept intact, but an additional independent variable to indicate ideology is added: OTHERLIB = the percentage of cases not involving the attorney general in which a justice voted for the liberal side 168 This independent variable is used in estimating both models. It ought to be positively related to SUPLIB and negatively related to SUPCONS. The results are presented in Table 26.

TABLE 26 PARAMETER ESTIMATES FOR MODELS OF AG SUCCESS CONTROLLING ON IDEOLOGY DV = SUPLIB DV = SUPCONS b t b t Constant 25.14 1.52 81. 34 7.26*** RESPOND 0.19 2.12* 0.18 2.89** PARTICI -0.05 -0.54 0.20 3.24*** SAMEGOV 0.33 0.16 2.77 1.96* POWER 0.21 1.27 -0.39 -3.54 APPOINT 1.04 0.43 -0.46 -0.28 AGGOV -1.31 -0.53 3.39 2.02* CALIF 6.61 1.52 -20.08 -6.79*** OTHERLIB 0.16 1.97* -0.15 -2.72** 1987 5.12 1.53 -4.18 -1.84 1985 -1.69 -0.53 -1.14 -0.53 1983 9.24 2.83** -1.13 -0.51 1981 2.20 0.68 0.37 0.17 N C L II . 114 r2 = .351 * p < .05 ** p < .01 *** p < .001

The model is more successful at explaining AG success when representing the conservative position than when

representing the liberal position. This is as might be expected, since I have been consistently more successful in explaining criminal success than non-criminal success, and SUPCONS is a variable which subsumes those cases (r = .616). As a result, the parameter estimates for the SUPCONS model 169

look very similar to those obtained in the OVERALL and CRIME models. Similarly, very few of the variables in the SUPLIB model reach conventional significance levels. However, to a certain degree I am gratified by this analysis, since my ideological independent variable (OTHERLIB) is associated with both dependent variables to a statistically significant degree in the directions predicted. The more that a justice supports the liberal side in non-AG cases, the more that the justice will support the AG's office when it represents that liberal position. And conversely, the more that a justices supports the liberal side in non-AG cases, the LESS likely that justice will support the AG's office when it represents the conservative position. Thus, there does appear

to be some association between ideological votes in the two categories of cases. And the fact that the other substantive variables perform as they have in earlier analyses while controlling on ideology suggests that those variables have importance above and beyond a particular justice's policy views.

Conclusion — In this chapter, many of the findings suggested in Chapter IV were given further support. As before, I was more successful at explaining variations in support for the AG in criminal cases than in non-criminal cases. Beyond that, AGs received more support from justices of the governor's party than from justices of their own party, and that their support level rises a small amount when both 170 the AG and governor belong to the same party. Further, an AG's rate of participation before a court was positively related to their success rate on the individual level. However, contrary to expectations, appointed justices were no more likely to support the AG than elected justices. Finally, these effects remained even with a control for ideology. I ought to stress, however, that the importance of many of these variables was only evident in the multivariate analysis. Due to low rates of dissent, it is virtually impossible to discern any statistically significant differences in the voting behavior of individual justices on many state courts of last resort. In such states, the influences of things like how a judge was selected on the subsequent behavior of judges appear to be practically non­ existent on the individual level. It is now time in the analysis to shift gears to some extent. While some hypotheses have received support, others have not. The failure of certain variables to perform adequately may partially be a result of the diffuse nature of support that has thus far been investigated. To this point, I have focused on all cases in which the attorney general's office provided legal representation — regardless of whether the attorney general or even the executive branch was actually involved.8 For reasons explained more fully in the next chapter, my models may be more successful explaining outcomes in cases in which the incumbent governor or incumbent attorney 171 general is actually named as a litigant. It is to discussing and analyzing these cases that I now turn. 172

ENDNOTES

1. Naturally, the breakdown of the vote in important decisions is likely to be scrutinized by interest groups, the media, politicians, and other observers to try to predict how the court will vote on similar issues in the future. But the widow contesting her eviction notice just wants a win. 2. And "high" is a relative term — even the SSCs with some of the "highest" dissent rates are low compared with dissent on the U.S. Supreme Court.

3. It is possible, of course, for different justices to dissent in different cases, and yet still have identical support scores.This is one reason why I have examined criminal and non-criminal cases separately, and why later in this chapter, I will examine liberal and conservative cases separately.

4. For example, if one appointed justice supported the AG 10 times out of 20, and another supported the AG 20 times out of 20, overall support would be 75%. This also serves to make the presentation of the data easier.

5. I made one last attempt to find a consistent relationship between judicial selection and support for the attorney general. A justice is screened by the governor before being appointed, but once a justice joins the bench, the actor to which he or she responds is the attorney general. Thus, I compared those justices who were appointed by a governor of the same party as the attorney general with other justices. Again, few differences were found. 6. Including the California dummy variable did not produce as dramatic of increases of r2 in the individual level analysis as in the collective analysis, but for the sake of consistency was left in.

7. In the model that includes CRIME as the dependent variable, each of the dummy time-point variables reaches statistical significance. I re-ran this model while excluding those time point variables, and the reduction in r2 was minimal (from .473 to .432), and the performance of the remaining variables was not significantly affected. 173 8 . There were certain cases in which the AG represented members of the legislative or judicial branches of government. However, these cases were indeed rare. More importantly is the fact the governor was rarely personally named as a litigant in those cases involving the executive branch. CHAPTER VI THE GOVERNOR AND ATTORNEY GENERAL AS LITIGANTS

In the preceding two chapters, a fairly successful model of attorney general success was developed. Many hypothesized relationships were supported, but some received little support or were found to operate counter to expectations. As already suggested, I would argue that the lack of support for certain variables is due, at least in part, to the fact that including ALL cases in which the AG provided legal counsel to a litigant provided too insensitive an indicator of the dynamics which may be present in particular cases. For reasons explained below, my model should receive greater confirmation in cases in which either the attorney general or the governor is individually named as an actual party to the case. Essentially, I included a number of variables to attempt to account for structural differences between the state attorney general and the U.S. Solicitor General. Nonetheless, the differences remain profound enough to make focusing on the AG's entire caseload too broad. The specific reasons for this conclusion are as follows: 1) The Attorney General's Workload — "Prior to the 1970s, state attorneys general tended to look upon their role

174 as being merely ministerial functionaries of the state administration; they were in office to do the bidding of other political executives and defended the state's legal estab­ lishment from attack” (Morris 1987, 299). While it is true

that state AGs offices are more vital today than in the past (i.e. Silbey 1980/81), Morris' description appears to still hold true in the vast majority of cases that AGs pursue all the way to their state's court of last resort. Before state high courts, the AG's office is largely reactive rather than proactive. As shown in Chapter IV,

attorneys general represented the respondent in 77% of all cases in which they participated. By contrast, Salokar reports that the U.S. Solicitor General's office represented the respondent in only 42% of the cases in which it participated between 1959 and 1986, and that percentage has been decreasing in recent years (Salokar 1992, 125). The highly reactive nature of the AG's office is problematic, because our depiction of judges as goal-oriented policymakers assumes that judges are responding to political stimuli. However, in most cases involving the AG, this is not the case. When the AG's office represents the petitioning party, it is in control of the litigation and has greater discretion and ability to argue legal positions which more closely represent the policy agenda of the executive branch than when it responds to appeals pursued by others. Hence, U.S. Supreme Court justices have greater opportunities to respond to the

175 176 Solicitor General in a manner that is political or policy- motivated than the typical state supreme court justice will have to respond to the AG in a similar manner. Other evidence lends credence to the thesis that looking at an AG's overall caseload may not be a sufficiently narrow focus. While the number of cases was insufficient to draw firm conclusions, Chapter IV also revealed that the ideological stances taken by Democratic and Republican attorneys general in particular states did not vary significantly. If we attempt to look for grand aggregate shifts in the state's legal policy when an incumbent of one

party relinquishes his or her office to an incumbent of the opposite party, we are almost certain to be disappointed. A primary cause of this was discussed in Chapter I. It demonstrated that an AG's caseload in each state tends to be highly repetitive. On the one hand, I argued that this ought to increase an AG's rate of success because it would increase

the office's experience in handling that type of litigation. However, while this helps explain why AGs tend to have high rates of success, I would argue that to a great degree, this

moves the focus in these cases from the identity of the attorney general, to the subject matter of the case. For

example, a Republican (George Deukmejian) occupied the office for the first two years of the study period, and a Democrat (John Van de Kamp) occupied the office for the latter three years of the study period. However, the nature of the office 177 compelled both of them to defend criminal convictions (and hence, take a "conservative" position) in the vast majority of cases in which they participated. They had no choice but to defend these cases. Further, the positions of the justices regarding criminal justice matters during this period have been much discussed and analyzed (see Chapter III) . It would be difficult to argue that the California Supreme Court during this period was responding to the attorney general per se. The influence of the nature of the case was the predominant force in the vast majority of cases in which the AG participated. Similar, if less dramatic, conclusions can be drawn about AGs in other states.

However, there is a category of cases in which the identity and policy aims of the attorney general should make a greater difference, namely, those cases in which the incumbent attorney general is named as an actual party to the case. I argue this for two reasons. First, these cases are much more likely to go directly to the heart of the AG's ability to wield power. In many states, "the common law is the fountainhead of the Attorney General's authority to represent, defend, and enforce the legal interests of state government and the public" (Ross 1990, 27) . When the attorney general is named as a party to a suit (whether as petitioner or respondent), the authority of the office is directly called into question. Certainly the policy that is involved in the particular case will be of importance. But in these cases the 178 justices will be forced to more closely focus on the con­ sequences of their decision for the office of the AG itself than in cases generally. Second, many of these cases arise as a result of AG initiative — through their powers to pursue consumer protection, securities regulation, antitrust, or other affirmative legal actions. While I have no data to confirm the notion, it seems to me that an attorney general's policy agenda is most likely to be revealed in his or her selection of which cases to proactively pursue and appeal to the state's highest court.1 State AGs must decide how to utilize the finite resources of their offices. The allocation of these resources should in part depend upon the policy predilections of the incumbent. Reflecting upon his decision to run for the office of attorney general in Arizona, Bruce Babbitt said that he grew tired of seeing the AG's office pursue suits designed to disadvantage American Indians in the state, and he wanted to utilize the power of government to help this and other dis­ advantaged groups. These cases may not be a perfect

representation of an AG's policy agenda, but they are surely

more refined than their caseloads in the aggregate. Hence, they are more "political" and consequently more sensitive to the political considerations discussed in previous chapters. 2) The Independence of the AG from the Governor — Each of the previous two chapters found support for the notion that members of the governor's party tended to support the attorney 179 general more than members of the opposite party. Curiously, no similar relationship was found regarding partisan compatibility between justices and AGs. Variables regarding both actors were included to account for the fact that AGs are independently elected in most states, and thus there was the need to keep the two offices distinct in the analysis. But it is that very independence that makes the finding just mentioned counterintuitive. Since it is the attorney general in most states that ultimately controls the conduct of litigation, it seems surprising that the partisanship of the governor appeared to matter more than the partisanship of the AG.

