Setting the Supreme Court’s Policy Agenda

Abstract

We develop a new approach to understanding which legal questions the Supreme Court chooses to address. We show that the Supreme Court is more likely to resolve ideologically polarizing legal questions. This result is based on a new technique for estimating the ideology of a doctrine, which we implement using a dataset of intercircuit splits. We use this technique to identify legal issues that are ideologically polarizing and show they are more likely to be addressed by the Supreme Court than less polarizing issues. Our results demonstrate how the Supreme Court uses certiorari to advance its ideological policymaking goals.

The Supreme Court is not, and never has been, primarily concerned with the correction of errors in lower court decisions….The function of the Supreme Court is, therefore, to resolve conflicts of opinion on federal questions that have arisen among lower courts, to pass upon questions of wide import under the Constitution, laws, and treaties of the United States… to decide only those cases which present questions whose resolution will have immediate importance far beyond the particular facts and parties involved.

Chief Justice Vinson

Address to the American Bar Association

September 7, 1949

Members of the Supreme Court can further their policy goals because they…comprise a court of last resort that controls its own caseload.

Segal and Spaeth

The Supreme Court and the Attitudinal Model Revisited

The United States Supreme Court is a national policymaker. The Court makes policy by deciding cases. Policymaking permeates each step of the Supreme Court’s work, including as it chooses which cases to hear, for that is how the Court sets its policy agenda. When the Supreme Court chose in 2015 to address the constitutionality of same-sex marriage bans, it did so because it wanted to address that question and set a national policy. ¹ When the Supreme Court chose in 2022 to hear Dobbs v. Jackson Women’s Health Organization, it did so because it wished to revisit its role in regulating abortion – not because it wanted to review the Fifth Circuit’s decision. In other words, when the Supreme Court chooses to hear a case, it does so because it is interested in addressing the legal question at issue in the case in order to set a national policy.

The legal questions that most pique the Court’s interest may be ripe for attention, important, or causing legal confusion – or more than one of these. Beyond strictly legal considerations, the Court may particularly seek out or particularly avoid political questions. Addressing legal questions that have ideological import – that is, questions that activate the left-right ideological spectrum in contemporary American politics – allows Justices to advance their ideological goals and makes them players in the political arena.

We articulate a theory of certiorari of legal questions, that is, why the Supreme Court chooses to address some legal questions but not others. ² We propose that the Supreme Court is more likely to review those legal questions that are ideological, than those that do not activate the left-right ideological spectrum in contemporary American politics.

Our theory foregrounds the Court’s interest in policymaking. While many agree that the Court’s goals are to make policy, a focus on policymaking is still relatively new to the empirical study of certiorari in the political science literature. Extant literature has focused on which individual cases get reviewed, a question that is distinct – theoretically, conceptually, and empirically – from the question of which issues get addressed. Importantly, then, our theoretical claim is coupled with an empirical innovation for identifying which legal questions the Supreme Court chooses to address. Because our theory states the Supreme Court is focused on questions, not individual decisions, so too must our empirical analysis – and our understanding of certiorari – focus not only on individual decisions that are up for review but also on the broader precedent that governs each decision.

To evaluate our claim, we study intercircuit splits. Splits arise when different circuits address the same legal question but develop different doctrines. Splits are possible, and common, because although there is an expectation of stare decisis within circuits, there is no such requirement across circuits. Intercircuit splits are fertile ground for doctrinal development – they are much more likely to be addressed by the Supreme Court and are disproportionately important and contested questions.

We analyze a dataset of 119 intercircuit splits, some of which have been resolved by the Supreme Court and some of which have not been resolved. For each intercircuit split, we know the identity of every appellate judge who ever heard a precedent-setting case about that question and which legal policies those judges chose. With these data, we are able to identify legal questions that polarize Democratic from Republican appointees and distinguish ideologically polarizing legal questions from others that do not divide appellate judges along ideological lines.

We show that the more ideologically polarizing a legal question is, the more likely the Supreme Court is to review it. That is, from a pool of comparably important questions, the Court chooses to hear those with a political valence. We validate our measure of polarization using Supreme Court merits votes: we show that when the Court decides a case to resolve a polarized question, the Justices themselves polarize over the final decision.

Our results identify which legal questions the Supreme Court chooses to address when granting certiorari – those that divide lower court judges along ideological lines. By focusing on which questions are addressed, rather than which cases, our work is a new perspective in research on certiorari.

Ideology and policymaking in theories of certiorari

In this paper, we develop a theory about how a desire to make policy – and especially a desire to make ideological policy – would influence certiorari decisions.

