Predicting Supreme Court Behavior in Indian Law Cases

Grant Christensen February, 2021

Abstract: This piece builds upon Matthew Fletcher’s call for additional empirical work in Indian law by creating a new dataset of Indian law opinions. The piece takes every Indian law case decided by the Supreme Court from the beginning of the Warren Court until the end of the 2019-2020 term. The scholarship first produces an Indian law scorecard that measures how often each Justice voted for the “pro- Indian” outcome. It then compares those results to the Justice’s political ideology to suggest that while there is a general trend that a more “liberal” Justice is more likely to favor the pro-Indian interest, that trend is generally weak with considerable variance from Justice to Justice. Finally, the article then creates a logistic regression model in order to try to predict whether a pro-Indian outcome is likely to prevail at the Court. It finds six potential variables to be statistically significant. It uses quantitative analysis to prove that the Indian interest is more likely to prevail when the Tribe is the appellant, when the issue is framed as a jurisdictional contest, and when the case arises from certain regions of the country. It suggests that Indian law advocates may use these insights to help influence litigation strategies in the future.

It is almost always a mistake to seek answers to Indian legal issues by making analogies to seemingly similar fields. General notions of civil rights law and public land law, for example, simply fail to resolve many questions relating to American Indian tribes and individuals. The extraordinary body of law and policy holds its own answers, which are often wholly unexpected to those unfamiliar with it.¹

Introduction

Shortly after the turn of the last century, while still sitting as a district court judge, Learned Hand queried: “[h]ow long we shall continue to blunder along without the aid of unpartisan and authoritative scientific assistance in the administration of justice”?² Writing in 1911, Judge Hand could not have contemplated the development of empirical research techniques that are now permeating legal discourse. (He was in fact resolving a question of patent infringement.)³ Today, the development of empirical analysis to apply unbiased “scientific assistance” to understand the behavior of courts—and, by extension, justice—is fast becoming a critically important piece of legal scholarship.⁴

This paper applies some of these new empirical tools to understand the Supreme Court’s behavior in cases involving federal Indian law.⁵ Indian law is unusual. It is a body of law that has developed essentially sui generis from common law principles articulated by the courts,⁶ and for that reason it is incredibly difficult to predict. No other area of law has so sharply divided the Court’s recent conservative textualists⁷ and few other areas result in such lengthy litigation.⁸ Indian law is also critically important.⁹ Indian law cases raise important constitutional questions about the separation of powers¹⁰ and, because the extent of a tribe’s authority raises a federal question,¹¹ a disproportionate number of Indian law cases end up in the federal courts. Accordingly, although the total number of cases decided by the Supreme Court has fallen to less than eighty in recent years, the Court continues to hear between two and three Indian law cases each term.¹²

The empirical analysis presented here is divided into two parts. First, a new dataset was created by identifying every Indian law case decided by the Supreme Court from the beginning of the Warren Court in 1953 through July 2020. Each case was coded for whether the interest preferred by Indian tribes was successful on the merits. Indian tribes themselves are not always parties to this litigation and so rather than determine whether the “tribe” won or lost, it is more accurate to look to the pro-tribal or pro-Indian interest. Sometimes this interest is represented by an individual member of the tribe, a tribal business, employee, or governmental subdivision, or even the United States federal government representing the tribe pursuant to its trust responsibility.¹³ Each Justice’s vote was recorded for whether they ultimately supported or opposed the tribal interest. This dataset thus permitted a ranking of Justices in the order of their support for Indian interests and is a new contribution to the Indian law literature. While this analysis is in the realm of descriptive statistics, it is more immediately visual to most readers and challenges the notion that “liberal” Justices are more “pro-Indian” while “conservative” Justices are more likely to oppose Indian interests. Justice Gorsuch, a noted conservative, has the third highest level of support for tribal interests while Justice Stevens, a noted liberal, barely supported the tribal interest a third of the time.

The second part of the analysis then coded each of the 158 Indian law cases for eighteen different independent variables. A logistic regression model was created to try to explain the behavior of the Court and identify what factors might statistically significantly influence the outcome of Indian law cases. The regression model identified six variables that were statistically significant and so substantially advances the understanding of judicial behavior. Insights from this second analysis statistically confirm the more recent bias in the Court against tribal interests, the importance for tribes to frame their legal issues as conflicts between tribal and state jurisdiction, the advantage of being the Appellant in Supreme Court litigation, and potential sympathies the Court displays for cases coming from certain parts of the country.

This paper is broken down into four parts in addition to this introduction. Part 1 introduces Indian law cases as a category. Part II looks at the role of individual Justices. It employs descriptive statistics to determine proportionally how often each Justice voted in favor of the tribal interest and ranks the Justices on the basis of these votes. It makes some preliminary observations about the role of these rankings and examines the application of the Median Justice Theory to Indian law, ultimately suggesting that the median Justice in the Indian law context may not be the middle Justice from the purely ideological perspective used in other models.

Part III provides an extensive discussion of the methodologies employed in the more rigorous logistic regression analysis that follows. The methodology part is roughly broken down into three subparts consisting of documentation for how the population of Indian law cases was culled from the broader list of all Supreme Court opinions, how each of those cases was coded for the dependent variable (whether the tribal interest prevailed on the merits), and a discussion of the eighteen independent variables along with their accompanying null hypotheses.

Part IV interprets the results from the logistic regression. It provides an analysis of both the strength of the regression model as well as tests of each independent variable. It also recognizes the difficulty in interpreting coefficients from a logistic regression and so presents a contrast score to quantify the effect each statistically significant variable has on the success of the tribal interest at the Court. The model itself is able to explain about a quarter of the variance in the success of the tribal interest, which is a robust result and an independent verification of the importance of this paper to the contribution of the study of Indian law and judicial behavior. Finally, Part V provides some brief concluding remarks.

I. Indian Law Cases Are Different

Political scientists have devoted considerable effort attempting to explain the behavior of the United States Supreme Court.¹⁴ They have constructed various theories and models in order to predict each Justice’s voting behavior and the overall outcomes of Supreme Court opinions before they are formally handed down.¹⁵ By studying the joining behavior of Justices over time, this scholarship suggests that the most potent explanation for how a Justice votes is that Justice’s political ideology; i.e., conservative Justices tend to cast ideologically “conservative” votes, while liberal Justices tend to cast ideologically “liberal” ones. While such theories can explain a large portion of Supreme Court outcomes,¹⁶ Indian law opinions are particularly difficult to classify. Rather than falling into traditionally liberal or conservative positions, the sui generis nature of Indian law encourages the development of common law principles, presents complex federalism concerns, and employs a legal interpretation of history that does not perfectly mirror political ideology.

In these traditional models, decisions in favor of a tribe are often grouped in a civil rights category, where the model assumes that liberal Justices are more likely to side with the pro-tribal interest and conservative Justices are more likely to oppose it.¹⁷ Because these models attempt to account for all of the cases over a period of several terms, a natural court,¹⁸ or even a Chief Justice’s tenure,¹⁹ they apparently fail to question their assumption about the political alignment of Justices’ votes for Indian law cases. Professor John Hermann has raised this issue in several scholarly pieces, convincingly arguing that explanations of judicial behavior in Indian law cases require a more nuanced approach than most models use in order to account for the variance in individual voting patterns.²⁰

It is often impossible to use traditional models to predict Indian law cases. Consider a pair of 6-3 decisions where the dissents each consisted of one liberal, one moderate, and one conservative aligned against a majority that was able to unify otherwise ideologically disparate members of the bench. In Kiowa Tribe of Oklahoma v. Manufacturing Technologies , the Court held that a tribe could raise sovereign immunity as a defense to a breach of contract claim, with Justices Stevens, Souter, and Thomas dissenting.²¹ In Mississippi Band of Choctaw Indians v. Holyfield , the Court held that a tribe had exclusive jurisdiction over the child custody placement of the children of tribal members domiciled on the reservation but born in the state of Mississippi.²² Justices Stevens and Kennedy were joined by Chief Justice Rehnquist in dissent.²³ Such unusual joining behavior requires an Indian-law-specific examination into the behavior of the Justices.