However, this problem of an indirect linkage between a justice and a governor through the attorney general can be virtually eliminated by focusing solely and separately on those cases in which the governor is named as a party. In most legal actions coming before a state court of last resort, the governor is not likely to have a specific position (beyond, perhaps, a general campaign pledge to be "tough on crime" or the like) .2 When a governor is named as a litigant, however, there can be no mistaking which side the governor would prefer to see prevail. And again, as in cases directly involving the AG, the stakes in cases such as these are likely to be high. These, then are the best cases in which to look for judicial reaction to the incumbent executive. 180 This chapter will first analyze cases directly involving the attorney general, and then turn to cases directly involving the governor. The cases will be generally summarized, so that we will have a better idea as to the types of cases in which executives appear as litigants. Some of the more interesting cases will be described in some detail. I will then turn to a PROBIT analysis of the outcomes of these cases, utilizing a model similar to that developed in the previous chapters, with appropriate additional variables included. CASES INVOLVING THE ATTORNEY GENERAL Altogether, the twelve state attorneys general were

individually named as a party to a case (either by themselves or with others) in 40 cases, which represents a mere 1.87% of the 2,140 cases in which their offices participated. Furthermore, as reported in Chapter IV, they won 26 of these cases, lost 10 of them, with a mixed outcome in 4 cases, for a winning percentage of seventy percent. This is quite comparable to their winning percentage in all cases (67.6%). Generally, these cases do not tend to generate dissent on state high courts. Dissent, whether in whole or in part, occurred in only 8 cases (20%), somewhat less than dissent rates in all cases involving the AG. As the number of cases with dissent is so low, we may briefly examine each in­ dividually. First, however, a few words should be said about the unanimous cases. 181 Unanimous Cases — As noted in previous chapters, the dissent rates on a number of courts are rather low. Therefore, as we might expect, unanimity in cases involving the attorney general does not necessarily imply a court dominated solely by one political party. AGs tended to win more often in front of courts controlled by their own political party — of their 21 unanimous wins, 17 (81%) came before courts controlled by the political party of the incumbent attorney general. However, they were also handed unanimous defeats by such courts. For example, the Tennessee Supreme Court was dominated by Democrats during the entire study period, but they handed the Democratic attorney general a number of unanimous losses. Non-unanimous Cases — Cases decided unanimously may be

a result of an institutional norm that discourages dissent (i.e. Walker, Epstein, and Dixon 1988). It may also be partially due to the fact in certain cases, the law strongly compels one outcome or another (i.e. Leonard 1989). Whatever the case, scholars have long recognized that the influence of

personal and political factors on judicial behavior are most clearly observed in cases in which there is dissent. Each of the seven3 cases involving the AG that were decided with dissent, and the voting patterns in those cases, are discussed below. If the identity of the incumbent attorney general is an important consideration in judicial voting behavior, we would hope to find that 1) if there is multiple dissent, the 182 dissenters will belong to the same political party and 2) the dissenters will belong to the opposite party of the attorney general (because the AG won each of the seven cases described below). The seven cases, briefly, are: — Amoitheater United School District #10 v. Harte4 — In this 1981 Arizona case, the Republican AG Robert Corbin brought a civil-rights action against a school district. The school district argued that under Arizona law [A.R.S. 41- 1481(D)] the attorney general was prevented from bringing an

employment discrimination suit against any state agency. The court, with a Republican majority, disagreed and Cameron won the case, 3-2. However, the dissenters were not of the same

political party, and the majority opinion was also joined by members of both political parties. Although there are civil- rights overtones to the case, the opinions tend to focus on

the relatively technical question of the whether or not the school district was indeed a state agency under the statute in question. This may partially explain the lack of clear partisan divisions.

— People ex rel Deukmeiian v Brown5 — This case goes directly to the question of the powers of the attorney

general. In 1977, the Democratic attorney general had advised Democratic Governor Jerry Brown to sign into law the State Employer-Employee Relations Act of 1977. In 1979, two organizations filed suit to obtain a writ of mandamus to compel Governor Brown "to perform their constitutional and 183 statutory duties without regard to provisions of SEERA, contending the legislation was unconstitutional” (29 Cal.3d 150, at 154) . They were joined in their suit by the new Republican attorney general George Deukmejian. The impetus for this case, then, was the fact that the position of the AG's office had changed as a result in the change in the identity of the incumbent. As the court put it, The issue then becomes whether the Attorney General may represent clients one day, give them legal advice with regard to pending litigation, withdraw, and then sue the same clients the next day on a purported cause of action arising out of the identical controversy. We can find no constitutional, statutory, or ethical authority for such conduct by the Attorney General, (at 155) As the quote suggests, the court ruled in Governor Brown's favor, 7-2. As might be expected on this highly partisan court, the vote divided exactly along party lines, with Republican justices Frank Richardson and William Clark providing the only votes for Deukmejian. It is difficult to

know whether the important factor was political party or the identity of the appointing governor, since either Brown or his father had appointed each of the Democratic justices on the court at that time. — Cales v. Webster6 — this 1987 Missouri case was brought by a bailiff who was being sued in a wrongful imprisonment action. The bailiff sued Republican AG William Webster, who had refused to provide legal representation for the bailiff. Five justices voted in favor of Webster, one justice dissented, and another dissented in part. The 184 dissenter (Warren Welliver) was a Democrat, but the justice that dissented in part (Charles Blackmar) was a Republican. It is true that decisions that appear to be based on technical questions can serve to mask ideological motives. However, the opinions in this case focused on whether or not the bailiff was acting as an agent of the state (finding that, because he was paid by the county rather than the state, he was not) . And the voting patterns suggest this was not an ideologically divisive issue. — Menorah Medical Center v. Health and Educational Facilities Authority7 — this 1979 Missouri case is one in which Republican AG John Ashcroft defended the constitutionality of a state law regarding state aid to educa­ tional and medical facilities from an equal protection challenge. The opinions in the case focus solely on the equal protection claim and do not mention the role of the attorney general in the case. It may be as a result of this that again, the dissenters in this case came from both political parties: two Democrats and a Republican. — Paul Kimball Hospital v. Brick Township Hospital8 — in this 1981 New Jersey case, Democratic AG James Zazzali defended the constitutionality of the state's tax code from equal protection charges. The dissenters in this 5-2 decision included one Democrat and one Republican, but the Republican justice (Robert Clifford) had initially been appointed by Democratic governor Brendon Byrne. Again, the opinion 185 stressed the technical nature of the case (was a "grandfather clause" in the law constitutional?) rather than focusing on the attorney general's powers.

— State ex rel Celebrezze v. K & S Circuits. Inc.8 — operating under his powers to prosecute violations of environmental laws, Democratic AG Anthony Celebrezze sued a manufacturer for water pollution in this 1983 Ohio case. The company admitted liability, but was denied a jury trial on the civil penalty to be paid. The voting patterns in this case strongly indicate a partisan division on the court — the court's lone Republican on this seven-person court was also the lone dissenter in the case. A Democratic AG was bringing a "liberal" case before a court dominated by Democrats, and the Republican refused to acquiesce. — Osborn v. Shi1linger9 — Democratic attorney general

(and former state supreme court justice) A.G. Me Clintock successfully opposed a condemned man's final request for a stay of execution in this 1985 case. The one dissenter in this case was Democrat Robert Rose. The court's other two Democrats and one Republican voted to deny the stay. In looking for political responses to the attorney general, the pattern we would wish to see — a court clearly divided along partisan lines — occurs in only two of the seven cases (Deukmejian and Celebrezze). It is interesting to note, however, that these are the only two of the seven cases in which the AG is named as sole litigant. This seems to be 186 an important distinction between the cases above — those cases in which the AG was named as sole litigant tended to directly concern the powers of the AG's office. Those cases in which the AG is only one of a number of litigants tended to turn on the substance of the case as it affected the other litigants. This will be a factor to be considered in the multivariate analysis to follow. No stark patterns emerge from these seven cases. Support for the AG may or may not be dependent upon partisan considerations. In order to test more comprehensively, a more rigorous statistical analysis will be performed. I will utilize the case as the unit of analysis, with a dependent variable to indicate whether or not the AG won the case in which he or she was personally named as a litigant. The model to be estimated is: WIN = b0 + bjPOWER + b2RESPOND + b3AGADVAN + b4AGGOV + b5INITIATE + e where WIN = 1 if the AG won the case 0 if the AG lost the case POWER = Boersema's overall AG powerrating RESPOND = 1 if the AG was the respondent 0 if the AG was the petitioner AGADVAN = The number of seats above a majority the AG's political party controls on the court AGGOV = 1 if the AG and governor belong to the same party 0 if otherwise INITIATE = 1 if the case was initiated by action on the part of the attorney general's office 0 if otherwise 187 Each of the independent variables is hypothesized to be positively associated with the dependent variable. The final independent variable (INITIATE) is included to represent the fact that cases which are initiated by the attorney general's office are more likely to be more personally important to the AG than other cases. They are also more likely to represent the AG's policy agenda than cases such as Osborn. in which the AG is compelled to respond to actions commenced by others. Since the dependent variable in this case is dichotomous • rather than continuous, ordinary least squares regression is not appropriate (Aldrich and Nelson 1984). Thus, I will utilize probit, a maximum likelihood estimator, to generate parameter estimates for this model.

