Judges, Litigants, and the Design of Courts

Abstract

Two important perspectives on courts highlight fundamentally different elements of adjudication and yield distinct predictions about judicial outcomes. The Attitudinal Model of judicial voting posits judge ideology as a strong predictor of court outcomes. Alternatively, the Law and Economics perspective focuses on the settlement behavior of litigants and reasons that while judges may vote ideologically, litigants adapt to these ideological proclivities, nullifying the effect of judge ideology. This analysis focuses on reconciling expectations about the effects of judge ideology and litigant strategies by examining their contingent nature and the conditioning effects of institutional design. The analysis examines state supreme courts from 1995–1998 to identify empirical evidence supporting both perspectives. While some state supreme courts have discretionary dockets allowing judges greater opportunities to exercise their ideology, others lack discretionary docket control, making dockets and outcomes largely litigant driven. Support for each perspective largely hinges on this fundamental feature of institutional design.

Theories of judicial outcomes rightly and obviously focus on the decisions judges make, yet those decisions are rendered on an exceedingly small proportion of the cases initiated. “Most cases settle” before they get to judges at either the trial or appellate court level (Galanter and Cahill 1994).¹ The premise of this article is that court design and litigant strategies influence the types of cases appealed and, in so doing, condition the empirical impact of variables derived from two contrasting models of judicial decision making. We focus on civil cases in state supreme courts, as the bulk of civil litigation—including appellate litigation—occurs in state courts. Testing theories using state appeals is warranted in its own right because appeals provide the bulk of our knowledge of the legal system and, further, the great preponderance of legal disputes, both civil and criminal, take place in the state courts (Galanter 2004: 506).

Our study unfolds as follows: in the next section we present two competing theoretical perspectives on explaining court outcomes. In this section we also discuss court institutional design and its implications for those perspectives and court outcomes. Next, we set forth two tests: the first test assesses how court institutional design structures condition the influence of judicial ideology on court outcomes, the second test provides insight on why court outcomes (i.e., litigant win-rates) may deviate from a well-established equilibrium rooted in law and economics research. Following this, we discuss our results and offer conclusions regarding their implications for understanding judicial processes.

Two Perspectives on Court Outcomes

The Attitudinal Model

In the field of political science, Segal and Spaeth's (1993, 2002) Attitudinal Model (AM) argues that institutional features securing judicial independence allow Supreme Court judges to base their decisions on their policy preferences. This perspective is well-known and with a wealth of quantitative studies supporting this view (Hagle and Spaeth 1993; Howard and Segal 2002; Rohde and Spaeth 1976; Segal 1997; Segal and Cover 1989) it would be hard to overestimate the effect AM has had on judicial scholarship. As one review of the state of knowledge on judicial behavior stated, “in scholarship on the Supreme Court … the view that policy considerations are dominant over legal considerations has been taken by the most prominent work” (Baum 1997: 22). Another assessment echoes this view by concluding that, “the attitudinal model's systematic empirical shattering of the myth of mechanical jurisprudence permeates virtually all our work on judges and courts today” (Lawrence 1994: 3). While the attitudinal model has attracted critics (see, e.g., Gillman 1999; Whittington 2001), it continues to motivate influential studies of judicial decision making, providing support for, or limitations to, the basic argument (see, e.g., Bailey and Maltzman 2008; Bowie and Songer 2009; Brace and Boyea 2008; Richards and Kritzer 2002).

Law and Economics

Less known to those who study court outcomes in political science is an alternative but not necessarily contradictory perspective developed in Law and Economics (LE) which focuses on the impact of litigants on judicial outcomes (see, e.g., Eisenberg and Farber 1997; Kessler et al. 1996; Posner 1973; Priest and Klein 1984; Wittman 1985). Priest and Klein (1984) famously promoted the perspective, noting that taking cases to court incurs costs. According to their “selection hypothesis,” litigants seek to avoid expanding costs when possible, settling cases when the cost of litigation is high. Litigants estimate who is likely to win and lose, and simpler cases commonly settle in pre-trial. The remaining cases that go to trial are typically difficult, complex, and less predictable. A priori these cases reveal no obvious winners or losers, it is predicted that success for plaintiffs and defendants will hover around 50 percent.²

The “50 percent rule” has become a prominent idea in the LE perspective (Cooter and Ulen 1988; Cross 2003; Donohue 1988; Polinsky 1989; Posner 1998). Although this idea remains influential, a theoretical debate has emerged about strategy and/or rationality—that is, how the plaintiff/defendant might predict the probability of winning. Critics of the 50 percent rule commonly focus on forces shaping the ability of litigants to deal with uncertainty, including the legal environment and litigant resources as well as litigants’ asymmetrical information and variations in stakes at risk for litigants (e.g., Bebchuk 1984; Watts 1994).³ While the 50 percent rule has not found a wealth of empirical support when considered in its simplest (most strict) form, more sophisticated analyses of the rule (i.e., adjusted to consider factors affecting uncertainty such as asymmetrical information or stakes) have empirically confirmed its viability (e.g. Kessler et al. 1996).