A. Indian Law and Judicial Ideology

In order to explore the relationship between a Justice’s political ideology and a Justice’s voting behavior in Indian law cases, this study began by counting the number of pro-Indian votes cast by individual Justices and comparing that behavior to their respective political ideological score based on their voting record across their entire tenure on the Court. The results illustrate the imperfect relationship between ideology and voting in Indian law cases.

The tribal interest prevailed in 73 of the 158 Indian law cases (46.2%) decided since the beginning of the Warren Court. If political ideology were a perfect predictor of Indian law cases, then the most liberal Justices would vote for the tribal interest the largest proportion of the time, while the most conservative Justices would vote for the tribal interest the least often.

To test this observation, it is first necessary to determine how liberal or conservative each Justice’s jurisprudence is. While it might be obvious to people who study the Court that Justice Stevens was “liberal” and Justice Scalia was “conservative,” an empirical study demands a more rigorous approach. Andrew Martin and Kevin Quinn have created an ideological score for each Justice for each term that Justice has served on the bench through the 2018-2019 term.²⁴ Negative scores are associated with a liberal voting pattern while positive scores indicate a conservative one. The absolute value of the score measures how liberal or conservative the voting pattern is, with the larger scores indicating a more extreme ideology. The scores reported below are aggregated for all years a Justice served on the bench from 1953 through the 2019-2020 term to give each Justice a single ideological score.

the land these treaties promised remains an Indian reservation for purposes of federal criminal law. Because Congress has not said otherwise, we hold the government to its word.”²⁵ Justice Gorsuch’s jurisprudence is just the latest example of how the liberal-conservative understanding of the Court breaks down in Indian law cases.

The following table shows how each Justice’s voting patterns in Indian law cases compare with their average ideological score:

Justice Votes in Indian Law Cases

Graphing the pro-Indian vote and the Martin-Quinn scores for each Justice into a scatterplot provides a pictorial representation of the data above and shows a weak negative relationship. Essentially the graph suggests that, despite some outliers, generally as the Justice’s voting record becomes more conservative they are less likely to support the tribal interest. While this is the same trend predicted by general models that suggest that conservative Justices are less likely to support Indian interests because they are related to questions of civil rights, there is a lot of variance among the data points.

Justice Ideology and Indian Law Voting Martin-Quinn Score

In order to determine whether an observed relationship between two variables is significant, a Pearson Product Moment Correlation Coefficient, or ‘r’, can be calculated.²⁶ The correlation coefficient is a statistical measurement of the relationship between two variables expressed between -1 and +1. The larger the absolute value of the coefficient, the more likely the observed relationship is significant. In this case the Pearson coefficient was calculated to test how strongly correlated judicial ideology was with voting in Indian law cases. The Pearson coefficient for the 34 Justice data points presented above is -0.6344. If the seven Justices who decided fewer than ten Indian law opinions are excluded on the basis that the small number of cases might skew the results, the Pearson coefficient is -0.7214.

Each of these results is statistically significant even at the p (cid:149) 0.01 level.²⁷ The Pearson test begins with a null hypothesis that the two variables (ideological score and percent vote in favor of the tribal interest) are not correlated. Even for a dataset with just 26 variables the p (cid:149) 0.01 is significant when the coefficient’s absolute value ‘r’ is greater than 0.496.²⁸ Both 0.6344 and 0.7214 meet this test.

The Pearson coefficient suggests that the null hypothesis should be rejected, indicating that a Justice’s political ideology as measured by the Martin-Quinn score and their voting behavior in Indian law are correlated. The p (cid:149) 0.01 level of significance means that this result is true with a 99% level of confidence or, put another way, there is less than a 1% chance that the observed relationship is random (i.e. not correlated). The net result provides support to the proposition that a Justice’s political ideology is associated with their voting behavior despite the outliers identified above. The next task is providing some context for what this association might mean when it comes to the behavior of the entire Court and not just an individual Justice. How does the observed voting correlation relate to whether the pro-Indian party ultimately prevails at the merits stage by securing at least five votes?

B. Median Justice Theory

Having demonstrated a statistically significant correlation between a Justice’s political ideology and their voting behavior in Indian law cases, it is important to place that conclusion into context. The Median Justice Theory suggests that when the political ideology of the individual Justices is used to predict the outcome of a case, the ‘middle’ or ‘median’ Justice will control the outcome.²⁹ This is really just a matter of judicial math. Assuming that all of the Justices can be arranged along a single political spectrum (from left to right), then the Justice in the middle can control the outcome of the case because they are in the position to provide the fifth vote to either the more conservative four Justices to their right or the four more liberal Justices to their left.³⁰

For much of her time on the Court, Justice Sandra Day O’Connor was presumed to be the median Justice and was therefore in an often prioritized position to control the direction of the Court’s jurisprudence.³¹ However the Median Justice Theory only works if a Justice’s preferred position can be identified clearly on the traditional left-right spectrum. While Justice O’Connor may have held the “median” position on many of the issues facing the Court during her tenure, she was not always the median.³² For example, Justice Kennedy was often perceived to hold the swing vote on freedom of speech questions even if he was otherwise more conservative than Justice O’Connor on other issues before the Court.³³ Who is the median Indian law Justice?

No existing scholarship has applied the Median Justice Theory to the Court’s Indian law decisions. Understanding the development of Indian law through the Median Justice Theory brings new insights to the development of Indian law principles by the modern Court. For example, during the natural court that existed from Justice Breyer’s confirmation in 1994 to Chief Justice Rehnquist’s death in 2005, Justice O’Connor was the median Justice using the Martin-Quinn ideological scores for those terms,³⁴ but Justice Breyer (the third most liberal Justice during the natural court period) was the median Justice for Indian law cases.³⁵

This shift is indicative of the growing hostility by the Supreme Court to Indian interests in recent years.³⁶ From the beginning of the Warren Court in 1953 to the end of the century, Indian interests prevailed in 60 of 125 cases (48%). However, since 2000, the Indian interest has only been successful in 13 of 33 cases (39.4%). At the end of the October 2019 term the Court is among the most divided natural courts on Indian law cases since 1953 with three of the five strongest supporters (Justices Gorsuch, Sotomayor, and Kagan) and four of the five least supportive (Chief Justice Roberts and Justices Thomas, Alito, and Kavanaugh) all currently on the bench. The recent death of Justice Ginsburg and her replacement with Justice Amy Coney Barrett, who has an uncertain record in Indian law cases, will provide an opportunity to test the validity of the claims made in this Article.

What the Median Justice Theory does suggest is that it has become more difficult for the pro-Indian interest to succeed at the Court as the median Justice has become less supportive of Indian rights.³⁷ In 1971 Justice Harlan was the median Justice³⁸ and although he was fairly conservative (with a Martin-Quinn score of 1.62)³⁹ he voted in favor of the tribal interest in 10 of 17 cases (58.8%). In contrast, by this same measure Justice Breyer is the median Justice now.⁴⁰ While his ideological score is less conservative than Justice Harlan (compare -1.23 with 1.62), his support for Indian interests is markedly weaker, voting in favor of the tribal interest in 19 of 48 cases (39.6%).

Interestingly, Justice Breyer is not ideologically the median Justice⁴¹—that distinction belongs to Justice Kavanaugh who has never voted in favor of an Indian interest.⁴² Given that the correlation between ideological score and support for the pro-Indian party is statistically significant, the data suggest that advocates on Indian law cases should target Justice Breyer as the important swing member of their audience through their briefs, amici, and at oral argument. The appointment of Justice Amy Coney Barrett to replace Justice Ginsburg further shifts the ideological balance of the Court. Given the strong pro-Indian voting records of Justices Gorsuch, Sotomayor, and Kagan, Indian law advocates must now carry not only Justice Breyer but a crucial fifth vote to prevail in deeply contested cases.