The results of estimating the model of AG support using probit appear in Table 27a. The results are disappointing. The overall model is not significant, even at the .10 level. Furthermore, only 66.6% of all cases were correctly predicted by the model. If we had only predicted that the AG would win all cases in which he or she was personally involved, we would have been correct 70% of the time. Hence, there is no reduction in error. Further, none of the substantive variables reaches any conventional levels of statistical significance. This lack of findings may be partially attributable to the small sample size (n = 30). In the previous chapters, I had found that AGs tended to receive more support from members of the governor's party than 188 from members of their own party. While the governor is not involved in the vast majority of the cases presently under analysis, this was also true of most cases generally. Hence, I estimated an alternative model, one which substituted GOVADVAN for AGADVAN in the equation shown above. The results of this probit analysis are contained in Table 27b. This variable works no better. The explanatory power of the model is not improved, nor is the performance of any of the other substantive variables. Neither model adequately accounts for AG support in these types of cases. Hence, it is time to turn our attention to cases involving the governor. SUPPORT FOR THE GOVERNOR AS LITIGANT Governors were named as litigants in 31 cases falling under the scope of this study. While they had a winning record, their success rate was not as high as that enjoyed by attorneys general. As was the case with the attorney general, there appears to be somewhat of a partisan element to support for the governor: Eleven of the governors' 16 wins came before courts dominated by members of their own political party; only four came before courts dominated by members of

the opposite party. Conversely, governors were more likely to lose in front of courts dominated by members of the opposite party (3 wins, 7 losses, and 2 unclassified). As was true in cases involving the AG, most of these cases were decided unanimously: only 9 of these 31 cases were decided with dissent. Again, this is a small enough number of 189 TABLE 27a — PROBIT ESTIMATION OF MODEL EXPLAINING AG SUCCESS b Std. Error t Constant -4.26 4.80 -0.89 RESPOND 0.59 0.62 0.96 POWER 0.06 0.06 0.88 AGADVAN 0.08 0.17 0.47 AGGOV 0.66 0.59 1. 12 INITIATE -0.75 0.79 -0.95 Mean of the DV = 70.0 Percent predicted correctly = 66.6% Approximate chi-square =4.39 Pseudo r2 = .120 N = 30

TABLE 27b — PROBIT ESTIMATION OF MODEL EXPLAINING AG SUCCESS (revised) b Std. Error t

Constant -4.65 4.88 -0.95 RESPOND 0.59 0.62 0.95 POWER 0.06 0.06 1.01 AGADVAN 0.02 0.16 0.12 AGGOV 0.54 0.59 0.91 INITIATE -0.89 0.73 -1.21 Mean of the DV = 70.0 Percent predicted correctly = 66.6% Approximate chi-square = 4.18 Pseudo r2 = .114 N = 30 190 cases that we can consider them briefly to see if voting behavior conforms to partisanship or method of ascension to the bench:

— Rock v. Thompson10 — in this 1981 Illinois case, two state senators filed suit requesting the court to issue a writ of mandamus compelling Republican Governor James R. Thompson

"to convene the Illinois Senate to elect from its membership a President of the Senate as presiding officer" (85 111.2d. 410, at 411). It seems that the state senate had convened briefly and had elected a president apparently in violation of senate rules before adjourning. The two senators wanted a new election to be held according to senate rules. Gov. Thompson lost on a 4-3 vote. Despite the apparently political nature of the case, the case was not decided on a party-line vote: one Democrat joined two Republicans in dissent, and one

Republican joined three Democrats in the majority. Illinois is the only state in this study in which the governor has no role in the selection of justices, and this may account in part for the lack of partisan voting behavior in this case. Warrior v. Thompson11 96 111. 2d 1 (1983) — lower-status citizen from the state of Illinois sued Governor James R. Thompson in order to receive payment of medical expenses, in a case that seems to embody clear ideological elements. The voting pattern is generally partisan, with all of the courts'

Republicans voting to uphold the Republican governor, while 191 both of the dissenters were Democrats. However, two of the court's Democrats also backed Thompson. Gilbert v. Gladden12 — This case is unusual, in that the governor appears at his own request. This case clearly goes to the heart of the governor's veto power. In New Jersey,

"there has developed an unofficial custom of long duration, known as gubernatorial courtesy, whereby bills that have been passed in both houses of the legislature are not presented to the Governor for signature or veto until the Governor requests them" (at 279). Under the New Jersey constitution, the governor has no "pocket veto", but this tradition of gubernatorial courtesy essentially had the effect of giving him one de facto. Four members of the state legislature, along with others, sued to end this tradition and require the legislature to present bills to the governor in a timely manner. The Court found in favor of the governor. The majority declared that this was a political question for the legislature to resolve and hence not appropriate to judicial alteration or cessation. Contrary to expectations, both dissenters were Democrats, like the governor. Most of Byrne's appointees stuck by him, including his one Republican appointee, Robert Clifford. However, the dissenters did include one of Byrne's appointees, Sidney Schreiber. Hence, the effect of appointment in this case is not entirely clear. 192 Worthington v. Fauver13 — to alleviate overcrowding in state prisons, Gov. Byrne issued Executive Order 106, in which he invoked his emergency powers under New Jersey's Civil Defense and Disaster Control Act to permit the dispersement of state prisoners into county prisons. Various counties sued to have the Order overturned on separation of powers grounds. By a 6-1 vote, the court ruled that "the Governor's Order is authorized by the Disaster Control Act and does not violate the constitutional principle of separation of powers" (at 185) . The only dissenter belonged to the same political party as the governor.

Clark v. Cuomo14 — Democratic Governor Mario Cuomo of New York "implement(ed) a plan to facilitate the registration of potential voters by making registration forms and assistance available at state agencies" (at 186). This expanded form of the "motor voter" program was issued by executive order rather than by the state legislature, and was hence challenged in court by a former Republican state chairman. Although the opinion focuses on the separation of powers issues raised by the plaintiff, both the context and the identity of the litigants suggest that this cases has significant political aspects as well. However, the voting patterns do not reveal a strictly partisan division. A fellow

Democrat dissented in part, but two Republican justices supported the governor. However, one of these Republican justices (Richard Simons) was appointed by Cuomo, and the 193 other (Sol Wachtler) was appointed by Cuomo to the office of chief justice. Higher Education Facilities Authority v. Gardner15 — in this 1985 Washington case, parochial schools attempted to compel Democratic Governor Booth Gardner and his Lt. Governor to "sign resolutions issuing bonds for the benefit of private- related educational institutions" (at 839). These private institutions included a number of religious schools, and thus Gardner opposed the issuance of the bonds on the basis that it would violate separation of church and state provisions of the Washington constitution. We would expect that when a Democratic governor defends a "liberal" policy such as this, partisan divisions on the court should be high — Democrats should support the governor, Republicans should oppose him. However, this does not occur. Gardner lost in a 6-3 decision, but both the dissenters and the majority were composed of members from both political parties. In fact, two of the three dissenters voting to uphold the governor's position were Republicans. Washington Public Employees Association v. Spellman16 — in this case, a Washington state employee's union brought a challenge to a piece of legislation that prohibited lump-sum payments in lieu of accrued vacation time, and named Republican Governor Spellman as a defendant. Again, a

Republican governor upholding a seemingly conservative position would seem likely to produce partisan divisions on a 194 court. However, again this is not the case. The court was composed of 5 Republicans and 4 Democrats, yet the decision was nearly unanimous (8-1) in opposition to the governor's position. (However, consistent with expectations, the lone dissenter was a Republican.) Luther v. Rav17 — this final case directly involves gubernatorial powers and has partisan dimensions. Mr. Luther was a member of the Board of Pilotage Commissioners, a state commission that regulates and licenses people serving as pilots in Washington state. He was initially appointed and then re-appointed by Republican Governor Daniel J. Evans. However, when his Democratic successor, Dixie Lee Ray, attempted to remove Mr. Luthor from the Commission, he sued, claiming that the governor was without authority to shorten the terms of gubernatorial appointees. Once again, partisan divisions are not as pronounced as we might expect. The court decisively (8-1) sided with Gov. Ray. The lone dissenter was a Republican, as expected.

From an examination of these cases individually, we are hard pressed to find much evidence of loyalty or disloyalty to the governor along party lines, even in those cases in which considerations of partisanship or gubernatorial power are evident. However, they were presented not as rigorous evidence, but to merely give examples of cases which animate dissent, and to present a description of the voting patterns in those cases. In order to more rigorously test these 195 hypotheses, a model of support for the governor, similar to those already developed, will be tested: WIN = b0 + bxPOWER + b2RESPOND + b3GOVDVAN + b4AGGOV + b5SOLO + e where

WIN = 1 if the governor won 0 if otherwise POWER = Boersema's AG power rating RESPOND = 1 if the governor was respondent 0 if the governor was petitioner GOVADVAN = the number of members of the governor's party above a majority seated on the court AGGOV = 1 if the AG and governor belong to the same party 0 if otherwise SOLO = 1 if the governor is named as sole litigant 0 if other litigants are named With the exception of SOLO, each of these variables should be familiar by now. Even though in this analysis I am considering the governor rather than the attorney general, two variables relating to the AG (POWER and AGGOV) were retained in this model. This was done because even though these cases involve the governor, in all but the most unusual of circumstances18 it is the AG's office that must actually represent the governor in court. Hence, the effects of the participation of that office must be considered. The SOLO variable is included as a kind of surrogate for the INITIATE variable included in the previous analysis. In many of these cases, the governor was simply one litigant named among many. When this occurred, it appeared from the 196 reading of the opinions that the governor was not closely connected with the litigation at hand. By contrast, most of the cases in which the governor was named as sole defendant did indeed appear to more directly impinge on the powers of the governorship itself, and hence ought to be more important. As before, since the dependent variable is dichotomous

(the decision in the individual case), probit analysis will be utilized rather than ordinary least squares regression. Table 28a shows that this model is somewhat more successful in predicting gubernatorial success than AG success. The fit is better, and the overall model is significant at the .10 level and nearly approaches the .05 level. Further, the 69.2% of cases that were accurately predicted indicate a small reduction in error. As for the substantive variables, it does appear that the more members of a governor's party on a court, the more likely it is that the governor will win the case — paralleling the findings of the previous two chapters. Contrary to expectations, governors seem less likely to win when they are represented by an attorney general of their own party. In the exposition of this dissertation, judicial appointment was hypothesized to be positively related to support for the executive. Table 28b contains the parameter estimates for the model in Table 28a, but with the variable

SAMEGOV = the number of justices on the court appointed by the current governor 197 TABLE 28a — PROBIT ESTIMATION OF MODEL EXPLAINING GUBERNATORIAL SUCCESS

b Std. Error t

Constant -7.21 5.29 -1.36 RESPOND 0.62 0.63 0.98 POWER 0.10 0.07 1.50 GOVADVAN 0.64 0.26 2.51** AGGOV -2.50 1.25 -2.01 SOLO -0.11 0.67 -0.17 Mean of the DV = 57.7 Percent predicted correctly = 69.2% Approximate chi-square = 10.72 (p<.10) Pseudo r2 = .303 N = 26 ** p < .01

TABLE 28b ~ PROBIT ESTIMATION OF MODEL EXPLAINING GUBERNATORIAL SUCCESS (revised) b Std. Error t

Constant -4.36 5.03 -0.87 RESPOND 0.70 0.61 1.15 POWER 0.04 0.06 0.67 SAMEGOV 0.43 0.22 1.93 AGGOV -0.24 0.83 -0.28 SOLO 0.04 0.06 0.76 Mean of the DV = 57.7 Percent predicted correctly = 65.3% Approximate chi-square = 8.59 Pseudo r2 = .243 N = 26