One factor not in much dispute in LE is that litigant strategy should cancel out the effects of judge ideology when ideology is known. Research within LE acknowledges that while judges may be motivated ideologically, litigant knowledge of judge ideology reduces its impact. This is because litigants decide whether to pursue their cases with judge ideology in mind (Priest and Klein 1984; Waldfogel 1998; Wittman 1985, 1988); such litigant strategies result in courts hearing non-random samples of cases. Litigants seeking liberal (or conservative) outcomes generally do not pursue their cases in conservative (or liberal) courts. If a litigant seeks a liberal outcome but faces a conservative court, then that litigant has greater incentive to either settle pre-trial or drop the case altogether. As a consequence, cases that do go to court are commonly ideologically ambiguous. Alternatively, some cases may result from miscalculations by litigants about the ideological dimensions of their case, or the preference of judges, or both. However it is expected that strategic miscalculations should be balanced between those seeking conservative and liberal outcomes in the long run. This leads to the expectation of approximately 50 percent win rates for petitioners and defendants in outcomes, producing no ideological bias despite the ideology of judges. Overall, the LE perspective leads us to expect that strategic litigants respond to incentives to produce cases that nullify the effects of judge ideology across repeated outcomes.

Within the AM perspective, research has likewise considered the rationality of litigants (e.g., Songer et al. 1995). Songer et al. (1995) find that criminal defendant litigants are rational when weighing the costs and benefits of waging an appeal; litigants convicted of crimes calculate success given several factors, including their personal estimate that an appeal will be upheld, disagreement in a lower appellate court decision, and costs in relation to the litigants’ personal resources. That important work provides an excellent starting point for addressing the questions that we endeavor to answer. However, it examines just one aspect of the larger courts/litigation process (i.e., litigants’ appellate strategies) with which we are concerned. It does not address the broader issue of the influence of litigant strategy on judicial outcomes over the long run, nor does it speak to the possible impact of significant differences in judicial institutional structure on the influence of judges’ ideologies on voting.

We must note that AM and LE approaches were derived from studying courts at the opposite ends of the judicial system. Overwhelmingly, the AM literature focuses on the U.S. Supreme Court, the court for which it was intended.⁴ Given its purpose, applying its logic elsewhere should not be construed as a challenge to its adequacy. However, the apparent primacy of ideology in that court makes it reasonable to consider its role in lower courts. Alternatively, the LE literature on case selection focuses primarily on the effects of litigant selection in trial courts (see Waldfogel 1998; but see, e.g., Hanssen 1999; Clermont 2000 for exceptions). However, it seems unrealistic to presume that litigant choices about pursuing their cases are not operating in other courts. While we know that the estimated ideologies of state supreme court judges vary widely and have differential impact across these courts (Brace et al. 2000), the widely varying impact of state supreme court judge ideology could be the result of strategic litigant case selection. Our premise, to be developed below, is that the effects of judge ideology and litigant strategies are jointly conditioned by the institutional design of courts, illustrating more systematically when judge ideology does and does not influence outcomes in these courts and why.

Judge Ideology, Litigant Strategy, and Institutional Design

These seemingly opposed perspectives can be productively reconciled by considering institutional design and forces driving cases reaching appeal. Past research on the U.S. Supreme Court underscores the important role docket discretion plays in their decision making.⁵ There is consistent evidence that the Court is more likely to grant certiorari to review lower court decisions that are ideologically inconsistent with the Court's current majority because the Supreme Court acts ideologically, and thus it is more likely to grant certiorari to reverse than to affirm the lower court (see, e.g., Armstrong and Johnson 1982; Black and Owens 2009; Boucher and Segal 1995; Brenner and Krol 1989; McGuire and Stimson 2004).

To date, one study has examined the effects of case selection on outcomes in state appellate courts. Eisenberg and Heise (2009) observe that at the state trial court level, selection theory reasonably describes the outcome distribution—at least at the aggregate level—as outcomes do not dramatically favor plaintiffs or defendants. At the appellate level however, defendants were far more successful than plaintiffs in securing a reversal of a trial decision, providing little support for litigant selection effects. Because these authors were primarily interested in features of trial litigation (jury versus non-jury on ultimate outcome), they employ the Civil Court Justice Survey of States (CCJS). This survey examines courts from the country's largest trial courts. Unfortunately, the data on state supreme courts in the CCJS survey analyzed by Eisenberg and Heise all have lower appellate courts.