II. Methodology

The remainder of this Article transitions from analyzing decisions made by individual Justices into an examination of the behavior of the entire Court. It develops and tests a model that explains the outcomes of Indian law cases using logistic regression. Documentation of the methodology is critically important to any empirical work on judicial behavior.⁴³ This Part therefore lays out the methodological decisions taken to construct the dataset that forms the basis of the regression model.

This section is divided into two subparts. The first subpart discusses how cases were selected to ensure that the dataset collected the entire population of Indian law cases decided by the Supreme Court. The second subpart includes a discussion of how each case was coded for the dependent variable (whether the tribal interest won) and for the values of each of the eighteen independent variables which are hypothesized to help explain the Supreme Court’s behavior on Indian law cases. The end result is that the study has identified 158 Indian law cases, determined whether the prevailing party represents the pro-Indian interest, and coded each case for the presence of each of the eighteen independent variables.

A. Determination of All Relevant Cases

Before anything can be said empirically about the Supreme Court’s behavior involving Indian law, a complete dataset needed to be built that includes the population of cases, a determination of whether the Indian interest prevailed in each case (the dependent variable), and codes each of those cases for potentially explanatory binary independent variables. This study has selected the beginning of the Warren Court (October 5, 1953) as its starting point and includes all Indian law cases decided by July 9, 2020.⁴⁴

The modern era of Indian law can be roughly traced to the beginning of the Warren Court.⁴⁵ In 1959 the Court in Williams v. Lee held that Indian tribes have the right “to make their own laws and be governed by them.”⁴⁶ While not necessarily clear at the time it was decided, Williams v. Lee would eventually mark the beginning of what many Indian law scholars now term the era of self-determination,⁴⁷ where federal Indian policy shifted from a concerted attempt to terminate tribal relationships to federal assistance with the promotion of tribal selfgovernance and self-determination.

A complete list of Indian law cases decided on the merits by the Supreme Court is a necessary precursor to a thorough examination of the Court’s voting behavior.⁴⁸ This task is more complicated because while ‘Indian law’ is itself an explicit discipline, by necessity the substantive cases involve questions pulled from fundamentally divergent areas of law: criminal and civil jurisdiction, family law, taxation, water rights, equal protection,⁴⁹ double jeopardy,⁵⁰ sovereign immunity,⁵¹ contract enforcement, casino gaming, trusts, title to real property, and more.

The task is made easier in that several, albeit slightly incomplete, lists already exist.⁵² In order to ensure that this study identified all relevant Indian law cases a Lexis search was performed to identify all Supreme Court cases containing the word “Indian” or “Tribe” from October 5, 1953 (the start of the Warren Court) to July 9, 2020. These results were then culled to eliminate all entries that did not result in a relevant opinion.⁵³ Cases were eliminated if they did not squarely address core Indian law issues. For example, United States v. R.L.C. involved a 16-year-old member of the Red Lake Band of Chippewa Indians stealing a car and causing a car accident that resulted in a fatality on the Indian reservation.⁵⁴ However, the issue before the Supreme Court was whether a juvenile could be detained for a period longer than an adult who committed the same offense.⁵⁵ While an Indian was involved, and the conduct occurred in Indian country,⁵⁶ the question was not an Indian law question and accordingly the case was not included in this dataset.

After culling the cases from Lexis, the list of cases was compared with existing lists of Indian law cases maintained by Matthew Fletcher,⁵⁷ by John Hermann,⁵⁸ and collected in a comprehensive appendix by Grant Christensen.⁵⁹ This resulted in the addition of one case involving Native Hawaiians.⁶⁰ The databases also referenced several cases that, upon further evaluation, are not included here. For example, the Fletcher index includes cases that were pulled by the parties before a final decision was reached,⁶¹ affirmed by an equally divided court,⁶² and a case involving the hunting of moose by hovercraft in Alaska but not implicating Indian law.⁶³ None of these cases required votes on the merits of core Indian law issues and are therefore omitted from this dataset.

So what is an ‘Indian law’ case? Broadly speaking, the world of Indian law cases includes those cases that address the relationship between tribes and either states or the United States, cases that interpret an Indian treaty or a statute which singles out Indians or Indian tribes for special treatment, and cases with legal issues that affect Indian reservations. Additionally, cases dealing with the legacy of colonialism and the status of Native Alaskans and Hawaiians are also included, as are cases between a member Indian and their tribe that raise federal questions. Essentially, to be classified as an Indian law case the case needs to do more than simply have an Indian party; the legal question must raise issues unique to Indian tribes or Indigenous people.

The final dataset contains 158 Indian law cases. Just four of those cases were decided before Williams v. Lee and so the Supreme Court has decided on average 2.6 Indian law cases every year from 1959-2020.⁶⁴ Despite a reduction in the total number of cases decided by the Supreme Court in recent years, the Court’s interest in Indian law has remained strong. For example, the Court scheduled three Indian law cases during the 2017-2018 term⁶⁵ even though it decided fewer than eighty cases total.⁶⁶

B. Coding the Variables

Having identified a complete list of cases, the next step was to code those cases for use in an empirical inquiry. While the data could arguably be coded to test many different kinds of judicial behavior, this study is trying to identify which factors are statistically significant predictors of the outcome in Indian law cases. Accordingly, each case was coded for the dependent variable: whether or not the pro-Indian interest prevailed on the merits.

Each case was also coded for eighteen independent variables. A null hypothesis is reported suggesting that each of the variables does not impact the outcome of Indian law cases at the Court and the logistic regression model will test to see if we can confidently reject each null hypothesis. While the rejection of a null hypothesis does not prove causation between the independent and dependent variables, it is valuable to identify which variables are statistically significant explanators of the Indian law outcome within the model.

1. The Dependent Variable: A Pro-Indian Outcome

The dependent variable is whether the Supreme Court ultimately reached the outcome urged by the tribe or was otherwise protective of the tribal interest. The dependent variable is coded as a binary “0” or “1” for each case, with a “0” being associated with a decision that went against the tribal interest and a “1” being an opinion that held in favor of the legal interpretation urged by the tribe.

In many cases the tribal interest is easy to identify. A majority of cases in the dataset include a tribe, or one of its members, participating directly as a named party.⁶⁷ Other cases have the United States, pursuant to its trust obligation,⁶⁸ arguing on behalf of the tribe. In these cases, decisions in favor of the tribe (or in favor of the United States representing the tribal interest) are obviously representative of a pro-Indian outcome and coded as “1.” In the small number of cases where the dispute was between a tribe and one of its members, the tribal interest was coded as the outcome urged by the tribe itself.⁶⁹ In this manner each case was coded with a binary dependent variable, and it is this variable that the regression model attempts to explain.

2. The Independent Variables

So how might a model explain whether the tribal interest wins or loses? In order to evaluate and explain the dependent variable, each case was also coded for eighteen independent variables, each attached to a possible theory/explanation for why the factor may alter the outcome from the Court. Each independent variable was also binary, with each case having the variable coded “0” or “1” following the general rules described below.