Certiorari is the process by which the Supreme Court decides what to decide. Perry (1991), a foundational political-science account of the certiorari process, understands the criteria for certiorari as being divisible into jurisprudential considerations and outcome considerations. The prototypical example of a jurisprudential consideration is the existence of a true, deep intercircuit split. An outcome consideration, by contrast, is motivation to change policy or make ideological moves.

Certiorari for policymaking

How does the Supreme Court decide to decide? First and foremost, the Supreme Court seeks questions where the law is contested: unclear, unsettled, non-existent, or ripe for change (see e.g. Lax and Rader 2010). The Court intentionally populates its discretionary docket with questions that “tender plausible legal arguments on both sides.” (Segal and Spaeth 2002, 93). Also, with vanishingly few exceptions (e.g. Bush v. Gore, 531 U.S. 98 (2000)), the Court addresses questions that have implications for future disputes (Perry 1991, 36). The Court plays an “‘Olympian role,’ announcing rules and standards from on high” and avoids “examining how the lower courts apply those rules or standards, nor does the Court itself routinely apply the rules and standards it announces” (Shapiro 2006, 272–273). ³

In his foundational account of certiorari, Schubert (1959) conceived of the court as “political,” meaning that, “the Court’s decisions have a significant effect upon the making or reshaping of various aspects of public policy” (211). In Schubert (1959)’s account, certiorari is part of the policymaking process. He describes a Court granting cert in order to create new legal rules and argues that political preferences about what those new rules should be influence certiorari decisions (215). Perry (1991); Hammond, Bonneau and Sheehan (2005); Sommer (2011); Carrubba and Clark (2012) add additional detail to the claim that Justices are policy-oriented at the certiorari stage, arguing that individual petitions are evaluated with an eye to policy outcomes.

Others describe how the Supreme Court pursues its policymaking goals through learning about or adopting policies articulated in lower court decisions (Clark and Carrubba 2012; Clark and Kastellec 2013; Beim 2017; Strayhorn 2020). The Court does so by considering petitions for certiorari from individual cases (Owens and Sieja 2017), but what they are attentive to is the legal issues presented by the cases, where the same legal issues might come up across multiple cases. As Perry (1991) notes, Justices and their clerks use the term “case” and “issue” interchangeably because “it is the issue the case raises that is important, not the case itself” (220–221).

Our empirical innovation is to explain how outcome considerations shape the choice of which legal question – not just which petition – to address. But most extant literature, which we review now, is focused on which individual petitions are addressed.

Ideology in theories of certiorari

Since most political scientists believe outcome considerations prevail at the merits stage, much of their analytical attention to the certiorari stage has also focused on what Perry (1991) calls the outcome mode – and especially on the role of ideology in certiorari. The notion that ideological motivations influence the Supreme Court’s certiorari decisions goes back at least to Schubert (1959). Several theoretical accounts of the role of ideology in certiorari use a principal-agent model (see, e.g. Cameron 1993; Songer, Cameron and Segal 1995; McNollgast 1995; Cameron, Segal and Songer 2000; Spitzer and Talley 2000). In those accounts, the Supreme Court grants certiorari sparingly, to review cases that seem particularly worthy of reversal but also to induce behavior in lower courts. ⁴ In particular, principal-agent accounts assume that fear of review and reversal may cause lower courts to abide by the Supreme Court’s preferences.

Principal-agent models of certiorari offer predictions about which lower-court cases will be reviewed based on the ideological valence of the case outcome or the ideologies of the judges who decided the case, or both. For example, Cameron, Segal and Songer (2000) and Sommer (2014) predict ideologically distant lower courts should be reviewed more often than proximate lower courts, especially when they make decisions that are consistent with their relative ideological predisposition. Relevantly, Klein and Hume (2003) show that circuits that are reversed do fall in line – circuits that have experienced higher rates of reversal exhibit more compliant decisions over the following two years.

Ideology and certiorari, empirically

Empirical work has explored whether and how ideological factors act as “cues” (Tanenhaus et al. 1963) that help the Supreme Court sift through a large docket and identify those petitions most worthy of certiorari, from an outcome-mode perspective. ⁵ For example, in the spirit of the principal-agent theories described above, empirical scholarship shows that under conservative Chief Justices Burger and Rehnquist, the Supreme Court disproportionately reviewed cases with liberal outcomes (see e.g. Caldeira and Wright 1988; Songer, Segal and Cameron 1994; Caldeira, Wright and Zorn 1999; George and Solimine 2001; George and Yoon 2003) and reviewed fewer cases from relatively conservative circuits (Lindquist, Haire and Songer 2007). In the area of search and seizure, the Court was more likely to review liberal decisions made by liberal panels than by conservative panels (Cameron, Segal and Songer 2000). Black and Owens (2012) corroborate this result, showing that in death penalty decisions between 1986 and 1993, Justices’ own ideologies, the ideology of the lower-court judges, and the decision those judges had made interacted to predict certiorari votes. ⁶