* p < .05 198 substituted for GOVADVAN. This variable attempts to go beyond pure partisan kinship and more directly test the effects of judicial appointment. As Table 28b, reveals, the change does not significantly alter the results. The new variable is also statistically significant, although it is somewhat weaker, as accuracy of prediction decreases. However, all of these parameter estimates — both those which support and contradict expectations — must be consumed with the awareness of the small sample size (n=26). CONCLUSION Proceeding on the notion that justices' reactions to the executive branch would be more sensitive to the stimuli as described in the first two chapters in cases directly involving members of the executive branch than in other cases, this chapter focused on the former cases exclusively. It was found that both AGs and governors won the majority of their cases, but that AGs tended to be slightly more successful than governors. Further, although none of these courts was staffed exclusively by members of one political party, the great majority of cases involving AGs and governors tended to be unanimous, reflecting the overall norm of unanimity on these courts. A closer, descriptive examination of the non- unanimous cases revealed that in most instances, the voting divisions in cases involving the AG and/or governor did not clearly reflect partisan differences on those courts. A model of AG and gubernatorial success was formulated and estimated utilizing probit, with little success. The only finding of any significance was that governors do tend to win more when they have more of their own appointees, or more members of their political party on the court. 200

ENDNOTES

1. As discussed in Chapter 3, there are cases in which an attorney general must occasionally defend policies which he or she does not personally support. To control for this somewhat, the following multivariate analyses include a dummy variable for legal actions initiated by the AG's office, which, presumably more closely reflect the policy goals of the incumbent.

2. We cannot even assume that he or she would always want the executive branch actor to win, because that actor may be pursuing a policy that is against policy of the administration.

3. One "mixed" case was decided with dissent,, but since neither the majority or the dissenters gave the AG a clear victory, I won't describe it here. 4. 624 P.2d. 1281 (Arizona 1981) 5. 29 Cal 3rd. 150 (1981) 6. 727 SW 2d. 901 (Mo. banc 1987) 7. 584 S.W. 2d. 73 (Mo. 1979) 8. 86 NJ 429 (1981) 8. 6 Ohio St. 3d. 354 (1983)

9. 705 P.2d. 1246 (Wyom 1985) 10. 85 111. 2d. 410 (1981) 11. 96 111. 2d. 1 (1983) 12. 87 NJ 275 (1981) 13. 88 NJ 183 (1981) 14. 66 NY 2d. (1985)

15. 103 Wn. 2d. 838 (1985) 16. 98 Wn. 2d. 677 (1981) 201 17. 91 Wn. 2d. 566 (1979)

18. As mentioned already, on some occasions the attorney general may be the actor or may be representing the actor who is opposing the governor in court, but these cases are rare indeed. CHAPTER VII CONCLUSIONS AND AGENDA FOR FUTURE RESEARCH

The success of any litigant in court depends upon a host of factors. Many of these factors also condition the fate of the state when it participates in a judicial proceeding. The state, however, whether in the form of the executive or legislative branches, brings with it a unique and interesting set of characteristics which may exert effects on judicial behavior that are independent of other influences. While this dissertation has not attempted to provide a comprehensive account of governmental success in court, it has attempted to identify some of these factors and test them empirically. In so doing, I have taken into account both variables that have been previously found by other scholars to influence judicial behavior, and factors relating to the unique situation of the dual executive in the American states. Below, I shall comment on the main conclusions of my analysis. I shall then offer general comments regarding the comparability of conducting judicial research at the federal and state levels. I shall conclude by offering suggestions for further research.

202 203

CONCLUSIONS First, by virtually any measure, the state attorney general is a successful litigant. State attorneys general virtually always win at least half of the cases in which their offices provide legal representation. In the aggregate, the success of the attorneys general in this study rivaled the success of U.S. Solicitor General. They were also found to be successful in relative terms, winning more freguently than other, similarly-situated litigants. Regarding specific contours of AG success, they won more freguently in criminal than non-criminal cases, and when they represented the respondent rather than the petitioner. These findings were consistent with expectations and the previous scholarship. However, as I have argued, work on the Solicitor General has already done a good job of documenting executive success in court, while failing to sufficiently specifying reasons for that success (Segal 1993). Turning now to my specific hypotheses explaining AG success:

Repeat Plaver Status — In conceptualizing the AG as repeat player, I had argued that the RP's advantages would be related to: 1) the freguency with which the AG appeared before the court, resulting in more expertise and other assets, and 2) the ability of the AG's office to set its own legal agenda. Support was found for the first proposition, but not for the second. The larger proportion of a court's docket an AG's caseload consumed, the higher the AG's success 204 rate in criminal cases and cases overall (the same did not hold true for non-criminal cases). Naturally, this variable is not a perfect surrogate for all the dimensions of RP advantages, but it appears to have been a meaningful one. Contrary to expectations, the other dimension of "repeat playerness", as measured by the index created by Craig Boersema, was inversely related to AG success. This was consistent with Boersema's own research. However, Boersema suggested this was due to the fact that weaker AGs appeared in a higher percentage of cases in which they were the appellee attempting to retain a criminal conviction, two factors associated with AG success. However, even after I controlled on these variables, Boersema's index is consistently and negatively associated with AG success. This strongly suggests another measure would be useful to capture this dimension. The Influence of the Attorney General vs. the Governor — directly measuring the "influence" of an executive branch official on the judiciary is difficult. In this study, however, I found that attorneys general received increased support from justices of the governor's political party, but did not receive increased support from justices of their own party. Similarly, the more members of the governor's party that sat on a court, the more success the AG enjoyed; no similar relationship was found regarding members of the AG's party. This finding was consistent in virtually every 205 analysis, including the analyses conducted at the individual case level. This was initially surprising. Although I had clearly argued that the dual nature of the executive branch at the state level needed to be accounted for conceptually, I thought that the party of the attorney general would be more important than the party of the governor. This is because, as discussed in Chapter II, the state attorney general is typically far more independent of the governor than the U.S. Solicitor General is of the president. None of the AGs' offices which returned my questionnaire indicated that the governor had a very strong influence over formulation of legal strategy or decisions to pursue appeals (see also Thompson 1974). From all accounts, the attorney general is divorced from the governor in most states. Since it is the attorney general who is more closely connected to the litigation that serves as the stimuli to the court and its justices, I had imagined that the AG's political brethren would cooperate more often. Nevertheless, the contrary finding was not totally unexpected and can plausibly be explained. First, although the state AGs have more independence from the governor in most states than Solicitors General have from the president, their ability to select cases laden with ideological import is far more limited. Considering the vast diversity of litigation involving the federal government (as well as the large number of cases in which it may participate as amicus), the U.S. 206 Solicitor General, in his "gatekeeper" role can take positions on a much broader range of issues than the typical state AG. As a consequence, the SG's caseload is much more "political" than the AG's. Further, since the SG's office has total control over its caseload, that "political" caseload is known to the Court to represent the agenda of the SG (and consequently, the president). Things are much less clear at the state level. Although the AG is free from the control of the governor in most states, its relative lack of control over its own caseload undercuts the office's ability to act as an active agent of legal change. Naturally, there are a few cases which have been initiated by the AG and seem to reflect an attempt to change policy through litigation. However, like cases in which the AG actively opposes the governor, these too are rare (and my model was unsuccessful at explaining AG success in these cases) . For the most part, justices may fairly perceive AGs as dutiful agents of the executive branch, stressing (in the words of Michael Dukakis), "competence over ideology". If this is true, it should not be a huge surprise that justices of the governor's party give increased support to the state when it is represented by the attorney general. The governor is the chief executive. He or she sets the tone and agenda of the executive branch. The governor may not have a specific stance upon many, if not most, of the cases that involve the state. But many of the lawsuits in which the AG 207 participates are initiated by the governor's agents, acting upon the governor's policy agenda, whether explicit or implicit. If the AG's office defends these suits, it then be­ comes a mere conduit for the governor's position. Hence, although much has been made of the differences in the relationship between the chief executive and the chief legal officer at the federal and state levels, it appears that they may be more similar in practice than previously acknowledged. In spite of all of this, the attorney general still has an important role to play in the development of a state's legal policy. This project was only concerned with litigation before a state's highest court. However, the AGs office handles litigation in lower state courts as well, and the AG's office may greatly affect policy through these lower-court actions. For example, if an AG aggressively prosecutes antitrust actions, the business community will perceive it even if no cases ever reach the high court on appeal. Further, AGs in most states have the power to issue advisory opinions, which are taken seriously by legislators and others when drafting and implementing legislation. And finally, even though the AG's legal caseload is generally routine and involves little discretion, there still occur those exceptional, isolated lawsuits in which the AG's office establishes new judicial precedent which charts a new course for government or the state in some area. The AG's office provides plenty of opportunities for an ambitious state 208 politician to make his or her mark on policy and politics (Boersema 1987).

Judicial Selection System — This study takes its place in the pantheon of works failing to document any significant systematic effects attributable to the mode of judicial selection. Even after controlling on political party, I found no evidence for the proposition that judicial appointees sup­ port the executive branch at higher rates than justices elected separately. My failure is likely due to the same reasons that other studies have also failed, as discussed in the first chapter: due to the fact that many justices in elective system are initially appointed, combined with the fact that even appointed justices must face a retention election, the two types of selection systems are quite similar in practice. There is no need to linger on this point.

ON THE CONDUCT OF COMPARATIVE JUDICIAL RESEARCH One of the most reliable phenomena to emerge in this research was my consistent ability to better explain support for the attorney general in criminal cases than in non­ criminal cases. Upon reflection, it seems to me that this may be partially (if not predominantly) due to the fact that criminal cases are more directly comparable across states than non-criminal cases, for a number of reasons. First, in criminal cases there is more jurisdictional comparability across states. Whereas some courts have jurisdictions which are heavily discretionary while others are 209 highly mandatory, each state's court of last resort must accept appeals regarding capital punishment, thus making these types of cases somewhat more similar. Second, "criminal” cases usually present issues within a very limited jurisprudential range — largely the 4th, 5th, 6th, and 8th amendments to the U.S. Constitution, or their state constitution equivalents. Hence, these cases the legal stimuli that they represent, and the judicial responses that they evoke, are similar across states. "Non-criminal cases" is a much more amorphous category, representing a much wider range of legal issues and varying quite a bit more from state to state. For example, both the New Jersey and Ohio AGs had caseloads dominated by non-criminal cases. But the Ohio AG handled a large number of workers' compensation cases whereas the New Jersey AG handled very few. This problem could have been alleviated, of course, by creating issue categories that were more precise. For the purposes of this project, however, that was an impractical solution because of the vastly different caseloads of the various AGs offices. For example, comparing how different AGs fared in tax cases would not have been possible, because the number of tax cases handled in many states was so small as to make comparative analysis fruitless. Much of the work that has been done on judicial behavior on state courts has focused on judicial responses to the death penalty (i.e Hall 1992; Hall and Brace 1992). This is 210 certainly a strongly justifiable choice — these cases involve important values, and are fairly comparable across states. However, we must keep in mind that death penalty cases are a "special" example of litigation. They are highly visible, but they also make up only a small proportion of any state court of last resort's docket. To fully appreciate the dynamics that motivate judicial behavior, it is essential that we widen the scope of the inquiry. This dissertation has attempted to do so, but widening the scope of the inquiry simultaneously increases the danger of non-comparability.