Also associated with each independent variable is a null hypothesis. In each case the null hypothesis suggests that the independent variable will not have an effect on the outcome. Logistic regression analysis tests each of these independent variables in order to determine when that null hypothesis can be rejected.⁷⁰ Those independent variables where the null hypothesis is rejected at a level of statistical significance form the basis for the discussion of the results that follows in Part IV.

a. The Tribe as the Appellant

Tribal Appellant: (TribeAppellant). It has long been true that the appellant wins more often than the appellee at the Supreme Court.⁷¹ This makes sense when one considers that Supreme Court decisions are really the product of two votes. The Supreme Court controls which cases it hears through the process of granting certiorari, where it takes four votes from the Justices in order to add the case to the Supreme Court’s docket.⁷² Professor Matthew Fletcher has written an exceptional piece on the behavior of Indian law cases at the certiorari stage, ultimately concluding that many Indian law cases which were arguably wrongly decided by the lower court are not granted cert because they are “factbound”⁷³ and/or “splitless.”⁷⁴

Moreover, much empirical work has established that Justices employ strategic voting at the certiorari stage.⁷⁵ Justices are most likely to vote to hear a case when they believe that the case below was wrongly decided and that their fellow Justices will agree to reverse.⁷⁶ This strategic behavior results in the appellant winning almost two-thirds of all Supreme Court cases that reach the merits⁷⁷ because judges are more likely to vote to hear a case when they believe that the lower court erred in its interpretation of the law and that there exists a majority for overturning the decision. Even when a Justice believes the lower court wrongly decided the case, the Justice may not vote to hear the case if he or she does not believe a majority of the Court would vote accordingly.⁷⁸

Ho: Appearing as the appellant does not change the

likelihood the tribal interest succeeds at the

Supreme Court

Given the success of the appellant generally, and the existence of strategic voting at the gatekeeping stage of appeal, the model tests whether the outcome of the Supreme Court’s Indian law cases might be explained by whether the tribe acts as the appellant when the case is petitioned to the Court. Cases are coded as “1” when the tribe is the appellant and “0” if any other party is the appellant. Cases where the appellant is representing the tribe’s interest, but is not the tribe itself, are still coded “0”.⁷⁹ All cases arising under the Court’s original jurisdiction are also coded as “0” because even though no party ‘appeals’ from a lower court judgment, Indian tribes do not qualify for the Court’s original jurisdiction and therefore could never be considered the petitioning or appealing party.⁸⁰

b . The Solicitor General in Support of the Tribe

Solicitor General Supporting the Tribe’s Position: (SCProTribe). The Solicitor General is responsible for representing the United States before the U.S. Supreme Court, and “has long exerted significant influence over both the Court’s case selection decisions and its substantive decisions on the merits,” supporting the winning party almost 90% of the time during the early years of the Roberts Court.⁸¹ The power of the Solicitor is so great that he or she is sometimes referred to as the “Tenth Justice.”⁸² The Solicitor General is given “great leeway” to determine whether to participate when the United States is not itself a party,⁸³ and when it chooses to participate it sends a signal to the Court about the merits of the claim.⁸⁴

Ho: Having the Solicitor General appear on behalf of the tribal interest does not change the likelihood the

tribal interest succeeds at the Supreme Court

The model tests whether the Solicitor’s participation in favor of the tribal interest may affect the outcome of Indian law cases. The model codes each case as “1” if the Solicitor General participated in the case and supported the tribal interest and “0” for all other cases. The Solicitor’s participation was determined by examining whether the Solicitor appeared at oral argument. The case does not code for whether the Solicitor may have submitted a brief or responded to a call from the Court for the views of the Solicitor. The case is coded as “1” whenever the Solicitor General appeared to support the pro-Indian position regardless of whether the tribe was the appellant or the appellee.

c. The Solicitor General Opposing the Tribe

Solicitor General Opposing the Tribe’s Position: (SGNegTribe). For the same reasons described above it is possible that the Solicitor General’s participation opposing the tribe’s interest might be an important signal to the justices about the stakes of the case and the strength of the tribe’s argument. The Solicitor is particularly likely to participate when the tribe or an individual Indian brings suit against the United States itself.⁸⁵ Even together these two variables do not cover all cases in the dataset. In 79 of the 158 cases (exactly half) the Solicitor did not participate at all.

Ho: Having the Solicitor General appear in opposition to the tribal interest does not change the likelihood the

tribal interest succeeds at the Supreme Court

The model tests whether the Solicitor’s participation against the tribal interest may affect the outcome of Indian law cases. The model codes each case as “1” if the Solicitor General participated in the case against the tribal interest and “0” for all other cases. The Solicitor’s participation was determined by examining whether the Solicitor appeared at oral argument. The case does not code for whether the Solicitor may have submitted a brief or responded to a call from the Court for the views of the Solicitor.

d. The Number of Justices Deciding the Case

A Decision Made by Fewer Than Nine Justices: (ShortCourt). Supreme Court decisions may be impacted by a vacancy on the bench, whether created by a recusal or after the retirement or death of a sitting Justice and pending the confirmation of a replacement. Professor Jeffrey Stempel has explained that this problem is particularly acute at the Supreme Court level because further review of a decision by a panel unfamiliar with the issue is impossible.⁸⁶ The Court is not unaware of the problem, it has ordered cases to be reheard during a following term in an attempt to ensure a full complement of Justices can hear particularly contentious cases.⁸⁷ Professor Justin Pidot adds that vacancies create challenges to the institution of the Court, and raise the specter of tied votes which fail to provide definitive guidance on important legal questions.⁸⁸ This is no hypothetical matter for Indian law. Justice Scalia’s unexpected death resulted in a 4-4 tie on an important question of tribal court jurisdiction in 2016;⁸⁹ Justice Kennedy worked on a case involving tribal treaty rights forty years ago when he sat on the Ninth Circuit and had to recuse himself when a version of the case reached the Supreme Court in 2018,⁹⁰ resulting in a 4-4 tie; and Justice Gorsuch’s previous vote not to hear a case en banc when he was on the Tenth Circuit resulted in a 4-4 tie in a diminishment case in 2019.⁹¹ That case was scheduled for reargument, and then the Court granted and decided a related case raising the same legal issue but where Justice Gorsuch was not recused, resulting in a 5-4 victory for the Muscogee Nation.⁹²

Ho: Having the case decided by a panel of fewer than nine justices does not change the likelihood the tribal

interest succeeds at the Supreme Court

The model tests whether a bench with less than a full complement of nine Justices may play a particular role in determining the outcome of Indian law cases by coding for whether the opinion was decided by fewer than nine Justices. Admittedly this will miss those cases that ended in 4-4 ties since those cases were affirmed by an equally divided Court.⁹³ The data is coded “1” for each Indian law case where fewer than nine Justices voted and “0” if all nine Justice’s had recorded votes. The coding of this variable was not impacted by whether the decision was unanimous or fractured. A 5-4 opinion and a 9-0 opinion are each coded as “0” because a full court of nine Justices voted while an 8-0 or a 4-3 opinion would each be coded “1.” In no Indian law case during this period did fewer than seven justices cast a vote. 4-4 opinions were omitted from the dataset entirely as they provided no precedential value nor did they generate opinions.⁹⁴

e. Indian Gaming and the Rise of the “Rich Indian” Stereotype

A Decision Made After Indian Gaming Was Implemented: (AfterGaming). The development of casino gaming on Indian reservations has fundamentally changed some tribal communities by injecting resources and new business opportunities.⁹⁵ The tribal casino gaming industry emerged after a Supreme Court decision in California v. Cabazon Band of Mission Indians held that states could not regulate tribal gaming activity occurring in Indian country.⁹⁶ Indian gaming has brought wealth to some Indian communities but has also encouraged the development of a new stereotype—the ‘rich casino Indian.’⁹⁷ Could the development of this new stereotype change the Court’s approach to Indian law cases?