Eisenberg and Heise (2009: 124) present a differential attitude hypothesis to refer not to judicial background or belief variables but rather to group-level attitudinal differences between appellate and trial court judges. They do not directly test the effects of judge ideology on these biases. Moreover, as noted above, their results derive from particular types of state supreme courts (those with docket discretion), albeit the modal type, with discretionary dockets. Their findings are consistent with those on the U.S. Supreme Court where the attitudinal model finds most consistent support. There, the Justices enjoy discretionary jurisdiction and choose almost all of the cases they hear (Guthrie and George 2005). They do not consider the effects non-discretionary dockets have on the forces affecting court decision making and outcomes.

The neo-institutional literature suggests that institutional structures are important influences on judicial outcomes affecting everything from the business of courts to the impact of judges’ preferences (e.g., Brace and Hall 1997, 2001; Yates, Tankersley & Brace 2010). To consider the influence of institutional design and litigant strategy, we examine all tort cases decided by state supreme courts in the years 1995 through 1998, distinguishing between state courts with and without discretionary jurisdiction over torts.⁶ Dockets on courts without discretionary jurisdiction over torts are not filtered by their judges and are thus litigant-driven: cases that make it to appeal are the result of litigant calculations influenced by the degree of uncertainty that litigants bear. Following the logic of the LE perspective, litigants should adjust their strategies to incorporate and effectively nullify judge ideology: appeals are costly and knowledge of court ideological preferences should figure into their strategies and incentives for pre-trial amelioration. Alternatively, other state supreme courts have control over their tort case agenda and authority to let lower court decisions stand or to hear a case. Although litigant strategies still play a large role in shaping the pool of cases eligible for review, judges exercise discretion reflecting their ideological preferences, ultimately determining which cases get on the docket in a manner resembling the U.S. Supreme Court. Consequently, we expect the effects of judge ideology to be significant in these courts. We divide our analysis to compare state supreme courts with discretionary authority over tort litigation and courts with constrained discretion. According to Flango and Rottman (1998), sixteen state supreme courts have complete discretion to accept or reject tort appeals and another thirty-four state supreme courts are required to accept each tort appeal or have limited authority over their tort agendas.

Further related to the discretion of state supreme courts is the connection between agenda authority and the presence of lower appellate courts. Logically, state supreme courts with lower appellate courts may be more likely to independently select tort disputes on a case by case basis, while state high courts without lower appellate courts should be constrained by mandatory jurisdictions. While the relationship between agenda discretion and the presence of lower appellate courts may seem straightforward, agenda authority within state supreme courts actually varies considerably whether or not they receive assistance from lower appellate courts. While thirty-nine states have lower appellate courts, only fourteen of those states permit state supreme courts complete discretionary control over their tort agenda. Among the eleven states without lower appellate courts, two state supreme courts have authority to decide their tort caseload. Thus, discretionary jurisdiction is not always connected with lower court assistance.

While we acknowledge the more complex choices facing litigants where operating in a state with an intermediate appellate court, we believe the impact of lower appellate courts is overshadowed by the capability of judges in state high courts with discretionary tort jurisdictions to select disputes on a case by case basis. While litigants in non-discretionary jurisdiction states determine those cases the state supreme court hears, where courts have discretionary authority to single out cases, judges make the final decision about which cases they adjudicate. We discuss the impact of this discretion more fully below.

Tests Using Tort Litigation in State Supreme Courts

Our analysis focuses on tort cases for several reasons. Tort cases involve damage, injury, or a wrongful act other than breach of contract for which a civil suit can be brought and for which courts will provide a remedy, usually in the form of damages. These cases lend themselves to the types of strategic calculations presumed to affect the distribution of cases that come to courts (Priest and Klein 1984)—that is, they involve calculations of the monetary costs and financial benefits of appeal. In addition, tort cases commonly have an ideological direction (McGuire et al. 2004). A decision in favor of a plaintiff (i.e., the party or parties that originally initiated litigation) is commonly viewed as a liberal outcome while one favoring defendants is construed as a conservative outcome. Thus, it is reasonable to expect judges’ ideology to exert an influence (e.g., Sheehan et al. 1992; Yates, Tankersley, & Brace 2010). Litigants face conventional strategic imperatives in cases with ideological dimensions. Finally, torts are an area of law commonly examined in LE analyses of case selection processes and win-rates and provide the best evidence for the hypothesized processes (Eisenberg 1990). As such, these are ideal conditions for assessing how strategic case selection processes and judge ideology may interactively operate in American state supreme courts.