In deciding upon a phenomenon to analyze, the researcher can opt to focus on variations in judicial support for a particular type of litigation (i.e. civil rights, obscenity and so on), or to explain variations in judicial support for a certain type of litigant, as this dissertation has sought to do. Focusing upon a particular type of litigation goes a long way to solving the problem of inter-state comparability. Focusing on a particular category of litigant may also have this effect. For example, litigation involving labor unions will quite frequently involve questions of labor relations and labor law.

However, choosing a government or governmental actor as the subject of study often provides no such focusing of cases, since the government is typically involved in many different types of litigation. While the model developed here was fairly successful at accounting for variations in one 211 homogenous class of cases (criminal cases), adequately explaining state success in non-criminal litigation would require a somewhat different approach. One would select a specific area of law, non-randomly include those states in which the AG participates frequently in this type of case, and increase the time frame of the study. Eventually, even "legal" variables could be included. However, as work has begun only in the last decade or so to attempt to construct such theories at the U.S. Supreme Court level, one must assume that, with the exception of the death penalty, similar research on state supreme courts will be long in coming. AN AGENDA FOR FUTURE RESEARCH Although we already knew that state governments fared well before their own state courts of last resort (Boersema 1987; Kagan et al 1987), this is the first work to attempt to systematically test hypothesized reasons for that success, with mixed success. Naturally, this dissertation is not the final word on the subject of executive - judicial relations in the states. There is much yet to be done.

One avenue of future research was suggested above. In addition, we know very little about the structure and internal functioning of state attorneys general's offices, including how cases are processed and decisions to appeal are made, and the AG's relationship with both the governor and the court regarding litigation. Such an account might be difficult to obtain (as suggested by my own experiences) , but it seems 212 essential to understanding the role of the state attorney general in the process. Another fruitful thing to do would be to drop the focus on the attorney general altogether, and include ALL cases in which the state participated. The AG's relative lack of discretion, coupled with the findings of this project, indicate that although the office of attorney general may be a "political" one, that political dimension does not clearly reveal itself in litigation. This would also be a manner in which to further test our finding that state litigation is conditioned by judicial responses to the governor. Since he or she is the head of the entire executive branch, focusing on all cases concerning that branch should produce similar results. If I were to continue conducting research on judicial - executive relations in the U.S. states, I would clearly de- emphasize the role of the AG and emphasize the role of the governor. And although there are certain structural differences which make the link between justices and governors less direct than the link between justices and presidents, that link can be more directly re-established by focusing on litigation directly involving the governor, as in Chapter VI. This model of gubernatorial success indicated that some of the hypothesized factors were at work, but the sample was too small and unrepresentative to state this conclusively. Conducting such an analysis would require focusing on more 213 cases from more states, a relatively simple data collection task with the use of LEXIS.

Although much of this dissertation emphasized the differences between the executive - judicial relations at the federal and state levels, Iam now struck by their similarities. It is true that many justices are elected rather than appointed, and the AG is a political actor independent of the governor in most states. However, evidence presented in this dissertation suggests that these differences are more evident in theory than in practice. The difference in behavior between judicial appointees and electees was slight. Similarly, the state AG generally works as the defender of the policies of the executive branch, of which the governor is head. At least with regard to litigation, the "dual executive" in the American states is usually not schizophrenic. APPENDIX A 215

LETTER AND QUESTIONNAIRE SENT TO STATE AG'S OFFICES

Dear State Attorney General, My name is James Brent, and I am currently a graduate student at the Ohio State University. I am in the process of writing my dissertation, which concerns the relationship between state attorneys general and their state courts of last resort. Essentially, I am attempting to explain why some state attorneys general have more success in court than do others. In order for me to perform this task meaningfully, it is necessary that I have some fairly detailed knowledge of the manner in which state attorneys general's offices operate. Unfortunately, very little has previously been written on this subject. I am therefore respectfully requesting your assistance. Enclosed you will find a brief questionnaire and a self- addressed stamped envelope. The questionnaire should take less than 20 minutes to complete. Some of the items are of the multiple-choice variety, while some of the others ask for a short written reply. You have my guarantee that all of your responses will remain COMPLETELY CONFIDENTIAL, and no quote will be directly attributed to you or your office. You are obviously not obligated to answer any questions which make you uncomfortable. However, I assure you that any information I receive will be used for academic purposes only, and I will be deeply grateful for as much assistance as you (or a knowledgeable member of your staff) can offer.

Yours is one of 12 states that I am including in my study. Your state was chosen because it fits certain criteria that are essential to the design of my research. Hence, for my purposes, any information which you may be able to provide will be extremely valuable (indeed, critical), and appreciated. In exchange for your cooperation, I will be happy to mail you a summary of the results of my study when I complete it later this year. If you have any questions, please do not hesitate to call me at (614) 292-2880. Thank you very much for your time. Sincerely,

James Brent, M.A. 216 When a state administrative agency or local prosecutor wishes to appeal a case to your state's court of last resort, is it required to seek the approval of your office? YES NO When a state administrative agency or local prosecutor wishes to appeal a case to your state's court of last resort, is it customary to seek the approval of your office? YES NO Approximately what percentage of all cases lost by the state in the lower court does your office appeal to the state supreme court?

%

When other governmental actors appeal a case, do they normally retain their own counsel, or do they normally receive legal representation from your office? ______All of them retain their own counsel ______Most of them retain their own counsel ______About half and half ______Most of them are represented by my office ______All of them are represented by my office

When deciding whether or not to appeal, what are the main criteria you and your office utilize in making your decision? Are there certain things you look for in a case (i.e. the amount of money involved, the chances of winning, etc.) that increase the likelihood of pursuing a case further?

How much discretion would you say that you and your office have in deciding which cases the executive branch will appeal to the supreme court? total discretion — all agencies must get our approval a great deal of discretion a fair amount of discretion a small amount of discretion no discretion — no agencies must get our approval 217

How much involvement does the attorney general personally have in the decision to appeal? In what percentage of cases is he or she personally involved? %

Any comments you wish to add about your office and the decision to litigate would be helpful.

Do you feel that the members of your staff have a fairly good idea of the thinking of individual justices on your state's court of last resort? Do you feel that you can predict their stances on certain issues? Does this knowledge on occasion influence your decision to litigate?

Do you feel that the justices on your state's court of last resort perceive your office to be more credible than most other counsel? Do you feel that they believe your briefs and arguments are more expertly crafted than that submitted by other litigants?

When it comes to making decisions about the conduct of the state's litigation, how much influence does the governor have over your office?

a great deal of influence a fair amount of influence a small amount of influence virtually no influence 218

Who has the most influence over the conduct of the state's litigation in your state — the attorney general or the governor? (Circle one) Are there any instances in which the governor's judgment overrides the attorney general's when they are in disagreement? YES NO

If so, what might some of those instances be?

In what ways does the governor communicate his desires about the conduct of litigation to your office? Formally or informally? On the phone, in person, or on paper? Frequently or not?

Does your office feel it owes a greater duty to serve the governor as a member of the executive branch, or the supreme court as an officer of the court?

How much pressure does your office feel to bring its legal agenda into line with the governor's policy and political agenda?

a great deal of pressure a fair amount of pressure a small amount of pressure virtually no pressure 219 What is the procedure that is followed when your office is asked to defend a state action with which the attorney general personally disagrees? Can you refuse to defend it? Will you give it to an assistant? or what?

To your knowledge, since 1979 has there been any major change in the amount of control over litigation that your office enjoys? (This change could be a result of changes in either the constitutional, statutory, or common law, as well as changes in the internal operating procedures of your office.) If so, what were these changes and when did they occur? THANK YOU VERY MUCH FOR YOUR ASSISTANCE! APPENDIX B

220 221

CODING RULES I. CASES SELECTED FOR ANALYSIS As described in Chapter III, all cases decided in calendar years 1987, 1985, 1983, 1981, and 1979 with a written, published opinion more than one page long were coded. West's regional reporters were utilized for Alaska, Arizona, Missouri, North Dakota, Tennessee, and Wyoming. Decisions were taken directly from the state reporters for the states of California, Illinois, New York, New Jersey, Ohio, and Washington, cross-checked with Shepard's Citations to ensure that they appeared in the corresponding regional reporter as well. II. CASE INFORMATION CODED

For each case, the following information was coded: — the case citation — case type (grouped into general categories) — whether or not the attorney general's office was involved — the identity of the attorney general — which side (petitioner or respondent) AG represented — which ideological position (liberal or conservative), if any, the AG represented — whether the AG was a party to the litigation — whether the governor was a party to the litigation — the winning side (petitioner or respondent, conservative or liberal) — the identity of the dissenters (full and partial) — identity of justices not participating — whether the case was mandatory, discretionary, or an original action (where discernable) III. PETITIONER VS. RESPONDENT

In most cases, this was straightforward to code. The AG was coded as NEITHER petitioner or respondent when a) the office was cross-appellant or cross-appellee, and b) when two cases involving the AG's office, one in which it represented the petitioner and the other in which it represented the respondent, were consolidated. IV. CRITERIA FOR "LIBERAL" OR "CONSERVATIVE" Essentially, for the purpose of coding, the identity of the litigants was usually more important than the legal issue involved. I adopted an Ulmer-esque (1978; also similar to Nagel 1961; Leonard 1989) division of the universe into 222 upperdogs and underdogs, with a decision in favor of the upperdog coded as a conservative decision. When two competing litigants were essentially of equal standing, no ideology was coded. When I was in doubt, my decisional rule was to not code an ideological dimension for the case. Under these guidelines, the following situations were coded accordingly: — corporation vs. an individual — the individual is liberal — corporation vs. a government — the government is liberal — government vs. an individual — the individual is liberal — government vs. government — no ideology coded — individual vs. individual — no ideology coded — company vs. another company — ideology coded The above scheme applies almost without exception to the following types of cases:

— breach of contract — wrongful discharge — wrongful death/negligence — products liability — environmental protection — eminent domain — tax — utilities regulation — zoning — workers' compensation — creditor/debtor — landlord/tenant — mineral/water/land rights — consumer protection — medical or legal malpractice For example, if a case involved a landlord-tenant dispute between two corporations, no ideology was coded. If, however, it involved a private party as tenant pitted against a corporate landlord, a vote for the tenant was coded as liberal. No ideology was coded in the following types of cases: —estates/wills —custody — divorce — reapportionment — adoption — stockholder disputes — paternity — dissolution of partnerships 223 All attorney discipline cases were excluded from the analysis. In all criminal cases, a decision for the government was coded as conservative. When a labor union was involved, a decision for the union was virtually always coded as liberal. The only exception was when a member of the union was suing the union itself, usually over a matter of civil rights. In such cases, a decision for the union was coded as conservative. Particular types of cases which deviated from these general rules or which deserve special explanation include: 1) Election Cases — in most cases involving elections, no ideology was coded. The. one exception was in cases involving ballot access requirements. In those cases, any decision expanding access to the ballot was coded as liberal. 2) Libel Cases — This was one of the rare instances in which the issue involved was more important than the identity of the litigants. A decision that protected themedia was coded as liberal, a violation of the general rule that decisions protecting corporations were usually coded as conservative.