Professor Matthew Fletcher suggests that an important inflection point occurred around the Cabazon decision.⁹⁸ “From 1959, the generally recognized beginning of the modern era of federal Indian law, to 1987, when the Supreme Court decided the major Indian gaming case California v. Cabazon Band of Mission Indians , Indians and Indian tribes… won nearly 60% of federal Indian law cases decided by the Supreme Court. But since Cabazon , tribal interests have lost more than 75% of their cases.”⁹⁹ Indian success has improved a little since Fletcher’s piece was first published, but he has clearly recognized a critically important change in the Court’s behavior.¹⁰⁰

Ho: Having the case decided after Cabazon does not

change the likelihood the tribal interest succeeds at the

Supreme Court

The model tests whether the development of Indian gaming, and new concomitant attitudes about Indians generally, may impact the outcome of Indian law cases by coding for whether the opinion was decided before or after the Cabazon decision was handed down.¹⁰¹ The variable is coded “1” for each case decided after Cabazon and “0” for each case decided before it. This variable is coded binomially and does not attempt to represent how many years away from the 1987 opinion the case was decided. Instead it tests whether 1987 was an inflection point in Indian law at the Supreme Court. Cabazon itself is coded “0” because the effects of Indian gaming hypothesized here could not have occurred until after the decision was handed down. Admittedly, the reason for changing outcomes in Indian law cases after 1987 may have any number of factors (as discussed in more detail in the analysis part) but scholars point to 1987 as an inflection point because of the change in the economic circumstances of many tribes.

f . The Origins of the Appeal

An Appeal from the State Court System: (AppFromState). The appellate jurisdiction of the United States Supreme Court permits appeals from both state and federal courts.¹⁰² While the basis for judicial review of a lower court opinion finds its origins in Article III, Congress has codified different statutory provisions for appeals from the federal and state court systems.¹⁰³ Professor David Schlueter suggests that the Supreme Court proceeds more cautiously when reviewing decisions from state courts, occasionally deferring to state supreme court judgments.¹⁰⁴ Most Indian law questions present federal questions which permit litigation in federal courts,¹⁰⁵ but questions of contract enforcement, state taxation, criminal jurisdiction, and family law involving Indians may originate in state court¹⁰⁶ and result in a direct appeal from the state court system to the United States Supreme Court.¹⁰⁷

Ho: Appealing the case from a state court proceeding instead of from a federal proceeding does not change

the likelihood the tribal interest succeeds at the

Supreme Court

The model tests whether the Supreme Court treats appeals from state courts differently than appeals from federal courts when answering Indian law questions. The data is coded “1” if the case was decided by state courts and then appealed to the Supreme Court and “0” if the case arrived at the Supreme Court through other means. Cases arising under the Supreme Court’s original jurisdiction were coded “0” because they were not decided by a state court.¹⁰⁸ Importantly, not all of the state court cases were resolved by a state’s highest court.¹⁰⁹ In several cases, the state supreme court refused to hear the appeal from a state trial or appellate court and so the case that was actually appealed to the Supreme Court was a lower state court opinion.¹¹⁰ These cases are all coded “1” because while they may not have been decided by a state’s highest court they are appeals from the state system.

Was a State Sovereign Involved in the Litigation

A State Participating as a Party: (StateParty). A body of political science scholarship suggests that when a state government participates as a party in front of the Supreme Court it is disproportionately likely to be successful.¹¹¹ At least one scholar has attributed this success to the fact that states are repeat players before the Court and therefore have learned from experience how to improve their chances,¹¹² while others suggest that the higher rates of success are attributable to the development of particular divisions to handle appellate work which are capable of accumulating appellate expertise.¹¹³ This paper makes no attempt to explain the relative success of a state as a party to litigation, it merely asks whether the Indian interest is less likely to prevail when it is being opposed by a state entity.

Ho: Having a state party participate in the litigation

does not change the likelihood the tribal interest

succeeds at the Supreme Court

The model tests whether the Supreme Court treats Indian law cases differently when a state is a party to the litigation by coding for state participation. The data is coded “1” if the state was a named party directly participating in the litigation and “0” if the state was not a party. For purposes of this variable the term ‘state’ includes state agencies or other state owned/governed entities which would benefit from having the state’s attorney participate directly on their behalf. A common example of this in Indian law cases were state tax commissions,¹¹⁴ but also included human resource offices, state prisons, and environmental officers.

h . Tribal-State Relations and the Question of Jurisdiction

Question of State v. Tribal Jurisdiction: (StateJurQuest). Many Indian law cases directly confront jurisdictional questions centered around the interaction of competing tribal, state, and federal powers.¹¹⁵ The Court has often used these interactions to opine on issues of federalism,¹¹⁶ with important cases on the Eleventh Amendment¹¹⁷ or interpretations of due process and equal protection¹¹⁸ emerging from Indian law. Professor Matthew Fletcher suggests that in this way the Supreme Court may use Indian law cases in order to reach other important constitutional questions; “From the view of a national decision maker such as a Supreme Court Justice, there is much more to a simple Indian law case than a dispute between Indians, Indian tribes, and the non-Indian individuals, governments, and entities that oppose them. There are questions of equal protection, due process, federalism, jurisdiction, congressional and executive power, and more. Indian law disputes often are mere vessels for the Court to tackle larger questions.”¹¹⁹ Moreover, tribal resources (from casinos to water or natural resources) provide a large potential pool of revenue over which states regularly attempt to assert their authority.¹²⁰ At least one paper has suggested that individual Justice voting behavior could be influenced by the jurisdictional questions inherent in the conflict between tribes and states.¹²¹

Ho: Questions requiring the interpretation of state

jurisdiction are no more likely to result in a decision

favorable to the tribal interest than any other

legal issue

The model tests whether cases involving a dispute between state and tribal jurisdiction might be sufficiently unique to help explain the Supreme Court’s behavior when resolving Indian law cases. Each case was read carefully to determine whether the resolution of the certified question(s) required the determination of a jurisdictional question between states and tribes. Functionally this variable looks at those cases where a state and tribe are contesting their authority in Indian country. Cases where the state is attempting to tax, to impose its hunting and fishing regulations, to criminally prosecute, or to decide the custody of an Indian child are all examples of where a jurisdictional dispute is present. Recognizing that this is still an amorphous category, the coding of these cases was aided by additional verification from John Hermann and Karen O’Connor who recognized “procedure/jurisdiction” cases were one of the four most common categories of Indian law cases decided during the Burger and Rehnquist courts.¹²² Cases that did raise this question were coded “1” and all other cases were coded “0”.

i. Geography: Where the Case Originated

Tribes are not homogeneous. Differences in tribal culture, history, law, policy, religion, etc. may precipitate different kinds of legal claims. Tribes across the great plains were often subject to allotment,¹²³ while the Pueblos of New Mexico are notable for holding their lands communally through land grants they can trace back to Spain.¹²⁴ I created several variables to test whether the geography of a claim effects the outcome.

Tribes of the Pacific Northwest: (PacNW). This variable is specific to claims from tribes in Washington State and Oregon. The Pacific Northwest has been singled out in Indian law because of its shared treaty history and interrelated legal issues connected to the Columbia River.¹²⁵ The Stevens Treaties from the 1850s were negotiated with many of these bands in attendance and share some common characteristics.¹²⁶ Tribes in this region also have a unique relationship to contested natural resources (particularly salmon) and so form a distinct interest group.¹²⁷

Tribes from Oklahoma: (Okla). This variable asks whether cases from Oklahoma are treated differently than other cases. Oklahoma was originally the “Indian Territory” where tribes from across the country were moved during the mid-1800s.¹²⁸ Because of its status as the Indian Territory it was among the last regions of the country to be formally conferred statehood, and continues to have among the highest populations of American Indians.¹²⁹ The Five Tribes (Cherokee, Seminole, Choctaw, Chickasaw, and Muscogee (Creek)) are all located there among a total of 38 federally recognized tribes.¹³⁰ It is possible the Court might consider cases coming from Oklahoma differently given its unique American Indian history.