Although there is consensus that ideology is an important cue, there is no consensus as to the details. For example, there is disagreement over whether Supreme Court Justices prefer to review decisions made by lower-court judges who are ideologically distant. Owens and Simon (2011) find that as the ideological distance between the Supreme Court and lower-court panel grows, review becomes less likely. Carrubba and Clark (2012) find that the effect of ideology on the probability of review is parabolic – judges who are somewhat but not too distant are most likely to be reviewed. Bonneau et al. (2007) and Black and Owens (2009) show that a Supreme Court Justice may want to grant certiorari if she is ideologically closer to the median justice than to the appellate panel that made the decision in question. Bonica, Chilton and Sen (2023) show that the ideology of the appellate panel interacts with the ideologies of the petitioner and the respondent to predict certiorari.

In sum, empirical studies that attempt to relate the ideologies of case outcomes or lower-court panels to the likelihood of review have either been fairly general (e.g. the conservative-leaning Supreme Court reviews more liberal outcomes) or have been in conflict with findings from similar studies with the same aim.

The role of intercircuit splits in doctrinal development

We focus on ideologically motivated policymaking through the review of intercircuit splits because circuit splits occur over legal questions that are both important and beg doctrinal development. That is, splits compel the Court’s policymaking role.

An intercircuit split occurs when two or more circuit courts answer the same legal question differently. Splits can arise because of the decentralized and empowered nature of appellate court decision-making. Within each circuit, there is a norm of stare decisis: appellate judges are expected to follow their own circuits’ precedents. But across circuits, there is no expectation of stare decisis. A circuit judge is not obliged to follow other circuits’ precedents (see, e.g. Hinkle 2015, exploring the effects of ideology on citation behavior within versus across circuits).

Intercircuit splits are not merely judicial disagreements on paper. When circuits split, judge-made law is different in different parts of the country and federal law is not uniform. When the Supreme Court resolves an intercircuit split, it announces one doctrine that becomes the law of the land, abrogating any contrary doctrine in the circuits (see, for example, Lindquist and Klein (2006) for a discussion of the Supreme Court’s behavior when resolving intercircuit splits).

There are factors beyond legal uniformity that draw the Court to questions over which there is a split. First, circuit splits occur only when doctrine is unclear or nonexistent. ⁷ Moreover, circuit splits occur disproportionately in important areas of the law, since judges must care enough about the issue to split from their colleagues (Klein 2002).

For these reasons, cases that present legal questions over which there is a split are more likely to be reviewed by the Supreme Court than other cases (see, e.g. Roehner and Roehner 1953; Tanenhaus et al. 1963; Ulmer 1984; Estreicher and Sexton 1984; Caldeira and Wright 1988, and cites infra). Over the course of a term, the Court dedicates about at least 40% of its docket to resolving intercircuit splits (Frost 2008, 1632–1633, noting 45% to 70% of the Court’s docket is dedicated to splits). ⁸

Even so, not all intercircuit splits are resolved by the Supreme Court. Beim and Rader (2019) show that two-thirds of circuit splits are never resolved, even though almost all unresolved splits continue to generate significant litigation in federal courts. So how does the Supreme Court – as an ideological policymaker – choose which intercircuit splits to resolve? Though there is ample empirical work pointing to the role of ideology in certiorari decisions, there is little empirical analysis of which questions get addressed for ideological reasons. Grant, Hendrickson and Lynch (2012) compare the ideological divisions of granted certiorari petitions to denied certiorari petitions, and find that granted petitions are more ideologically polarized than denied petitions. Clark and Kastellec (2013) and Beim and Rader (2019) (whose data we use in this paper) study the resolution of intercircuit splits (as we do), but neither considers the role of ideology.

Making ideological policy on the Supreme Court

We present a new perspective on the role of ideology in certiorari decisions. Integral to our perspective is the recognition that the Supreme Court does not monitor individual decisions – or at least, not merely. A Court that cares about making policy cares about setting its policy agenda. The Court focuses on attributes of legal questions – the Court selects questions that would allow it to pursue its policymaking goals.

Unlike a monitoring court, the policymaking court does not care per se about the ideological direction of any one given lower-court decision, nor about the ideologies of lower-court judges who made a particular decision. When choosing whether to grant certiorari, the Court’s interest is in the policy it expects to make. We claim further that jurisprudential motivations do not mute outcome motivations. Outcome motivations can still power decision-making when the Court is in jurisprudential mode. Even when resolving intercircuit splits, the Court may seek out those intercircuit splits that are ideological.