The two perspectives we consider traditionally focus on differing features of judicial behavior. The AM approach commonly focuses on the manner in which judge ideology predicts case votes. Alternatively, the LE approach considered here is interested in court outcomes and the relative biases in those outcomes.

In the analysis that follows we consider these two dimensions of litigation phenomena. In the first, we examine the effects of judge ideology on voting in tort cases. In the second, we consider the ideological direction or bias of state supreme court outcomes.

Judge Votes—How Does Court Design Condition the Influence of Judge Ideology?

In American state supreme courts some judges get to pick the cases they wish to review while others do not. This distinction is not trivial because given the opportunity it is reasonable to expect judges to pick cases in a manner that reflects their ideological preferences. This argument was made cogently by McGuire and Stimson (2004) (see also McGuire et al. 2004; McGuire et al. 2009) who examined the United States Supreme Court (a court notable for its docket discretion). They show that the effects of judge ideology on case votes are conditioned by the nature of the appeal. Specifically, appellants may make accurate or inaccurate estimates of the preferences of judges (or the “winnability” of an appeal), and courts may choose to review cases or not. Reversals of lower court decisions occur when a petitioner accurately estimates the outcome of the appeal—i.e., the lower court will be overturned in the higher court. Because the appellate process operates primarily to correct error, strategic reversals by judges are expected to be more frequent in appellate courts with discretionary jurisdiction.⁷ Alternatively, petitioning parties sometimes misestimate the higher court's ideological position, leading the court to affirm a lower court ruling. Under these circumstances, petitioners are worse off than before because the status quo remains unchanged and they have incurred the additional costs of the appeal. These cases represent strategic miscalculations on the part of litigants and provide less evidence of ideological voting (McGuire and Stimson 2004). As a consequence, appellate courts should generally let lower court decisions they favor stand and most typically intervene with lower court decisions they oppose.⁸ It is here that we expect judge ideology to operate but note that its operation is contingent on having discretion over the docket.

The expectations for judge voting are thus straightforward:

H1: The impact of judge ideology on voting will be more pronounced in state high courts with discretionary jurisdiction and in the context of lower court decisions being reversed.

In sum, where courts have discretion like the U.S. Supreme Court to pick their cases, they should most commonly pick cases they wish to reverse. Their case selection should reflect their ideological preferences and it is here we should expect to see judge ideology operating in a significant and predictable manner. Alternatively, where courts do not exercise this discretion, their dockets are shaped exclusively by the strategic calculations of litigants and, for the reasons outlined above, we expect those calculations to cancel out the effects of judge ideology.

Court Outcomes—Win-Rate Equilibrium and Litigant Uncertainty

We also consider aggregate court outcomes. The LE perspective commonly focuses on win-rates, or plaintiff victories relative to defendant victories, with the expectation that these should hover at 50 percent due to the operation of litigant strategies. Recall from above that it is litigant uncertainty that leads cases to trial. We reason that departures from the 50 percent norm should be influenced by features of state supreme court environments that heighten or lessen litigant uncertainty regarding outcomes.⁹ It is expected that institutional and contextual factors reduce (or increase) uncertainty, which consequently reduces (or increases) departures from 50 percent win-rates for plaintiffs.¹⁰ Most notably, these effects should be observed where state supreme courts have no discretion because it is here that court dockets are driven exclusively by the strategic calculations of litigants. Accordingly, this analysis provides three general conditions that affect uncertainty among litigants. First, litigants may vary in their capability to estimate likely outcomes. Second, litigants should react to the preferences and voting behavior of judges thereby increasing their ability to make precise estimates. Third, features of courts may increase the accuracy of litigant estimates. The question then becomes what can reduce uncertainty for litigants about prospective outcomes; the following section considers several probable conditions of uncertainty.

Litigant Influences on Outcomes and Win-Rates

Litigants (and their attorneys) vary in terms of their capabilities, resources, and incentive structures (Galanter 1974). This variation should be particularly true in tort cases in which plaintiffs are typically inexperienced and are taking a “single-shot” against defendant businesses who are often well seasoned “repeat players” in court. While defendants should have greater resources and experience, plaintiffs should be relatively more plagued by uncertainty because of their disadvantage in terms of these same resources and experience in court. Accordingly, they should err more in their strategic calculation as to whether an appeal will be successful. As such,

H2: Departures from the 50 percent norm should increase as the proportion of original plaintiff appeals increases in courts.

Misjudgments by plaintiffs are not expected to promote as much bias in outcomes in courts with discretion over torts because courts would typically be disinclined to hear cases they would affirm anyway.