Tribes from the Northern Plains: (NPlains). Tribes across the Northern Plains speak different languages and share different cultures but this was the last region of the continental United States to have Indian claims settled.¹³¹ The last time the United States negotiated a treaty with an Indian tribe as the losing party was the Treaty of Fort Laramie in 1868—where the United States asked the Indians for peace.¹³² Given its less hospitable climate, Indians were pushed to the fringes of this region. I have used a definition of Northern Plains that comes from the Indian Health Service.¹³³ This variable includes all claims that come from tribes located in Michigan, Wisconsin, Minnesota, Iowa, North Dakota, South Dakota, Nebraska, Wyoming, and Montana.¹³⁴

Tribes from the Southwest: (Southwest). Although I recognize there is a great variation of tribal languages and cultures I have grouped tribes in Arizona and New Mexico together. This region includes some of the most populous reservations¹³⁵ as well as the Pueblos.¹³⁶ Both Arizona and New Mexico have comparatively large proportions of their population who identify as Indigenous and their shared desert climate often links these states together in the American mind.¹³⁷

Tribes from California: (Calif). California has the largest number of federally recognized tribes in the continental United States at 109.¹³⁸ It also has a large population of urban Indians¹³⁹ and many tribes have benefitted from casino gaming.¹⁴⁰ It is worth examining whether appeals from these communities might be treated differently by the Court given the unique status of California as home to so many American Indian tribes.

Indigenous People of Alaska/Hawaii Noncontiguous States: (Noncontig). Alaska and Hawaii actually did not become states until after the beginning of the Warren Court.¹⁴¹ Felix Cohen’s¹⁴² Handbook of Federal Indian Law¹⁴³ carves out Alaska and Hawaii as being treated uniquely given the complicated history of their acquisition. Alaska was purchased from Russia with little thought about the rights of the Native Alaskan people who have lives there since time immemorial,¹⁴⁴ while Hawaii was its own Indigenous nation-state until it was conquered by the United States at the end of the nineteenth century.¹⁴⁵ These noncontiguous states have incongruous histories which may change the legal basis under which their Indigenous citizens claim rights. There are no treaties with Native peoples from either state and much of the common law developed by the courts and applicable to Indians on the mainland has been supplanted by statute.¹⁴⁶

Ho: Cases that involve Indians and Indian tribes from

the Pacific Northwest are no more likely to prevail

than Indian law cases in general

Ho: Cases that involve Indians and Indian tribes from Oklahoma are no more likely to prevail than Indian

law cases in general

Ho: Cases that involve Indians and Indian tribes from the Northern Plains are no more likely to prevail than

Indian law cases in general

Ho: Cases that involve Indians and Indian tribes from

the Southwest are no more likely to prevail than

Indian law cases in general

Ho: Cases that involve Indians and Indian tribes from

California are no more likely to prevail than Indian

law cases in general

Ho: Cases that involve Indigenous people from Alaska

or Hawaii are no more likely to prevail than Indian

law cases in general

The model tests whether the region of the country from which a case arises changes its outcome. Each of the 158 cases was coded for where the issues in front of the court originated. This is notably different than the procedural history of a case because many Indian law cases involved federal agencies and were therefore appealed from the D.C. Circuit Court of Appeals,¹⁴⁷ or involved claims against the United States and so originated in the Court of Claims,¹⁴⁸ or claims filed for breach of trust in the Federal Circuit,¹⁴⁹ or took advantage of the Supreme Court’s original jurisdiction.¹⁵⁰ Each case was coded “1” if it originated in the target region and “0” if it arose elsewhere.

There was one case involving Native Alaskans that was decided after Chief Justice Warren was appointed but before Alaska became a state.¹⁵¹ It was coded “1” for the ‘noncontig’ variable in recognition that it raised questions of land rights and aboriginal title to Indians in Alaska which were unrelated to the question of statehood. One additional case, Morton v. Mancari ,¹⁵² involved a Bureau of Indian Affairs (BIA) policy to provide preference to American Indians in the promotion process within the agency.¹⁵³ This case did not really arise from any one tribe, although the litigation was appealed to the Supreme Court from New Mexico. Because the policy originated with the Bureau of Indian Affairs and was not tribe specific I coded it as if the case emerged from Washington D.C.

j. The Federal Appellate Courts

Cases from the Eighth, Ninth, and Tenth Circuits: (Eighth) (Ninth) and (Tenth). Indian tribes are not evenly distributed across the United States. Almost all American Indian reservations are located in just three federal appellate circuits: the Eighth, Ninth, and Tenth.¹⁵⁴ Moreover, political scientists have shown that the Supreme Court allocates its review of appellate decisions disproportionately—being more likely to review and to reverse some circuits more often than others.¹⁵⁵ Other scholars have noted that appellate courts may “shirk” from following Supreme Court precedent at the margins in order to advance their own ideological agendas.¹⁵⁶ The Supreme Court’s review of appellate court behavior is thus an important check on the independence of those courts.¹⁵⁷ Given that different appellate courts are generally treated differently by the Supreme Court, the model tests whether that is true for Indian law cases.

Ho: Appeals from the Eighth, Ninth, and Tenth

Circuits are no more likely than other Supreme Court

cases to result in a victory for the pro-tribal interest

The coding of these variables is simple. When a case was appealed from the Eighth Circuit it was scored as “1” for the ‘Eighth’ variable and all other cases were scored as “0”. This same process was followed for the variables ‘Ninth’ and ‘Tenth’. There was one case which consolidated appeals from both the Federal Circuit and the Tenth Circuit.¹⁵⁸ It was scored as a “1” for the Tenth Circuit because one of the cases was decided by the Tenth Circuit, and a “0” for both the Eighth and Ninth Circuits. The model did not independently test cases from the Federal Circuit, but if it had I would have coded the case “1” for the Federal Circuit as well.

k. En Banc Review

En Banc Review: (En banc). Less than one half of one percent of all federal appellate court cases are decided en banc (i.e. by the entire set of active federal appellate judges).¹⁵⁹ In contrast, 14 of the 158 Indian law cases (8.9%) decided by the Supreme Court were appealed from en banc consideration. Discounting appeals from state courts and cases accepted under the Court’s original jurisdiction and the proportion becomes even higher (14 of 112 or 12.5%). Political scientists have two theories about en banc opinions; some argue the Court is less likely to hear cases decided en banc because those cases have the benefit of review by a larger panel while others suggest that the acceptance of a case en banc might serve as a signal of the importance of the case and thus encourage Supreme Court review.¹⁶⁰

Ho: Appeals to the Supreme Court from an en banc

review are no more likely than any other case to result

in a pro-tribal outcome

This variable was very easy to score. Cases that were decided by an en banc panel before being granted certiorari to the U.S. Supreme Court were scored “1” and all other cases were scored “0”. Cases appealed from the state court system and cases decided in the first instance by the Supreme Court pursuant to its original jurisdiction were also scored “0” because they were not decided by an en banc panel, although admittedly it would have been impossible for them to be heard en banc given the nature of their procedural history.

III. Results

Having documented the choices made in the model and the hypothesized relationship between the variables, this part lays out the model’s results. It begins by sharing the raw data resulting from the logistic regression model and then provides discussion divided into three parts.

The first part discusses how logistic regression works and how to interpret the values presented in the results. The second part evaluates the fitness of the model itself to determine whether the variables identified explain the Supreme Court’s behavior in Indian law cases. The third part examines the independent variables tested in the model and suggests that the independent variables with the most statistically significant relationship to the dependent variable are; 1) whether the tribe is the appellant party, 2) whether the case was decided after the Supreme Court set the baseline for Indian casino gaming, 3) whether the case involved a question of state jurisdiction, and 4) whether the case originated from Oklahoma, the Northern Plains, or the Southwest.