Certiorari for ideological policymaking

Among contested legal questions, which would the Supreme Court most like to answer? We argue that the Court wishes to address contested legal questions with ideological import. While answering any legal question necessarily involves creating or choosing a legal rule, the legal question may not always activate ideological preferences: compare the constitutionality of bans on the use of affirmative action in college admissions to whether there is a time limit on foreign service of process. ⁹

In sum, as Posner notes, it is only over “unclear” legal questions that are “charged with political significance” that “the political prejudices and the policy preferences” of judges come into play (cited in Segal and Spaeth 2002, 93). In the following section, we explain how we use the behavior of lower-court judges to identify contested political questions that should be attractive to a Supreme Court that wants to make ideological policy.

Lower-court preferences and behavior

Like Justices on the Supreme Court, judges on the Courts of Appeals pursue ideological goals. However, they face more constraints in doing so. In particular, the Courts of Appeals must hear every case brought before them. Therefore, many of their decisions apply clearly established law and are not legally “close cases” like those that make it on the Supreme Court’s docket. In disposing of routine cases, there is little room for ideology to influence decision-making (Cross 2007). Therefore, the extent to which circuit judges vote differently based on their ideologies is small compared to the Justices and varies based on the decision context (Epstein, Landes and Posner 2013).

However, the Courts of Appeals often decide cases of first impression, in which they address open legal questions and develop policies. These cases are excellent opportunities for ideological decision-making, especially when the questions have a clear ideological valence (Epstein, Landes and Posner 2013; Sunstein et al. 2006). Appellate judges take full advantage of these opportunities to choose legal rules they prefer ideologically, and, absent clear Supreme Court precedent, do so with little regard as to what the Supreme Court might prefer (Klein 2002). Subsequent decisions within the circuit are made in light of circuit precedent and do not allow judges to express their ideology as freely.

Since there is no expectation of stare decisis across circuits, a case of first impression in a circuit allows the first panel to decide in a relatively unconstrained fashion. Of course, they may take into account the decisions colleagues in other circuits made (see e.g. Hinkle 2015). But the first panel in a circuit may vote with few constraints and may create an intercircuit split if it chooses. Subsequent panels, however, are expected to follow the circuit precedent.

The attractiveness of polarized intercircuit splits

Since appellate judges may express their ideological preferences when deciding cases of first impression, we can use their choices to distinguish which legal questions are ideologically tinged by observing whether a question triggers appellate judges’ ideologies.

Consider a legal question with two possible answers – policy A and policy B. ¹⁰ As different appellate courts hear cases about this legal question, judges sign on to either policy A or policy B. As they do, we (the researchers) observe how different judges react. We learn how divisive the question is: do judges tend to agree, signing on to the same answer to the legal question, or are they evenly split between sides? We can also see how judges’ own ideologies relate to the answer they pick – perhaps judges appointed by Democrats tend to pick one legal answer, while judges appointed by Republicans pick another. We call a legal question ideologically polarized when it offers one policy alternative that is more attractive to judges appointed by Democratic presidents than to judges appointed by Republican presidents, and another that attracts Republican judges more than Democratic judges.

Polarizing questions offer the Supreme Court an opportunity to make ideological policy. ¹¹ An ideologically polarized question offers a choice between a liberal policy and a conservative policy. Not all questions offer these kinds of options. For example, some questions offer a choice between two policies that are not especially distinct ideologically, or do not map neatly onto a liberal-conservative dimension. Therefore, when choosing which legal questions to address, we hypothesize that the Supreme Court will be more likely to answer those legal questions that polarize circuit judges in a left-right sense, than those that divide judges along other dimensions.

Ideologically polarized intercircuit splits may be particularly attractive for ideological policymaking, even as compared to ideologically polarizing questions on which there is no split. The Court may seek to preserve its institutional legitimacy even while making ideological policy (see, e.g. Epstein 1998). Questions that have a plausible and ideologically appealing answer offer a cheat code for the legal game (Posner 1993, 2005). Addressing intercircuit splits provides the cover to engage in ideological policymaking through the “legal” process of granting cert to a case to resolve doctrinal conflict in the lower courts. For a court that wishes to avoid perceptions of ideology, the cover provided by resolving an intercircuit split can allow them to seek out and abrogate ideologically unappealing doctrine. Even for the pure attitudinalist, that is, a justice who uses legal arguments instrumentally to achieve policy ends (Segal and Spaeth 2002), ideological splits guarantee that the justice’s ideologically preferred doctrine has plausible legal justification. We note that most scholars have followed Perry (1991) and considered resolving intercircuit splits to be a jurisprudential (legal) project, as opposed to an ideological project. An exemplar of this perspective is Epstein, Landes and Posner (2013), who argue the Court’s decisions are more often unanimous when resolving intercircuit splits because that is a less ideological activity (130) (see also, e.g. Black and Owens 2009). In contrast, we claim that resolving intercircuit splits can indeed be an attractive venue for making ideological policy: we argue the jurisprudential and outcome modes are not mutually exclusive.