Judge and Contextual Influences on Outcomes and Win-rates

The LE approach reasons that, although judges may vote ideologically, litigants adapt to these ideological proclivities, thus nullifying the effect of judge ideology. The reason for this is quite simple. Litigants with cases with liberal (or conservative) dimensions would be unlikely to appeal an adverse outcome to a conservative (or liberal) court. This litigant selection process, for example, means that liberal or conservative courts seldom receive cases in which they can exhibit their ideological preferences. It seems more likely that litigants seeking liberal (or conservative) outcomes would select only liberal (or conservative) courts to hear their claims. However, what if uncertainty exists about judge ideology? New membership on a court might produce ideological change, thereby increasing litigant uncertainty. Thus,

H3: Ideological change from court membership turnover should increase litigant uncertainty, resulting in greater departures from the 50 percent win-rate norm.

Consistent with these expectations of litigant response to judge ideology, this analysis considers the effects of more (ideologically) extreme courts on litigant certainty. In situations in which litigants face ideologically extreme courts (either collectively very liberal or very conservative), they have a reasonable opportunity to decrease their uncertainty as those courts are expected to provide clear policy signals to litigants.¹¹ In regards to reacting to ideological extremism, movement away from the 50 percent norm should decrease when the mean ideology of a particular court is furthest from the cross-state average of court ideology.¹² Thus,

H4: Where courts are ideologically extreme, deviations from the 50 percent win-rate norm should be smaller.

Litigants may also react to voting blocks on courts. Consistent patterns of consensus and dissensus should reduce uncertainty for litigants. Obviously, consistent patterns of unanimity should provide strong signals to litigants, as consistent divisions in court voting should reduce uncertainty. However, wide variance in patterns of dissent should increase uncertainty. For example, a nine-member court that consistently voted 5 to 4 should be less predictable than one with consistent outcomes of 9 to 0 or 8 to 1 voting blocks. Thus,

H5: A greater dispersion of dissent should increase uncertainty by making it harder for litigants to predict voting blocks, thereby producing greater departures from the 50 percent win-rate norm.

Institutional Influences on Outcomes

Of the institutional features that influence litigant certainty, one feature is likely to be the methods by which judges are kept on courts. Existing research makes a forceful case for the effects of judicial elections on case outcome (Brace and Hall 1997; Hanssen 1999). More specifically, Hanssen (1999) provides a compelling argument that elected judges are more predictable than their counterparts in appointive courts. Appointed judges, on the other hand, serve in courts that more likely promote independence and provide no interaction with the public; accordingly, these judges are less predictable (Landes and Posner 1975). Thus,

H6: Judicial selection through elections should provide greater predictability and cause smaller deviations from the 50 percent win-rate norm.¹³

State supreme courts vary by how they process litigation. While most state courts of last resort make decisions en banc, fourteen courts divide their membership into smaller panels of justices to facilitate the more effective processing of cases (Rottman 1995). Assignment to decision panels generally operates in a rotating manner with judges serving over a fixed duration. Where courts utilize panels, their decision outcomes should be less predictable as litigants will have less stable information about the ideological proclivities of a panel's members. Thus,

H7: Courts that utilize panels should reduce predictability among litigants and, as a result, should have greater deviations from the 50 percent win-rate norm.

The number of state supreme court judges should operate in a similar manner. The size of state supreme courts ranges from five to nine judges. Where courts have more judges, their case outcomes should not be controlled by the voting proclivities of one or two swing judges, thereby making court decisions easier to predict. Furthermore, larger deliberative entities are generally known as providing more accurate, consistent, and reliable decisions (Abramowicz 2000; Surowieki 2004). Thus,

H8: Larger courts should provide better and more plentiful information for prospective appellants and, consequently, should deviate less from the 50 percent win-rate hypothesis.

Lastly, state supreme courts vary in the length of judicial terms. The more litigants can learn about judge preferences, the fewer miscalculations will be made about pursuing their appeal. On courts where judges serve long terms, litigants and their counsel have a stronger basis for estimating prospective outcomes. Thus,

H9: Where courts have longer terms, deviations from the 50 percent win-rate norm should be smaller.¹⁴

Data, Modeling, and Estimation

We employ the State Supreme Court Data Archive (SSCDA), which contains all tort appeals and judge votes within those appeals from 1995 to 1998 for the systematic evaluation of decisions in tort appeals.¹⁵ To test our hypothesis regarding judge votes, a model of individual judge votes was developed employing a generalized estimating equation (GEE) logistic regression with an independent correlation structure.¹⁶ We use conditional state-level fixed effects with robust standard errors to estimate the event of a judge vote in favor of the original plaintiff.¹⁷ The dependent variable (liberal judge vote) captures the ideological choice that judges face in tort appeals—namely, either support for the original plaintiff or defendant. The use of the GEE procedure addresses the dilemma of interdependence within the data. This technique offers the benefits of more precise variance-covariance estimates even when the nature of dependence is unknown (Zorn 2006).¹⁸ Using this GEE logistic regression, the model is:

$\text{Pr}\left({\text{Liberal Judge Vote}}_{\text{it}}\right)=f({\text{Judge Ideology}}_{\text{it}}{\beta }_c+{\upsilon }_i)$

where β_C represents the effects of Judge Ideology and υ_i represents unit-specific (case) effects which captures subject-specific propensities toward the outcome variable. Here the focus is directed to the impact of judge ideology on liberal judge votes. This portion of our analysis compares the effect of judge ideology on judge votes in all decisions and then separately on decisions to affirm and reverse lower court decisions, with the expectation that the effects of ideology hinges on court discretion.

For our analysis of court outcomes, or more specifically, departures from the 50 percent equilibrium, an ordinary least squares regression is utilized on win-rates at the court level. The estimation design is a three stage panel fixed effects regression with vector decomposition and panel corrected standard errors (see Plümper and Troeger 2007). We use this procedure to efficiently estimate time invariant or nearly time invariant (“sluggish”) explanatory variables.¹⁹ The dependant variable (win-rates) is the percentage of wins by the plaintiff minus 50 percent, expressed as an absolute value. The model for this multivariate pooled cross-sectional time series analysis is:

$\begin{array}{c}\left|\text{Observed Win-Rates}-50\%\right|={\beta }_0+{\beta }_1\text{Original Plaintiff Suit}+\\ {\beta }_2{\text{Ideological Change}}_{\text{t-}1}+{\beta }_3\text{Ideologically Extreme Court}+\\ {\beta }_4{\text{Dispersion of Dissent}}_{\text{t-}1}+{\beta }_5\text{Elect}+{\beta }_6\text{Panel}+\\ {\beta }_7\text{Court Size}+{\beta }_8\text{Term}+ϵ\end{array}$

The expectation is that these variables exert significant effects when the appellate process is litigant driven (i.e., when there is no discretionary control) and less effectively when judges can act strategically in selecting cases for review as is the case when there is tort case selection discretion.²⁰

Descriptions of the variable measures are made available in Table A1 of the Appendix.

Results

Judge Votes in Tort Cases

We begin with this model which considers the effects of judge ideology on decisions to support the plaintiff in state supreme court tort appeals. Our analysis considers the nature of the case (reversals versus affirmances) as well as the nature of tort jurisdiction with the expectation that the effects of ideology will be strongest in reversals and evident only in states with discretionary control over tort litigation.

Table 1 presents the analysis of judge votes favoring plaintiffs, which we construe as a reflection of liberal behavior: liberal judges are expected to support “have-not” plaintiffs more than do conservative judges.²¹ The results are divided into votes in reversals in which theory suggests the strongest evidence of ideological voting occurs, and affirmances in which the same theory predicts a weaker relationship with ideological preferences. We further divide our results into courts with and without full discretionary docket control with the expectation that judicial discretion in case selection is a necessary condition for ideological influence on judge votes.

Table 1. Predicting Individual State Supreme Court Judges’ Votes in Tort Cases GEE Fixed Effects Logistic Regression, Robust Standard Errors. Dependent Variable: Judge's Vote Favoring Plaintiff = 1, Judge's Vote Favoring Defendant = 0*

Notes: Entries in bold are significant at the.05 level, one-tailed test.

^* Unit of analaysis is a judge's vote.

^† Represents the change in probability of votes favoring a plaintiff given a change from the minimum to maximum value of judge ideology.

The results in Table 1 conform well to the expectation offered by McGuire et al. (2004) and the supplemental institutional caveat offered here, and are robust with regard to model specification (i.e., random or fixed effects models) and estimation technique (i.e., standard or population-averaged logistic regression).²² Specifically, ideological voting in these tort cases is strongly evident in reversals, while the impact is weakened in affirmances. Substantively speaking, the effect of judge ideology is always stronger in reversals than affirmances; this effect is enhanced by the presence of discretionary jurisdiction over torts. Table 1 reports that the most liberal justice, compared to the most conservative justice, is 13% more likely to support a plaintiff in a reversal, while a similar difference in an affirmance increases the likelihood of a pro-plaintiff vote by approximately 8%. Clearly the impact of judge ideology is strongest in reversals and in those courts that have discretion over their docket. The comparative weakness of judge ideology in courts without discretion compared to courts with discretion corresponds to what we would expect if litigant strategies operated to cancel out the influence of judge ideology. In sum, the effect of judge ideology is most pronounced where judges have the opportunity to exercise their strategic discretion and least pronounced where litigant strategies are dominant.