The outcome of the logistic regression is reported here:

Std. P >

Ind. Variable Coef. Err. Z |z|

TribeAppellant 1.344 0.488 2.76 0.006 SCProTribe 0.569 0.506 1.12 0.261 SGNegTribe 0.680 0.604 1.13 0.260 ShortCourt 0.065 0.549 0.12 0.905 AfterGaming -1.015 0.440 -2.31 0.021 AppFromState -0.036 0.673 -0.05 0.957 StateParty 0.171 0.526 0.33 0.745 StateJurQuest 1.260 0.521 2.42 0.016 PacNW 0.655 0.682 0.96 0.337 Okla. 1.566 0.765 2.05 0.041 NPlains 2.024 0.695 2.91 0.004 Southwest 1.669 0.619 2.70 0.007 Calif. 0.152 0.838 0.18 0.856 NonContig. -0.916 1.233 -0.74 0.458 Eighth -1.056 0.863 -1.22 0.221 Ninth 0.341 0.606 0.56 0.573 Tenth 0.548 0.687 0.80 0.425 Enbanc 0.444 0.685 0.65 0.517 Constant -2.468 0.829 -2.98 0.003

No. of Obs. 158

Prob. > Chi^2 0.0000

R^2 0.2452

A. Brief Discussion of Empirical Metrics

This model employs logistic regression in order to measure which variables help explain the Supreme Court’s behavior in Indian law cases. Logistic regression was chosen because the dependent variable is binary (the Indian interest either won or lost at the Supreme Court on the merits). Because the variance between binary observations is homoscedastic,¹⁶¹ Ordinary Least Squares (OLS) regression is inappropriate.¹⁶² Unlike OLS, logistic regression is specifically designed to model binary data and so provides the best picture of how the independent variables described above explain (or fail to explain) the Supreme Court’s Indian law jurisprudence.¹⁶³

The results of the logistic regression test each null hypothesis identified in Part III. For each independent variable; a coefficient, standard error, z-score, and p-value ( P > |z| ) are calculated. The z-score is a test of statistical significance that determines whether or not the null hypothesis can be rejected. Z-scores measure standard deviation. The further the z-score is from zero, the more significant the independent variable. Whether the z-score is positive or negative is a reflection of the relarelationship between the independent and dependent variables. In this model a negative z-score would imply that if the independent variable is present, it decreases the likelihood of the tribal interest succeeding on the merits while a positive z-score increases that likelihood.

The coefficient is an indicator of how much of an effect a one unit change in the independent variable has on the dependent variable. Because the dependent variable and each of the independent variables are binary these coefficients can be measured against each other to provide a gauge of relative or comparative importance. The coefficient’s sign always matches the z-score, where a negative coefficient means the presence of the independent variable reduces the likelihood the tribal interest prevails at the Supreme Court and a positive coefficient increases that likelihood.

However the z-score and the coefficient alone are insufficient to interpret the data. The p-value takes the z-score and provides a probability that the null hypothesis was falsely rejected. In other words the p-value provides the degree of confidence that the reported result is actually the result of the independent variable operating on the dependent variable instead of an observation observed by chance. The smaller the p-value the greater likelihood that the reported relationship is meaningful and that the null hypothesis can be rejected. The results reported here will accept a 95% level of confidence and discuss in more detail those relations that are statistically significant at a p-value of 0.05 and below.

B. Evaluating the Logistic Regression Model

Before reviewing which variables are statistically significant in explaining the behavior of the Supreme Court in Indian law cases it is important to evaluate how robust the model ultimately is. First, the number of observations is 158. Logistic regression only works if each case has a value for every variable tested by the model. Any independent variable without a value would be omitted when reporting the results of the data. The dataset contains 158 cases and the number of observations used in the model is equal to 158. This indicates that every identified case has a value for each independent variable tested and so no cases were omitted when generating the model.

Taken together the model does a respectable first attempt of modeling the Supreme Court’s outcome in Indian law cases. The data taken together is robust. This represents an important step forward in the understanding of Supreme Court behavior.

C. Explaining Supreme Court Behavior in Indian Law Cases

While the results of the logistic regression provide coefficients and contrast values for each of the eighteen independent variables, the relationship is not statistically significant for twelve of them and therefore we cannot reject the null hypothesis associated with those variables.¹⁶⁴ We therefore reject the hypothesis that en banc decisions, cases coming from the Pacific Northwest, California, or noncontiguous states, the Solicitor general’s participation, a court with fewer than nine voting members, appeals from state courts, the existence of a state party, or the specific appellate court have a statistically significant impact on whether the court reaches a pro-Indian outcome. For these variables we cannot reject the null hypothesis.

The remaining six variables are statistically significant. The results for these six variables are discussed below in the order of the confidence with which the null hypothesis can be rejected. These results provide insight into the behavior of the Supreme Court that is useful for both the scholar and the practitioner. They suggest that when trying to obtain an outcome at the Court that favors a tribal interest, advocates should seek cases where the Court is willing to grant a case when the Tribe is the appellant, try to frame the issue as one raising a jurisdictional question, target appeals that come from tribes located in Oklahoma, the Northern Plains, or the Southwest, and be generally cautious about taking cases to the Supreme Court in the post- Cabazon era.

Cases Where the Tribe is the Appellant: (P-Value of .006). The model provides support for Professor Matthew Fletcher’s suggestion that because the appellant wins in a majority of cases in which the Court grants certiorari, Indian tribes should win more often when they are the appealing party.¹⁶⁵ An Indian tribe was the appellant in 66 of the 158 cases (41.77%). For the first time it is possible to quantify the benefit of having a tribal appellant.

Admittedly, these results provide no insight into how the Court decides to grant certiorari, which is a condition precedent to the resolution of the matter on the merits.¹⁶⁶ Moreover, when the tribe wins at the appellate level, it lacks the ability to appeal its case to the Court and therefore tribes often have little control over whether they are the appellant when they have not lost in the proceeding below.¹⁶⁷ However for Supreme Court observers, especially those tasked with identifying or predicting a winner before a case is finally resolved, knowing that the tribal interest is more likely to prevail when an Indian tribe is the appellant is certainly useful.

Cases Requiring the Resolution of a Question of State Jurisdiction: (P- Value of .016). The model suggests that when the Supreme Court is presented with questions of conflicting jurisdiction between tribes and states, all other things being equal the tribal interest is more likely to prevail. This conclusion is consistent with almost two centuries of Supreme Court precedent where the Court has considered its role to protect tribes from states.¹⁶⁸ Remarkably, questions of jurisdiction were present in 77 of the 158 cases collected in the dataset (48.73%).

The high level of statistical significance is particularly noteworthy. The model cannot tell us why the Court seems more amenable to cases framed as jurisdictional struggles. Perhaps, as Professor Robert Williams suggests, the Court is playing its role in mediating conflicts of competing sovereigns in order to protect Indian tribes from the excesses of the state.¹⁶⁹ Certainly the Supreme Court has a legacy of being solicitous to tribal concerns when they are threatened by states. From Chief Justice Marshall writing in 1831, “If courts were permitted to indulge their sympathies, a case better calculated to excite them can scarcely be imagined”¹⁷⁰ to Justice Gorsuch in 2020 “[o]n the far end of the Trail of Tears was a promise,”¹⁷¹ the Court has long viewed its role as protecting tribes from states.

Tribal advocates are advised to consider framing their issues to the Court as questions of state encroachment on tribal jurisdiction. The framing of legal issues matters to the Court. The data suggests that if tribes can get the Supreme Court to frame the issues in Indian law cases as being jurisdictional in nature, the tribal interest is more likely to prevail, especially when that jurisdictional issue pits an Indian tribe against a state.

Cases originating from the Northern Plains (P-Value of .004), Southwest (P-Value of .007), or Oklahoma (P-Value of .042) . These variables are interesting. Recall that the pro-Indian interest won 46% of the 158 cases decided since 1953. Thirty-eight of the 158 cases came from the Northern Plains and the tribal interest prevailed in 23 of them (60.5%). Thirty of the 158 cases from the Southwest and the tribal interest prevailed in 17 of them (56.6%). Sixteen of the 158 cases originated from tribes in Oklahoma and the tribal interest prevailed in 10 of them (62.5%). Tribes in these regions performed considerably better than cases arising from tribes in other states. This is valuable information from a purely descriptive level but becomes even more informative when applied to a logistic regression model.

The regression model asks whether the observed effect of greater success might be attributable to something other than the region of the country. For example, could the Solicitor General’s participation, or the jurisdictional nature of the cases, or the fact that the cases were appealed from state or federal court explain the change in outcome. The model rejects with a greater than 95% level of confidence that any of the other independent variables provides an explanation for the more successful outcome in cases that arise from these states.