Data

Our thesis is that the Supreme Court prefers to address legal questions with an ideological valence, and that polarization of circuit judges can indicate which questions have an ideological valence. A recent dataset allows us to build a unique measure of polarization with which we can answer our question. The data, collected by Beim and Rader (2019), consist of 136 intercircuit splits, 43 of which were resolved and 93 of which have not been resolved. These 136 splits are a pseudo-random sample of acknowledged splits that began between 2005 and 2013. ¹²

Beim and Rader (2019)’s process for identifying intercircuit splits begins with the Seton Hall Circuit Review, which publishes quarterly summaries of new intercircuit splits. The Seton Hall Circuit Review identified these intercircuit splits by searching Westlaw for key terms and reading the results. The summaries note the legal question and include some relevant appellate case citations. The resultant data are a small sample of those intercircuit splits that exist.

The structure of Beim and Rader (2019)’s data is unique and particularly well-suited for our research design. In particular, their data capture every lower-court case about each of many specific legal questions. This detail allows us to measure the ideological polarization of each intercircuit split – and then to see which legal questions are addressed and which are left unaddressed.

For each of the 136 intercircuit splits, Beim and Rader (2019)’s data include all circuits involved and the order in which they joined the split, all precedent-setting cases (cases of first impression) within each circuit that addressed the legal question over which there was a split, and all citing cases from each relevant circuit – that is, cases that applied the rulings from the precedent-setting cases. The final dataset includes 692 precedent-setting cases and 2,627 citing cases, nested in 136 intercircuit splits. The result is a unique dataset that includes all relevant circuit cases for each specific legal question, where some legal questions were addressed by the Supreme Court and others were not. With these data in hand, we are able to build a very detailed impression of how lower courts treated these legal questions, and then to evaluate which of those legal questions were addressed by the Supreme Court.

Dependent variable

Our interest here is in which intercircuit splits the Supreme Court chooses to address, and so our level of analysis is the intercircuit split. Our dependent variable is a dichotomous measure of whether an intercircuit split was resolved. Like Beim and Rader (2019), we code the dependent variable $ Resolved=1 $ if the Supreme Court grants certiorari to a case involved in a split, and upon deciding the case, ultimately resolves the split. If there are no petitions for certiorari, if all petitions are denied, or if a petition is granted, but the resulting opinion does not resolve the intercircuit split, we code the dependent variable $ Resolved=0 $ . ¹³

Independent variables

Our measure of polarization

Our primary question is whether the Supreme Court is more drawn to resolving intercircuit splits that are more ideologically polarizing than those that are less ideologically polarizing.

We use appellate judges’ decisions – their “endorsements” of a doctrine in written opinions, in cases of first impression in their circuits – as a measure of that doctrine’s ideological valence. When one doctrine is very popular with Democratic appointees but very unpopular with Republican appointees, while the opposing doctrine is very popular with Republican appointees but very unpopular with Democratic appointees, we call that an ideologically polarized intercircuit split. ¹⁴

We begin by counting the number of Democratic appointees who heard cases of first impression ¹⁵ on each of our 119 questions, then also the number of Republican appointees who heard cases of first impression on each question. The results are in Figure 1a. We include this figure to show that in most intercircuit splits, the number of judges to hear the question – from each party – is sufficient to give us a good sense of a legal answer’s popularity with that party. The median number of Democratic judges is seven, and the median number of Republican judges is 10. (We had to remove one in which no Democratic appointees decided cases of first impression. In all other splits, at least one Democratic appointee and one Republican appointee heard a case of first impression in the split.)

Figure 1. Our Measure of Polarization.

Then, for each intercircuit split, we calculate the proportion of Democratic appointees who heard a precedent-setting case in that split who took side A (which side is A is arbitrary and is chosen without loss of generality). We include Democratic appointees who sided with the majority’s endorsement of side A and also Democratic appointees who dissented from majority decisions that took B. ¹⁶

Having estimated how popular a doctrine is among Democratic appointees, we then construct the same measure for Republican appointees – we calculate the proportion of Republican appointees who heard a case in the same split who took side A. Our measure of polarization is the absolute value of the difference between the proportion of Democratic appointees taking side A and Republican appointees taking side A.