Litigant Win-Rates in Tort Cases

We now turn our attention to court outcomes and factors influencing departures from the 50 percent equilibrium. Table 2 first presents descriptions of plaintiff win-rates in twelve areas of tort litigation appealed before state supreme courts and it illustrates the pervasiveness of the 50 percent norm.²³ In all but one area the 50 percent norm is evident: only one category of tort (i.e., toxic injury) is statistically different from 50 percent; yet this occurs only in one of the most infrequent tort areas.²⁴ In sum, with the noted exception, tort outcomes reveal no significant bias favoring plaintiffs or defendants.

Table 2. Plaintiff Success Rates (%) by Areas of Tort Litigation and Tort Discretion*

Note: ^*Unit of analysis is a tort case by area of policy and tort discretion.

Entries in bold are statistically different from 50 percent at the 0.05 level.

This preliminary evidence suggests that strategic litigants are acting in the manner hypothesized by the 50 percent rule. A more in-depth analysis can determine if, in fact, variations in win-rates are influenced by factors that could be reasoned to shape uncertainty. Although significant departures from 50 percent win-rates are the exception and not the norm, nevertheless, variation exists in the degree of departure from this norm across courts. As reasoned earlier, smaller departures from the 50 percent norm are expected when litigants have a stronger basis for estimating potential outcomes. The magnitude of departure from 50 percent should increase as a function of uncertainty that litigants face if case selection is litigant driven, as hypothesized in courts without discretionary jurisdiction. However, this expectation does not extend to outcomes in courts with docket selection authority over torts because case selection in these courts is a joint function of litigant and court strategies; thus, litigants need to pursue their appeal, and the court has to grant review. This process resembles that operating in the U.S. Supreme Court. This distinction is not trivial. The majority of literature concerning the effects of ideology on court outcomes centers on the U.S. Supreme Court while the preponderance of literature on the 50 percent rule focuses on trial courts (where dockets are shaped exclusively by litigant strategies). Thus, conflicting expectations concerning the effects ideology can be resolved by considering the appellate structure and the nature of the judicial vote.

In examining deviations from the 50 percent hypothesis, the dependent variable is the percentage of wins by the plaintiff minus 50 percent, expressed as an absolute value. The effects of our explanatory variables for all courts, as well as courts with and without discretionary jurisdiction over torts, are then compared.²⁵ The results concerning all courts are presented in the first column of Table 3. The model performs reasonably well. The correctly signed and statistically significant coefficients for ideologically extreme courts, dispersion of dissent, elective methods of retention, panel, size of the court, and judge term length provide support for the hypotheses. The results conform to the expectations: in courts with ideological distinctions, consistent patterns of consensus, elected judges, an absence of panels, more judges, and longer terms, litigants face less uncertainty and win-rates stay significantly closer to 50 percent, all other factors being equal.

Table 3. Predicting Deviations from Fifty Percent Plaintiff/Defendant Wins Panel Fixed Effects Regression, Panel Corrected Standard Errors (PCSEs). Dependent Variable: Absolute Value [Obs. Plaintiff Win-rate—50 percent]*

Note: ^*Unit of analysis is the observed plaintiff win-rate deviation from 50 percent by state and year.

RMSE = Root Mean Squared Error; Entries in bold are significant at the.05 level, one-tailed test.

When we divide our analysis into courts with and without discretionary jurisdiction, striking patterns are revealed. In courts without discretionary dockets the model explains 30% of the variance and performs extremely well. As hypothesized, the variables representing original plaintiff suits, ideologically extreme courts, dispersion of dissent, elective method of retention, panels, size of the court, and longer judicial terms receive directional support and are statistically significant. Substantively speaking, for every standard deviation increase in the number of plaintiff appeals, there is approximately a 2% departure from the 50 percent norm. Since plaintiffs are commonly “single-shot” individuals going up against “repeat player” organizations, this finding is consistent with capability theories of litigation. In terms of the ideological characteristics of courts, ideologically extreme courts decrease departures from the 50 percent norm: where courts are more ideologically extreme they are less likely to produce outcomes favoring one side or the other, as hypothesized. As discussed previously, this is likely due to litigants having better information on the probable ideological tenor of outcomes before the court and, accordingly, tending to lean toward negotiated settlements, which in turn push win-rates toward 50 percent. Similarly, a standard deviation increase in the dispersion of dissent (indicating more uncertainty regarding court preferences) increases departures from the 50 percent norm by more than 2%. As for estimates of institutional influence on litigant certainty, four structural features strongly influence the balance between plaintiff and defendant success. Courts with elective methods of retention reduce departure from 50 percent by 4% and those with panels likewise reduce deviations from the 50 percent norm by almost 4%. Further, a standard deviation increase in the number of justices serving decreases departures from 50 percent by more than 1%, while a standard deviation increase in term length of 4.31 years is associated with approximately a 2% decrease in deviation.