Admittedly the model captures only about a quarter of the variation in the Supreme Court’s ultimate behavior in Indian law cases and so there may be other uncoded variables which are doing some of this work. However the high level of statistical significance even when controlling for the other independent variables does suggest that all things being equal a case has a greater chance of success at the Supreme Court if it arises from one of these three regions no matter how it gets to the Supreme Court (federal appellate courts, state court, or other federal courts, etc.).

Cases decided after 1987 and the legalization of large scale Indian gaming (P-Value of .021). This is the least surprising of the statistically significant variables. Since the Cabazon case affirmed the inherent sovereignty of tribes to regulate gaming as long as not prohibited by their surrounding state, Indian interests have prevailed in just 22 of 69 cases (31.9%), but before the Cabazon decision Indian tribes had prevailed in 51 of 89 cases (57.3%). The descriptive data alone clearly indicates that Indian tribes have been less successful recently at the Supreme Court.

What the model has added is a confirmation that this trend against tribal success holds true even after accounting for the other independent variables in the model like the participation of the Solicitor General, whether the appeal came from state or federal court, what region of the country the appeal originated in, whether jurisdictional issues were present, etc. The fact that the relationship remains true indicates that there is some element of bias in the Court since 1987 that did not exist in the same way before 1987. The rise of Indian gaming is commonly used as that inflection point.¹⁷²

These results trigger closer scrutiny of the data surrounding 1987. While it is possible that Indian gaming is the reason (or part of the reason) for the lack of success of tribal interests at the Court in the years following 1987, the model merely instructs that the null hypothesis be rejected. It does not necessarily imply that Indian gaming or the Cabazon decision was exactly the reason for the change in the Court’s behavior. Astute observers might note that a number of other things happened around 1987. Justice Rehnquist was elevated to Chief Justice in 1986, Justice Kennedy succeeded Justice Powell in 1988, and Justice Thomas succeeded Thurgood Marshall in 1991. The Court therefore got more conservative at the same time that it decided the Cabazon decision.

To place this variable in context it might be helpful to look at the element of time and the changing nature of the Court. The following table breaks down the descriptive statistics of the ultimate success of Indian tribes by decade and by Chief Justice.

# Pro- Tribe Total Indian %

Opinions Cases Win

Tribal Success by Chief Justice

Warren Court (1953-1969) 10 16 62.5% Burger Court (1969-1986) 39 71 54.9% Rehnquist Court (1986-2005) 15 51 29.4% Roberts Court (2005-Present) 9 20 45.0%

Tribal Success by Decade

1953-1959 3 5 60.0% 1960-1969 7 11 63.6% 1970-1979 23 38 60.5% 1980-1989 20 44 45.5% 1990-1999 6 25 24.0% 2000-2009 5 20 25.0% 2010-2020 9 15 60.0%

The model’s observation that 1987 was a relevant turning point in tribal success is accurate given the current data, but there is reason to believe that this may be changing. The appointment of Justice Gorsuch to replace Justice Scalia has altered the balance in Indian law cases, with the tribal interest prevailing in the five most recent Indian law cases¹⁷³ (although Justice Gorsuch voted against the tribal interest in one of them).¹⁷⁴ It is worth continuing to watch the Court’s behavior in Indian law cases to see if the 1987 inflection point identified in the data loses some of its relevance with the composition of the current Court.

Conclusion

This paper marks the first attempt in the literature to explain the success (and failure) of tribal interests at the Supreme Court over the span of four Chief Justices. The logistic regression analysis has drawn upon a new dataset created specifically to examine the Court’s behavior in cases presenting Indian law questions. The results are suggestive of patterns that are unique to Indian law and provide new insights for both practitioners and scholars.

The use of empirical data is designed to usher in a new and important era in the advocacy of Indian interests. Because Indian appellants are more likely to prevail, more effort and resources should be focused on getting the Supreme Court to accept Indian law cases when the Indian interest is the appellant party. Other scholarship suggests that additional amicus briefs¹⁷⁵ may help cases gain attention at the certiorari stage of litigation. Advocates for the advancement of Indian law should build on these strategies to help move more cases from filing to cert worthy when the tribal interest is the appellant party.

The data also provides other important insights for Indian law advocates, some of which may appear counterintuitive. Appellate advocates have the ability to frame legal issues in a variety of ways in an attempt to gain the attention and sympathy of the judiciary.¹⁷⁶ The model presented here suggests that the Court is particularly sympathetic to tribal interests when they are framed as struggles between tribes and states, especially when the fault lines coalesce around issues of jurisdiction. Lawyers should use this information to frame their arguments to the court in terms of jurisdictional disputes or as cases which pit tribes against states. Doing so suggests marginally improving the chances of securing a pro-Indian outcome.

Finally, when possible, advocates for reform in Indian country should try to cultivate cases from the Northern Plains, the Southwest, or Oklahoma. Cases originating in states with larger reservations and more rural populations appear to attract the Court’s attention and sympathy differently than cases from the West coast—even after accounting for factors like the Solicitor General’s participation, whether the tribal interest is the appellant, whether the case came from a state or federal court, etc. The model does not provide a concrete explanation for why this is true. It may be that the treaty relationship or Congressional treatment of these areas is somehow different, or that the Southwest, Oklahoma, and the Northern Plains play a particularly important role in the imagination of what Indian country is in the minds of the Court. Regardless of the reason, groups like the Native American Rights Fund who focus on promoting tribal interests should try to cultivate litigation from some places more than others to advance Indian interests.

In addition to explaining the Supreme Court’s behavior in Indian law cases, this paper also provides an empirical evaluation of individual Justice’s voting behavior in Indian law cases. By ranking the Justices based purely on the proportion of votes cast in support of Indian interests, the Indian law community can better evaluate the Court and its members during the era of tribal self-determination. The data provides general support for the proposition that more liberal Justices are more likely to support the outcome preferred by Indian tribes, but also highlights some noted outliers. For example Justice Stevens, long revered as a champion of civil rights and liberal values, has a decidedly mixed record on Indian law issues siding with the tribal interest in only 39 of 107 recorded votes. In contrast Justice Gorsuch, a recent Republican appointee with a generally conservative voting record, has sided with the tribal interest in 4 of his first 5 Indian law cases.

The rankings show that the current Court contains three of the most supportive Justices for outcomes which favor Indian country, President Obama appointees Justices Sotomayor and Kagan and President Trump’s appointee Justice Gorsuch. Since Justice Sotomayor was named to the bench the tribal interest has prevailed in 9 of 15 cases (60%) reversing a trend of declining success in Indian law cases. However, the paper also shows that four of the five most hostile Justices to Indian law issues currently sit on Court. This bifurcation of judicial attitudes in Indian law cases will likely present few unanimous or 9-0 opinions in the coming terms. With such a currently hostile bench, this paper and its conclusions become even more important—attempting to identify the issues, parties, and conditions that are most likely to maximize success at the Court for Indian tribes.

Let me end with a call for more scholarship. The canon of empirical work on judicial behavior is deep, but almost none of it singles out Indian law for consideration. This paper provides the first robust dataset in twenty years that looks at Indian law cases as a distinct subset of the Court’s jurisprudence, and has begun to explain only part of the variance of the Court’s behavior in Indian law cases. Indian law needs to catch up.

Future scholars could do more to predict/explain individual Justice’s votes in Indian law cases, explore the role of amicus briefs in Indian law cases at the certiorari stage and on the merits, evaluate the role of brief writers or oral advocates, mine the language of the Court’s opinions for nuances about attitudes toward Indian law, further explore the connection between political ideology and Indian law, examine dissenting or concurring opinions, etc. There is so much more to learn about the Supreme Court’s behavior in Indian law cases that has the potential to transform how Indian law advocates go about presenting their claims. Further empirical analysis of these questions would be warmly welcomed by both scholars and practitioners.