We illustrate our measure with an example. In Schuette v. Coalition to Defend Affirmative Action (134 S.Ct. 1623), the U.S. Supreme Court resolved an intercircuit split about the use of affirmative action in college admissions. In 1997, a Ninth Circuit panel ruled that a constitutional amendment banning the consideration of race in college admissions did not violate the Equal Protection Clause (122 F. 3d 692 (CA9 1997)). Three judges – O’Scannlain and Leavy (both appointed by Ronald Reagan) and Kleinfeld (appointed by George H.W. Bush) – participated in the decision; none dissented. We thus want to classify this doctrinal position as being quite popular with Republican appointees. In 2012, the Sixth Circuit split from the Ninth when it issued an en banc decision, that an amendment to the Michigan constitution, which banned the consideration of race in college admissions, did violate the Equal Protection Clause (701 F. 3d 466 (CA6 2012)). All of the Democratic appointees voted with the majority. We thus classify this doctrinal position as being quite popular with Democratic appointees. Of the eight Republican appointees on the en banc panel, only one voted with the majority. The other seven Republican appointees dissented, and instead agreed with the Ninth Circuit panel. This reinforces our estimate that the Ninth Circuit’s position is popular with Republican appointees. This is a very polarized split, as “Unconstitutional” was endorsed by 100% of Democratic appointees who heard this question, and only 9% of Republican appointees. We subtract popularity among Republican appointees from popularity among Democratic appointees to find the difference in popularity; $ 100-9=91. $ Therefore, our polarization score for this split is 0.91.

Most splits do not exhibit such a high degree of polarization over doctrinal choices. Figure 1b shows the distribution of our polarization measure for all intercircuit splits. The median value of this score is .19. That is, in the median polarized split, Democratic appointees support a doctrine at a rate that is 19 percentage points higher than Republican appointees. Some of the least polarized intercircuit splits in our data are about whether Article 12 of The Hague Convention on Civil Aspects of International Child Abduction is subject to equitable tolling (resolved in Lozano v. Montoya Alvarez, 134 S.Ct. 1224) and whether there is a time limit on foreign service of process (unresolved).

This is a relatively large degree of ideological voting among appellate judges. Existing scholarship on ideological behavior on the Courts of Appeals has studied outcomes votes – that is, votes on the merits, not doctrinal choice. That scholarship has found a difference of 4–15 percentage points between Democratic and Republican appointees (see, e.g. Sunstein et al. 2006; Cross 2007; Epstein, Landes and Posner 2013). That we find a larger difference is as expected. We are observing judges’ votes on legal questions that are controversial and unsettled, in cases of first impression, with no existing circuit precedent. ¹⁷

Other variables

We control for several characteristics of intercircuit splits that the Supreme Court might consider when deciding whether to address a split.

First, we control for the Number of Circuits involved in the intercircuit split. ¹⁸ Gressman et al. (2007)’s handbook Supreme Court Practice distinguishes “deep” splits, in which many circuits have participated, from “shallow” splits, in which few circuits have participated, noting the central relevance of this to the Supreme Court’s decision, as deep splits entail greater non-uniformity in the law. (The average number of circuits in an intercircuit split in our data is 5).

The distribution of those circuits within splits may also concern the Court. Are they equally divided in the split? Or is the split imbalanced, with just one circuit against several? Lindquist and Klein (2006) find that split imbalance is predictive of which doctrine the Supreme Court ultimately chooses; and so perhaps imbalance is also predictive of which intercircuit splits the Court will hear. Lopsidedness may also frame the Court’s perception of polarization. We account for Lopsidedness as the absolute value of the difference in the number of circuits on each side of the split, at resolution or truncation. ¹⁹

Beim and Rader (2019) find that active intercircuit splits are more likely to be resolved than inactive intercircuit splits. They note that some intercircuit splits begin with multiple decisions across multiple circuits taking opposing sides in a short time span, while other splits begin years after any circuit took the initial side. In the former instance, circuits are developing opposing doctrines simultaneously. The latter is less troubling from the perspective of legal uniformity, and so is less indicative of a pressing legal question. We account for this by measuring the conditions under which the split began. Dormant start indicates whether the first side of the split was active or dormant when the split began. It is an indicator that is coded “1” if 3 or more years had passed since the last case on the first side at the time the split began. Splits in which the first case on each side of the split occurred in the same year, or fewer than 3 years apart, receive a “0” for Dormant start. Twenty-four percent of circuit splits had a dormant start.