3) Termination of Parental Rights — decisions for the parent(s) were coded as liberal 4) Church/State Cases — I coded as liberal those decisions that upheld the separation of church and state, and the free exercise of more "exotic'', non-mainstream religions. However, when a church was seeking tax-exempt status, a decision in favor of the church was coded as conservative. 5) Licensing of Professionals — these cases usually involved allegedly incompetent or negligent doctors seeking their license from a state board or seeking to have their termination from a hospital overturned, while the doctors in these situations are indeed "underdogs" in terms of their resources, I have difficulty thinking of them as underdogs who need special protection by liberal justices. Thus, I considered these cases as examples of a "business vs. business", and hence coded no ideology. 6) Homeowners vs. homeowners associations — Somewhat arbitrarily, if one homeowner sued a homeowners' association, I coded a vote for the single homeowner as liberal. If, however, MORE than one homeowner sued an association, I assumed the resources were more equal and hence coded no ideology. 224 In those cases that had multiple litigants on each side, if ANY of the litigants had "equal" status on the two sides, no ideology was coded. For example, if an injured worker sued both a co-worker and his employer, no ideology was coded. Similarly, if one corporation sued another corporation and the government, no ideology was coded. Many times, cases involving private individuals appeared to actually be disputes between two small businesses. To avoid inconsistency, if the appellation "co." or "inc." did not appear in the case title, a litigant was not considered a business. V. CODING OF CASE OUTCOMES — WINNING AND LOSING Generally, coding case outcomes was straightforward. Cases were coded as "mixed" if a) there were no cross appeals, and if the lower court decision was "affirmed in part, reversed in part", or b) if there WERE cross appeals, and the lower court decision was either affirmed or reversed in its entirety. As indicated in the text, I occasionally deviated from this rule if it appeared that one party had won an overwhelming victory and the other party won only a minor point of law. However, as I desired to avoid as much arbitrary coding of cases as possible, this was done only rarely. In instances of any doubt, the rule was to code the case as "mixed". VII. CODING OF JUSTICES' POLITICAL PARTIES

Justices were only included in the analysis for those years in which they sat on the bench during the entire year. Also, those justices who, for whatever reason, failed to participate in at least 90% of the cases decided by the court during that year were excluded from the analysis for that calendar year, so that the justices were essentially responding to the same set of stimuli. As indicated in the text, information of judges' political party and method of ascension to the bench was collected from a variety of sources. If I could find independent, published verification that a justice belonged to a particular political party, he or she was coded as such. Of the 122 state supreme court justices who sat on the bench for at least one calendar year and who are hence included in this study, I could independently determine the partisanship of 104 (85.2%) of them. Of the remaining justices, for those who were appointed, I utilized the party of the appointing governor as a surrogate. Naturally, this is not a perfect surrogate — Gryski and Main report that the correlation 225 between a justice's party and the party of the governor who appointed them was .822 (1986, 533) — high, but not perfect. Nevertheless, it is the best surrogate available. Utilizing it, I was able to assign a partisanship to all but 4 justices. BIBLIOGRAPHY

Abraham, Henry J., and Robert R. Benedetti. (1969). "The State Attorney General: A Friend of the Court?" Pennsylvania Law Review 117:795-828. Adamany, David W. (1969). "The Party Variable in Judges' Voting: Conceptual Notes and a Case Study." American Political Science Review 63:57-73. Aldrich, John, and Forrest Nelson (1984) . Linear Probability, Logit, and Probit Models. Beverly Hills: Sage Publications. Allen, David W. (1991). "Voting Blocs and the Freshman Justice on State Supreme Courts." Western Political Quarterly 4:727-47. Armstrong, Virginia C., and Charles A. Johnson (1982). "Certiorari Decisions by the Warren and Burger Courts: Is Cue Theory Time Bound?" Polity 15:141-50. Asseo, Laurie (1987). "Arizona Governor, Chief Justice Spar Over Nominee." National Law Journal Feb. 2, 1987. Atkins, Burton M (1993). "Alternative Models of Appeal Mobil­ ization in Judicial Hierarchies." American Journal of Political Science 37:780-98. Atkins, Burton M., and Henry R. Glick (1974). "Formal Judicial Recruitment and State Supreme Court Decisions." American Politics Quarterly 2:427-49. Baum, Lawrence (1992). "Membership Change and Collective Voting in the U.S. Supreme Court." Journal of Politics 54:3-24.

______(1990). American Courts: Process and Policy (2nd. ed.) Houghton Mifflin Co.: Boston. ______(1987). "Explaining the Vote in Judicial Elections: The 1984 Ohio Supreme Court Elections." Western Political Quarterly 40:367-71.

226 227 Baum, Lawrence (1976). "Decisions to Grant and Deny Hearings in the California Supreme Court: Patterns in Court and Individual Behavior." Santa Clara Law Review 16:713-44.

Beiser, Edward, and Johnathan Silberman (1971). "The Political Party Variable: Workers Compensation Cases in the New York Court of Appeals." Polity 3:521-31. Bell, Griffin B. (1978). "The Attorney General: The Federal Government's Chief Lawyer and Chief Litigator, Or One Among Many?" Fordham Law Review 46:1049-70. Bernstine, Nancy (1977). "Prosecutorial Discretion in Consumer Protection Divisions of Selected State Attorney General Offices." Harvard Law Journal 20:247. Beyle, Thad L. (1983). Being Governor: The View From the Office. Durham: Duke Press Policy Studies. Bickel, Alexander (1962). The Least Dangerous Branch. Bobbs- Merrill Co.: Indianapolis. Boersema, Craig E. (1987). "A Comparative-State Study of the Powers and Backgrounds of State Attorneys General, and Their Impact on Public Policy." Unpublished Ph.D. dissertation, Washington State University. Brennan, William (1985). "A Defense of Judicial Activism." in William Lasser, ed. (1992) Perspectives on American Government. D.C. Heath and Co.: Lexington, Mass. Brenner, Saul, and John F. Krol (1989). "Strategies for Cert- iroari Voting on the United States Supreme Court." Journal of Politics. 51:828-40.

Brigman, William (1966). The Office of the Solicitor General of the United States. Unpublished Ph.D. dissertation, University of N. Carolina at Chapel Hill. Bunch, Kenyon D., and Gregory Casey (1990). "Political Controversy on Missouri's Supreme Court: The Case of Merit vs. Politics." State and Local Government Review 22:5-16. Burt, Jeffrey A., and Irving Schloss (1968-69). "Government Litigation in the Supreme Court: The Roles of the Solicitor General." Yale Law Review 78:1442-81. Caldeira, Gregory A. (1983). "On the Reputation of State Supreme Courts." Political Behavior 5:83-108. 228 Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes (1964). The American Voter New York: Wiley. Canon, Bradley C. (1991). "Courts and Policy: Compliance, Implementation, and Impact." in John B. Gates and Charles A. Johnson (eds.) The American Courts: A Critical Assessment CQ Press: Washington, D.C.

______(1983). "Defining the Dimensions of Judicial Activism." Judicature 66:236-47. ______(1972). "The Impact of Formal Selection Processes on the Characteristics of Judges — Reconsidered." Law and Society Review 6:479-93. Canon, Bradley C. , and Lawrence Baum (1981). "Patterns of Tort Law Innovations: An Application of Diffusion Theory to Judicial Doctrines." American Political Science Review 75:975-87.

Caplan, Lincoln (1987). The Tenth Justice. Aired A. Knopf: New York. Carberry, Charles M. (1975). "The State Advisory Opinion in Perspective." Fordham Law Review 44:81 Chamberlain, Ronald S. (1987). "Mixing Politics and Justice: The Office of Solicitor General." The Journal of Law and Politics 4:379-428. Champagne, Anthony (1986). "The Selection and Retention of Judges in Texas." Southwestern Law Journal 40:53-117. Chanin, Leah F. (1976). "Opinions of the State Attorney General" Wisconsin Law Review 70:298 Christenson, Arlen C. (1970). "The State Attorney General." Wisconsin Law Review __:298-340. Cody, W.J. Michael (1987). "Tennessee's Law Firm." Tennessee Bar Journal. 23:12. Cook, Beverly (1977). "Public Opinion and Federal Judicial Policy." American Politics Quarterly 14:567-600. Council of State Governments (various years). The Book of the States. Lexington, KY: The Council of State Governments. Criag, Mickey, and Linda L. Norman (1986). "The Bird Court and Reapportionment: A Case Study in Judicial Partisanship?" Benchmark 2:155-63. 229

Culver, John H., and John T. Wold (1986). "Rose Bird and the Politics of Judicial Accountability in California." Judicature 70:80-9. Dahl, Robert A. (1957). "Decision-Making in a Democracy: The Supreme Court as a National Policy-Maker." Journal of Public Law 6:279-95. Danelski, David J. (1966). "Values as Variables in Judicial Decision-Making: Notes Toward a Theory." Vanderbilt Law Review 19:721-40.

DeLong, Earl H. (1934). "Powers and Duties of the State Attorney General in Criminal Prosecution." Journal of Criminal Law and Criminology 25:358-400. Douglas, Ross, and Michael Catalano (1988) . "How State and Local Governments Fared in the United States Supreme Court for the Past Five Terms." The Urban Lawyer 20:341- 52. Dubois, Philip (1984). "Voting Cues in Nonpartisan Trial Court Elections: A Multivariate Assessment." Law and Society Review 18:395-436.