Notably different results emerge in courts with discretionary dockets as illustrated in the second column of Table 3. Here the only statistically significant effects on deviations from the 50 percent norm are that for original plaintiff suit, panel, and the length of term. Further, each significant coefficient operates in the opposite direction of those for courts without discretion, further highlighting the conditioning effect of institutional structure. Lastly, ideological change among judges exerts no statistically discernable effects on win-rates whether courts do or do not have discretionary jurisdiction.

The most compelling features of these results are that factors impinging on litigant uncertainty affect win-rates in state supreme courts when those courts’ dockets are driven by litigant strategies. Most notably, in these circumstances, judge ideology operates to cancel rather than promote bias in court outcomes: the more ideologically extreme a court is, the less likely it is to produce ideologically biased outcomes. Moreover, this pattern does not hold where courts exercise discretion over their dockets.

Conclusion

Scientific progress involves some mix of problem solving (Polanyi 1957) and the art of oversimplification (Popper 1992: 42). In studies of law and courts, identification of “problems” risks the appearance of challenging established formulizations, while simplification carries the danger of insulting those with great appreciation for nuance and detail. Noting these hazards, in this study we contrast two important perspectives on judicial outcomes that by their very nature are simplifications. These perspectives highlight fundamental but by no means exhaustive features of litigation. Our approach is not to treat these as dueling adversaries with the expectation that one will vanquish the other in a march toward “vertical progress” (Dryzek 1986) in theorizing about courts. Instead, we highlight their contingent strengths in differing institutional settings seeking “lateral progress” (Dryzek 1986) to broaden our understanding of forces operating on adjudication in differing institutional settings.

Thus, we have not posed a competition between alternative explanations of judicial behavior and court outcomes, rather we have presented a puzzle and developed an argument to solve that puzzle with the intent of synthesizing our understanding of the forces operating on judicial outcomes. The puzzle concerned the roles of judge ideology and litigant strategies—alternatively reasoned to be central for understanding judicial outcomes. Careful examination of hypotheses regarding these two perspectives required analysis consistent with each perspective: judge votes and court outcomes. The results illustrate when judge ideology operates and when it does not and provides a theoretical synthesis to help understand why. The findings of this study highlight the important role of institutional design in conditioning the empirical relevance of each perspective. A fundamental feature of court institutional design—namely, the presence or absence of discretionary jurisdiction—is critical in shaping the contexts in which judicial behavior occurs and the effects of judge ideology in judicial processes.

The findings presented in this research provide lateral progress in theoretical development by illustrating the conditions under which each theoretical piece better fits features of judicial behavior. The synthesis of litigant strategies, judge ideologies, and court structures provides a more comprehensive understanding than looking at any single facet alone. In light of these findings, the frustrations of students of lower courts that seek to “export” thoughtful and well-developed theories derived from studies of the U.S. Supreme Court can be understood in a new light. Our results demonstrate that litigant strategies and court institutional design in state supreme courts play a decisive role in shaping the agendas of these courts and these agendas condition the influence of judge preferences. While this may disappoint students of state supreme courts desiring parsimonious explanations of their behavior, it should not. A unique advantage to studying state courts generally, and state supreme courts specifically, is that they allow us to consider the comparative impact of differing institutional structures on judicial processes. Given this, these results should encourage those who integrate the important role that context and institutional structure plays in their interpretations of courts below the U.S. Supreme Court.

As the proponents of the AM approach note very clearly, the strength of their model hinges on the unique institutional setting of the U.S. Supreme Court. Our findings reaffirm this by illustrating how comparable structures in state supreme courts provide strong evidence in support of the attitudinal model. At the same time, our analysis indicates why the attitudinal model performs poorly where structures are notably different from the U.S. Supreme Court. Most notably, in these courts litigants and not judges shape court dockets and under these circumstances judge ideology works to reduce rather than promote ideological bias in court outcomes. From this perspective it is clear that a one-size-fits-all approach to judicial decision making at the subnational level is inappropriate. A general theory of judicial decision making for these courts must integrate significant features of institutional structure and their effects on judge and litigant strategy. Armed with this understanding, we hope that more fruitful interpretations of state supreme court decision making will be possible.