Several individual actors can signal to the Court that it should resolve an intercircuit split. Among these are the Solicitor General and individual lower court judges. A lower court judge may dissent from his colleagues’ decision; dissents are known to increase the likelihood of certiorari (see e.g. Tanenhaus et al. 1963; Caldeira and Wright 1988). Therefore, we note whether there was a Dissent in any case in the intercircuit split. Seventy percent of intercircuit splits have at least one dissent. The Solicitor General may file a petition for certiorari; the SG is known as the “tenth justice” since these petitions are so often granted (see, e.g. Bailey, Kamoie and Malztman 2005, and cites infra). We code $ SGPetition=1 $ if the Solicitor General petitioned for certiorari in any case in the split. ²⁰ Ten percent of intercircuit splits have at least one petition from the Solicitor General.

Finally, we account very roughly for the Issue area addressed in the intercircuit split, as in Beim and Rader (2019). They categorize the intercircuit splits by whether they are aboutCriminal Procedure, Economic Activity, or something else (including, for example, questions about Civil Rights or the First Amendment). Eighteen percent of the intercircuit splits are about economic activity. This issue includes several splits about ERISA and splits about the Securities and Exchange Commission, for examples. Forty-five percent are about criminal procedure, many of which are about sentencing (note, Beim and Rader (2019) data begin in 2005, so many of the intercircuit splits are about the circuits trying to interpret and apply Booker). Thirty-seven percent are about other issue areas (the omitted category in our regression); these include a split about age discrimination in employment, splits about procedural questions (like class certifications and attorneys’ fees), and constitutional questions like affirmative action in college admissions. ²¹

Model and results

Is the Supreme Court more likely to resolve polarized splits than non-polarized splits? Yes. In the raw data, 58% of splits in the top decile of polarization ( $ Polarization>0.5 $ ) are resolved, compared to only 17% of splits in the bottom decile of polarization ( $ <0.02 $ ).

Table 1 presents the estimated coefficients and 95% confidence intervals from a linear probability model where the dependent variable is Resolved. The regression indicates which intercircuit splits are more likely to be resolved. ²² The Supreme Court is 45 percentage points more likely to resolve a maximally polarized split than an unpolarized split ( $ p=0.02 $ ), conditional on other factors thought to predict split resolution. By comparison, around 5% or 6% of certiorari petitions overall are granted (see, e.g. Bonica, Chilton and Sen 2023). This seems to suggest that polarized intercircuit splits are reviewed at a very high rate, compared even to questions over which there is no split. However, we urge caution in this comparison. Ours is the first empirical paper to quantify how likely it is that some particular question will be resolved; others’ results are at the level of the petition.

Table 1. Linear probability model: Which intercircuit splits does the Supreme Court resolve?

Notes: *p < 0.1; **p < 0.05; Number of Circuits is standardized. HC3 robust standard errors shown.

Our findings are consistent with Grant, Hendrickson and Lynch (2012), who observe that the apparent ideological distance between sides in cert petitions alleging splits is larger among a sample of granted petitions than among denied petitions from the 1986–1994 terms. ²³

As for those other factors, we note, consistent with Beim and Rader (2019), splits involving many circuits are more likely to be resolved. Interestingly, splits that evenly divide lower courts are no more likely to be resolved than lopsided splits, and neither are splits with dissents below. Splits in which the Solicitor General filed a petition for cert in a case were much more likely to be resolved than those without Solicitor General involvement. As in Beim and Rader (2019), the likelihood that a split is resolved also varies by issue area, with those involving economic activity more likely and criminal procedure less likely to be resolved than splits in other issue areas. We note also that our measure of polarization seems not to be merely a measure of issue importance or issue divisiveness. The Solicitor General is no more likely to petition for certiorari in polarized intercircuit splits than in less-polarized intercircuit splits. As Table A5 shows, in both a bivariate regression and a regression that includes all our control variables, polarization fails to predict when the Solicitor General petitions for certiorari. The coefficient is negative and is not statistically significant. Polarization does not increase the likelihood that a split has a dissenting opinion, nor that a split has an en banc decision (see Table A6.) Results from these regressions are in the Appendix.

Measure validation

We claim our measure distinguishes those splits that are ideological in a left-right sense, from those that are not. It is reasonable for a reader to wonder if our measure has accurately captured this. As a validation exercise, we test whether Supreme Court Justices themselves polarize when resolving polarized intercircuit splits. That is, we regress polarization among the Justices on the polarization of intercircuit splits. In the Appendix, we present the results of three bivariate linear regressions using split polarization (among resolved splits) to predict polarization among Supreme Court Justices on the final merits vote in those cases that resolved the splits.