______(1983) . "The Influence of Selection System and Region on the Characteristics of a Trial Court Bench: The Case of California." Justice System Journal 8:59-87. ______(1980) . From Ballot to Bench: Judicial Elections and the Quest for Accountability. Austin: The University of Texas Press. Egelko, Robert (1987). "Duke's Court." California Lawyer 7:28.

Eiler, Edward J., and Ernest 0. Vincent (1986). "The California Supreme Court and the Death Penalty." Benchmark 2:143-53. Emmert, Craig F., and Carol Ann Traut (1994). "The California Supreme Court and the Death Penalty." American Politics Quarterly 22:41-61.

Epstein, Lee, and Karen O'Connor (1988). "States and the United States Supreme Court: An Examination of Litigation Outcomes." Social Science Quarterly 69:660-74. Fahy, Charles (1942). "The Office of the Solicitor General." American Bar Association Journal 28:20-22. 230 Fino, Susan (1987). The Role of State Supreme Courts in the New Judicial Federalism Westport, CN: Greenwood Press, Inc.

Flango, Victor E., and Craig R. Ducat (1979). "What Difference Does Method of Judicial Selection Make? Selection Procedures in State Courts of Last Resort." Justice System Journal 5:25-44.c. Flango, Victor Eugene, Lettie McSpadden Wenner, and Manfred W. Wenner (1975). "The Concept of Judicial Role: A Methodological Note." American Journal of Political Science 19:277-90.

Friedelbaum, Stanley H. (1982). "Independent State Grounds: Contemporary Invitations to Judicial Activism." in Mary Cornelia Porter and G. Alan Tarr, eds. State Supreme Courts: Policymakers in the Federal System. Westport, CN: Greenwood Press. Frohnmeyer, Dave (1988). "Public Interest Comes First." The Journal of State Government 61:91-3. Galanter, Marc (1974) . "Why the 'Haves' Come Out Ahead: Speculations on the Limits of Legal Change." Law and Society Review 9:95-160.

Galie, Peter J. (1982). "The Other Supreme Courts: Judicial Activism Among State Supreme Courts." Syracuse Law Review 33:731-93. George, Tracy, and Lee Epstein (1992). "On the Nature of Supreme Court Decision Making." American Political Science Review 86:323-37. Gibson, James L. (1978). "Judges' Role Orientations, Attitudes, and Decisions: An Interactive Model." American Political Science Review 72:911-24. Glick, Henry R., and Craig F. Emmert (1987). "Selection Systems and Judicial Characteristics: The Recruitment of State Supreme Court Judges." Judicature 70:228-35. Glick, Henry R. and Kenneth N. Vines (1973). State Court Systems Englewood Cliffs, NJ: Prentice-Hall, Inc. Griffin, Kenyon N., and Michael J. Horan (1983). "Patterns of Voting Behavior in Judicial Retention Elections for Supreme Court Justices in Wyoming." Judicature 67:68-77. 231 Griswold, Erwin N. (1975). "Rationing Justice: The Solicitor General's Caseload and What the Court Does Not Do." Cornell University Law Review. 60:335-54. ______(1969). "The Office of the Solicitor General: Representing the Interests of the United States before the Supreme Court." Missouri Law Review 34:527-36.

Grossman, Joel (1966). "Social Backgrounds and Judicial Decision Making." Harvard Law Review 79:1551-64. Gryski, Gerald S., and Elanor C. Main (1986). "Social Backgrounds as Predictors of Votes on State Courts of Last Resort: The Case of Sex Discrimination." Western Political Quarterly 39:528-37. Haflett, William C. Jr. (1984). "Tice v. Deptartment of Transportation: A Declining Role for the Attorney General?" North Carolina Law Review 63:1051-60.

Hagle, Timothy M. (1991) . "But Do They Have to See It to Know It? The Supreme Court's Obscenity and Pornography Decisions." Western Political Quarterly 44:1039-54. Hall, Melinda Gann (1992). "Electoral Politics and Strategic Voting in State Supreme Courts." Journal of Politics 54:427-446. ______(1987). "Constituency Influence in State Supreme Courts: Conceptual Notes and a Case Study." Journal of Politics 49:1117-24.

Hall, Melinda Gann, and Paul Brace (1992). "Toward an Integrated Model of Judicial Voting Behavior." American Politics Quarterly 20:147-68.

Hanushek, Eric A., and John E. Jackson (1977). Statistical Methods for Social Scientists. Orlando: Academic Press. Harmon, John M. (1988). "Memorandum Opinion for the Attorney General: The Role of the Solicitor General." Loyola of Los Angeles Law Review 4:1089-97.

Heiser, Peter E. Jr. (1982). "Opinion Writing Functions of Attorneys General." Idaho Law Review 18:9-41. Herndon, James (1962). "Appointment as a Means of Initial Accession to Elective State Courts of Last Resort." North Dakota Law Review 38:60-73. 232 Hinkley, Barbara, Richard Hofstetter, and John Kessel (1974). "Information and the Vote: A Comparative Election Study." American Politics Quarterly 2:131-58. Horowitz, Donald L. (1977). The Courts and Social Policy The Brookings Institution: Washington, D.C. Ignagni, Joseph A. (1990). "Explaining and Predicting Supreme Court Decision Making: The Establishment Clause Cases, 1970-1986." Presented at the annual meeting of the Midwest Political Science Association.

Interview (1983). "John Van de Kamp: Crime and Environment Top New Attorney General's Agenda." California Lawyer 3:36. Jacob, Herbert (1984). Justice in America: Courts, Lawyers, and the Judicial Process. Little, Brown, & Co.: Boston. ______(1964). "The Effect of Institutional Differences in the Recruitment Process: The Case of State Judges." Journal of Public Law. 13:104-19.

Jenkins, John A. (1983). "The Solicitor General's Winning Ways." American Bar Association Journal 69:734-8. Johnson, Alan (1992). "Attorney general defending abortion law despite his personal opinions." Columbus Dispatch May 31. Johnson, Charles A. (1979). "Judicial Decisions and Organiza­ tional Change: Some Theoretical Notes on State Court Decisions and State Administrative Agencies." Law and Society Quarterly 14:27-56.

Johnson, Charles A., and Bradley C. Canon (1984). Judicial Policies: Implementation and Impact. CQ Press: Washington D.C. Johnson, Charles A., Roger G. Shaefer, & R. Neal McKnight (1978). "The Salience of Judicial Candidates and Elections." Social Science Quarterly 59:371-8. Johnson, Richard M. (1967). The Dynamics of Compliance: Supreme Court Decision-Making from a New Perspective. Northwestern University Press: Evanston, 111.

Jordan, James E. (1985). "United States Supreme Court Litigation Activities of State Attorneys General." Paper presented at the annual meeting of the Midwest Political Science Association. 233 Kagan, Robert A., Bliss Cartwright, Lawrence M. Friedman, and Stanton Wheeler (1978). "The Evolution of State Supreme Courts" Michigan Law Review 30:121-56.

Kazanjian, John H. (1973). "Consumer Protection by the State Attorneys General: A Time for Renewal." Notre Dame Lawyer 49:410-27. Kearney, Richard C., and Reginald S. Sheehan (1992). "Supreme Court Decision Making: The Impact of Court Composition on State and Local Government Litigation." Journal of Politics 54: 1008-25. Kleinschmidt, Thomas, and Stanley Feldman (1992). "Changing Face of the Judiciary." Arizona Attorney 28:13-5. Korenberg, Rosanne (1982). "Attorneys' Fees to State Attorneys General in Antitrust Actions." Boston University Law Review 62:493. Krislov, Samuel (1959). "Constituency vs. Constitutionalism: The Desegregation Issue and Tensions and Aspirations of Southern Attorneys General." Midwest Journal of Political Science 3:75-92. Lee, Rex E. (1988) "Lawyering in the Supreme Court: The Role of the Solicitor General" Loyola of Los Angeles Law Review 4:1059-67. (1986). "Lawyering for the Government: Polemics and Principle." Ohio State Law Journal 47:595-601. Leonard, James (1989). "Ideology and Judicial Behavior: A Statistical Study of the Ohio Supreme Court: 1970, 1975, 1980, and 1985 Terms." University of Cincinnati Law Review 57:935-86. Leshy, John D. (1988) "The Making of the Arizona Constitution." Arizona state Law Journal 20:1-113. Levy, Edward (1949). An Introduction to Legal Reasoning. The University of Chicago Press: Chicago. Lopez, Larry (1987). "Ariz Gov Backs Down on Court." National Law Journal March 2, 1987, p.41. Lovrich, Nicholas P. Jr., and Charles H. Sheldon (1983). "Voters in Contested Nonpartisan Judicial Elections: A Responsible Electorate or a Problematic Public?" Western Political Quarterly 36:241-56. 234 Macauley, Stewart (1963). "Non-Contractual Relations in Business: A Preliminary Study." American Sociological Review 28:55-67. Mann, Thomas E. (1978). Unsafe at any Margin: Interpreting Congressional Elections. American Enterprise Institute for Public Policy Research: Washington, D.C. Mason, M.P. (1978). "Courting Reversal: The Supervisory Role of State Supreme Courts." Yale Law Journal 47:595-601. McConnell, Michael W. (1988). "The Rule of Law and the Role of the Solicitor General." Loyola of Los Angeles Law Review 4:1105-18. McGuire, Kevin (1991). "Obscenity, Libertarian Values, and Decision Making in the U.S. Courts of Appeals." American Journal of Political Science 36:235-58.

Meese, Edwin III (1985). "End Judicial Activism." in William Lasser, ed. (1992) Perspectives on American Government D.C. Heath and Co.: Lexington, Mass. Miller, Thomas J. (1988). "Telemarketing: Reach Out and Cheat Someone." The Journal of State Government 61:98-9. Morris, Thomas R. (1987). "States Before the U.S. Supreme Court: State Attorneys General as Amicus Curiae." Judicature 70:298-304. Muir, William K. Jr. (1967). Prayer in the Public Schools: Law and Attitude Change. University of Chicago Press: Chicago.