Table A7 shows the results of our regressions. The Supreme Court’s decisions are more polarized when resolving polarized intercircuit splits than when resolving less polarized intercircuit splits. This general result is true irrespective of the measure we choose for Supreme Court polarization. In each model, our measure of polarization is positively associated with polarized merits voting, and significant at .101 (analogous measure), .137 (Martin-Quinn), and .019 (connectedness measure). We take this to be reinforcing that our measure is capturing polarization – and perhaps even as an indication that the Supreme Court is itself more polarized when resolving polarizing questions.

Conclusion

The Court is an observer of doctrinal development in the lower courts, looking for legal questions that provide the most attractive opportunities for policymaking. Resolving intercircuit splits over legal questions that have an ideological valence provides such opportunities. These policymaking opportunities allow the Supreme Court to operate simultaneously in jurisprudential mode and outcome mode – answering a pressing legal question that has political implications.

We shift from the extant literature’s focus on the Court as a resolver of individual disputes to a focus on the Court as an executive promulgating legal policies. In resolving legal questions, the Court creates doctrines – judge-made laws – that affect the disposition of future cases and potential or actual real-world disputes. We advance the literature on certiorari by modeling the policy considerations that arise as the Supreme Court makes its docket – which legal questions to address, when, and why.

Our conclusions stand in direct contrast to those of Bonica, Chilton and Sen (2023). They show that the Supreme Court is more likely to grant certiorari when the lawyer for the petitioner is ideologically distant from both the appellate court and the respondent, than when the respondent is ideologically distant from the appellate court and the petitioner. They interpret these results as indicating that the Supreme Court has a preference for cases that do not trigger left-right ideological behavior by the Courts of Appeals. Taken together, our results and theirs suggest a need for continued research about the Supreme Court’s role in an increasingly polarized time.

The politics of circuit splits are of particular interest because these are the legal issues most likely to be addressed by the Court. Scholars typically think this is because the Court wishes to resolve legal ambiguity. We show that (at least some of the time), it’s the Court weighing in on ideology wars in the lower courts.

Specifically, the Court should prefer to review splits over legal questions that are more ideologically polarized. By recording many appellate judges’ votes on the same legal question (indeed, by observing and recording the vote of every judge who heard a particular question), we are able to characterize the ideological valence of legal questions, and the degree to which judges polarize over the choice of competing doctrines. Using a dataset of 119 intercircuit splits, a unique dataset that captures every case that presents a given legal question, we find support for our prediction that the Court is more likely to resolve polarized splits.

Our perspective also has important implications for research on the Supreme Court. Certiorari is about which legal questions the Court will address.

We encourage attention to which level of analysis may be correct in future research on certiorari. Individual certiorari petitions are vehicles for answering legal questions, and the attributes of the specific petition may or may not be relevant to the justices. When a researcher wishes to study which issues get addressed, or when a researcher assumes the Supreme Court cares primarily about policymaking on legal questions, then analysis at the petition level may misspecify the data generating process behind certiorari decisions. The Court must choose from filed petitions – they are limited in their powers by what parties bring. But the Court can read other petitions, and even read decisions and pleadings from cases that never led to a cert petition. This peripheral vision means the Court can see attributes of the legal question – not just attributes of the case. Therefore, an ideal dataset for studying which legal questions are addressed should include these question-level characteristics.

Extant empirical scholarship is about which cases are granted certiorari, and this is a different question in very important ways (see Johnson 2022). As we described in Section 1, existing scholarship understands how the Court sets its agenda by looking only at the ideologies of the deciding panels and the ideologies of the outcomes in particular cases. We show that doing so risks misunderstanding both how ideology is at play in cert decisions and whether ideology is in play.

Furthermore, we find that the justices themselves split along ideological lines on the final merits vote when they resolve polarized splits. That is, ideological decision-making on lower courts sets the table for ideological decision-making on the Supreme Court. The implications of this finding for the development of the law are profound. The data we study are time-bound; in particular, we study a moment in time with a conservative Supreme Court and heterogeneous lower courts. Lower courts are becoming increasingly political through new appointments, as Bonica and Sen (2021) and Kastellec, Cameron and Park (2013) show. As judges become increasingly dissimilar across party lines, intercircuit splits may become both more common and more polarized. It would be fruitful and important to learn whether the Supreme Court’s attention to ideologically polarized intercircuit splits maintains as the ideological make-up of the federal judiciary changes. In any case, that increasing politicization will have many repercussions for American law. Among them, as lower courts become more ideological, so too will the Supreme Court’s docket, the Supreme Court’s decision-making, and doctrine itself.