Murphy, Walter (1964). The Elements of Judicial Strategy. University of Chicago Press: Chicago. Myers, Ken (1991). "Rumpus in Columbus as Justices Battle." National Law Journal Nov. 25, 1991. page 3 col. 1. National Association of Attorneys General (1977). Common Law Powers of State Attorneys General. Raleigh, N. Carolina. National Center for State Courts (1987). State Court Caseload Statistics. Washington, D.C. Nagel, Stuart S. (1961). "Political Party Affiliation and Judges' Decisions" American Political Science Review 55:843-50. 235 Neuborne, Burt (1988). Testimony to the Senate Judiciary Committee on March 19, 1987. in Loyola of Los Angeles Law Review 4:1099-1103. (no author) (various years). Who's Who in American Law Chicago: Marquis Who's Who. Note (1977/78). "The Solicitor General and Intergovernmental Conflict." Michigan Law Review 76:324-64.

Note (1974). "The Role of the Michigan Attorney General in Consumer and Environmental Protection." Michigan Law Review 72:1030. Oberly, Charles M. Ill, and Fred S. Silverman (1988) . "Avoiding Public Showdowns." The Journal of State Government 61:94-5. . O'Connor, Karen (1982/83). "The Amicus Curiae Role of the United States Solicitor General in Supreme Court Litigation." Judicature 66:256.

Palincsar, John (1979). "Separation of Powers in Illinois and the Inability to Abrogate Common Law Powers of the Attorney General." Illinois Bar Journal 68:264-71. Pritchett, C. Herman (1969). "The Development of Judicial Research." in eds. Joel B. Grossman and Joseph Tanenhaus. John Wiley: New York. Provine, Doris Marie (1980). Case Selection in the United States Supreme Court. University of Chicago Press: Chicago.

Puro, Steven (1971). "The Role of the Amicus Curiae in the United States Supreme Court." Ph.D. Dissertation, State University of New York at Buffalo. Reincke, M. and J.C. Wilhelmi, eds. (various years). The American Bench: Judges of the Nation. Minneapolis: Reginald Bishop Foster and Associates. Rohde, David W. (1972). "Policy Goals and Opinion Coalitions in the Supreme Court." American Journal of Political Science 16:208-24.

Rohde, David W., and Harold J. Spaeth (1976). Supreme Court Decision Making. W.H. Freeman & Co.: San Francisco. Rosenthal, Alan (1981) . Legislative Life. New York: Harper and Row. 236 Ross, Lynne M. (ed.) (1990). State Attorneys General: Powers and Responsibilities. The Bureau of National Affairs, Inc.: Washington, D.C. Rowland, C.K., and Robert A. Carp (1980). "A Longitudinal Study of Party Effects on Federal District Court Policy Propensities" American Journal of Political Science 24:291-305. Sabato, Larry (1983). Goodbye to Good-time Charlie. Washington, D.C.: CQ Press. Sallen, Pat (1987). "Gordon Takes Over as Chief Justice." Arizona Bar Journal 22:7-11. Salokar, Rebecca Mae (1992). The Solicitor General: The Politics of Law. Temple University Press: Philadelphia. ______(1988). "The Solicitor General: Balancing the Interests of the Executive and Judicial Branches, 1959- 1982." Unpublished Ph.D. dissertation, Syracuse University.

Scheb, John M. III., Thomas D. Ungs, and Allison L. Hayes (1989). "Judicial Role Orientations, Attitudes, and Decision-Making: A Research Note." Western Political Quarterly 42:427-3 5. Schnapper, Eric (1988). "Becket at the Bar — Conflicting Obligations of the Solicitor General." Loyola of Los Angeles Law Review 4:1187-1271. Schubert, Glendon (1962). "The 1960 Term of the Supreme Court: A Psychological Analysis" American Political Science Review 56:90-107.

______(1954). Quantitative Analysis of Judicial Behavior. The Free Press: Glencoe IL.

Schwartz, Joshua I. (1988). "Two Perspectives on the Solicitor General's Independence." Loyola of Los Angeles Law Review 4:1119-65.

Scigliano, Robert (1971). The Supreme Court and the Presidency. The Free Press: New York. Segal, Jeffrey A. (1989). "Supreme Court Support for the Solicitor General: The Effect of Presidential Appointments." Western Political Quarterly 42:137-52. 237 Segal, Jeffrey A. (1986). "Supreme Court Justices as Human Decision-Makers: An Individual Level Analysis of Search and Seizure Cases." Journal of Politics 48:938-55.

______(1984). "Predicting Supreme Court Cases Probablistically: The Search and Seizure Cases, 1962- 1981." American Political Science Review 78:891-900. * Segal, Jeffrey A., and Albert D. Cover (1988). "Ideological Values and the Votes of U.S. Supreme Court Justices." American Political Science Review_&3:557-65. Segal, Jeffrey A., and Harold J. Spaeth (1993). The Supreme Court and the Attitudinal Model New York: Cambridge University Press.

Shaffer, Stephen D. (1982). "Voting in Four Elective Offices: A Comparative Analysis." American Politics Quarterly 10:5-30. Sheehan, Reginald S. (1992). "Governmental Litigants, Underdogs, and Civil Liberties: A Reassessment of a Trend in Supreme Court Decision Making." Western Politics Quarterly 45:27-39. Sheehan, Reginald S., William Mishler, and Donald R. Songer (1992). "Ideology, Status, and the Differential Success of Direct Parties Before the Supreme Court." American Political Science Review 86:464-71. Sheldon, Charles H. (1986). "The Recruitment of Judges to the Washington Supreme Court: Past and Present." Willamette Law Review 22:85-128. Silbey, Susan S. (1980/81). "Case Processing: Consumer Protection in an Attorney General's Office." Law and Society Review 15:849-81. Slotnick, Elliot E. (1984). "Judicial Selection Systems and Nomination Outcomes: Does the Process Make a Difference?" American Politics Quarterly 12:225-40. Sobeloff, Simon E. (1955). "Attorney for the Government: The Work of the Solicitor General's Office." American Bar Association Journal 41:229-79. Songer, Donald (1979). "Concern for Policy Outputs as a Cue for Supreme Court Decisions on Certiorari." Journal of Politics 41:360-82. 238 Spaeth, Harold J., and Michael F. Altfeld (1985). "Influence Relationships Within the Supreme Court: A Comparison of the Warren and Burger Courts." Western Political Quarterly 38:70-83. Stern, Robert L. (1960). "The Solicitor General's Office and Administrative Agency Litigation." American Bar Association Journal 46:154. Squire, Peverill, and Eric R.A.N. Smith (1988). "The Effects of Partisan Information on Voters in Nonpartisan Elections." Journal of Politics 50:169-79. Sudnow, David (1965). "Normal Crimes: Sociological Features of The Penal Code in a Public Defender's Office." Social Problems __:255-76. Swinford, Bill (1991) . "A Predictive Model of Decision Making in State Supreme Courts: the School Funding Cases." American Politics Quarterly 19:336-52. Tanenhaus, Joseph, Marvin Schick, Matthew Muraskin, and Daniel Rosen (1963). "The Supreme Court's Certiorari Jurisdiction: Cue Theory" in Glendon Schubert (ed.) Judicial Decision Making. Free Press: Glencoe, Illinois.

Tarr, G. Alan, and Mary Cornelia Aldis Porter (1988). State Supreme Courts in State and Nation. New Haven: Yale University Press. Tate, C. Neal (1983). "The Methodology of Judicial Behavior Research: A Review and Critique." Political Behavior 5:51-82.

______(1981). "Personal Attribute Models of the Voting Behavior of U.S. Supreme Court Justices: Liberalism in Civil Liberties and Economic Decisions, 1946-1978." American Political Science Review 75:355-67. Tate, C. Neal, and Roger Handberg (1991). "Time Binding and Theory Building in Personal Attribute Models of Supreme Court Voting Behavior, 1916-1988." American Journal of Political Science 35:460-80. Teger, Stuart, and Douglas Kosinski (1980). "The Cue Theory of Supreme Court Certiorari Jurisdiction: A Reconsideration." Journal of Politics 42:834-46. Thatcher, Thomas D. (1931). "Genesis and Present Duties of the Office of Solicitor General." American Bar Association Journal 17:519-21. 239

Thompson, William N. (1974). "Should We Elect or Appoint State Government Executives? Some New Data Concerning State Attorneys General." Midwest Review of Public Administration 8:17-41. Ulmer, S. Sidney (1979). "Parabolic Support for Civil Liberties Claims: The Case of William 0. Douglas." Journal of Politics 41:634-9. ______(1978). "Selecting Cases for Supreme Court Review: An Underdog Model." American Political Science Review 72:902-10.

______(1973a). "Social Background as an Indicator to the Votes of Supreme Court Justices in Criminal Cases: 1947-1956 Terms." American Journal of Political Science 17:622-30.

______(1973b). "Parabolic Support of Civil Liberties Claims: The Case of Douglas." Journal of Politics 41:634-9. ______(1972). "The Decision to Grant Certiorari as an Indicator to the Decision on the Merits." Polity 4:429- 47.

(1971). " and the Brown Decision." Journal of Politics 33:689-702. Ulmer, S. Sidney, and David Willison (1985). "The Solicitor General of the United States as Amicus Curiae in the U.S. Supreme Court, 1969-1983 Terms." Paper presented at the annual meeting of the American Political Science Association.

Van de Kamp, John K. (1988). "Crime Fighting Goes Hi-Tech." The Journal of State Government 61:103-4. Walker, Thomas G., Lee Epstein, and William J. Dixon (1988). "On the Mysterious Demise of Consensual Norms in the U.S. Supreme Court." Journal of Politics 50:361-89. Wasby, Stephen L. (1988). The Supreme Court in the Federal Judicial System (3rd ed.) Nelson-Hall: Chicago. Watson, Richard A. (1986). "Observations on the Missouri Court Plan." Southwestern Law Journal 40:1-117.

Watson, Richard A., and Rondal G. Downing (1969). The Politics of the Bench and Bar: Judicial Selection Under the Missouri Nonpartisan Corut Plan. John Wiley: New York. 240 Webster, William (1988). "Protecting Consumers: Challenges Posed by New Federalism." The Journal of State Government 61:96-7. Welch, Susan, Michael Combs, and John Gruhl (1988). "Do Black Judges Make a Difference?" American Journal of Political Science 32:126-36. Wheeler, Stanton, Bliss Cartwright, Robert A. Kagan, and Lawrence M. Friedman (1987). "Do the 'Haves' Come Out Ahead? Winning and Losing in State Supreme Courts, 1870- 1970." Law and Society Review 21:403-45. Wilkins, Richard G. (1988). "An Officer and an Advocate: The Role of the Solicitor General." Loyola of Los Angeles Law Review 4:1167-86. Witt, Elder (1986). A Different Justice: Reagan and the Supreme Court. CQ Press, Inc.: Washington, D.C. Wold, John T. (1974) . "Political Orientations, Social Background, and Role Perceptions of State Supreme Court Judges." Western Political Quarterly 27:239-48.