I owe deep thanks for valuable insights and conversations to Aaron-Andrew Bruhl, William N. Eskridge, Abbe R. Gluck, Rebecca M. Kysar, Margaret H. Lemos, Katherine Shaw, Lawrence Solan, Nelson Tebbe, and Adam Zimmerman. I am especially indebted to my husband, Ron Tucker, for his patience with this project. Special thanks to Dean Michael A. Simons and St. John’s University School of Law for generous research assistance and to participants at workshops and colloquia at Yale Law School and the Benjamin N. Cardozo School of Law for their comments on earlier drafts of this Article. Christina Corcoran, Sade Forte, Ilya Mordukhaev, Jennifer Roseman, Samuel Sroka, Rita Wang, Kim Friedman, Lissa Yang, Peter Ryan, Vince Nibali, Christine Sammarco, and Thomas Combs provided terrific research assistance. Thanks also to the editors at The University of Chicago Law Review for their excellent work. All errors are my own.
This Article provides the first empirical study of the Roberts Court’s use of substantive canons in statutory interpretation cases. Based on data from 296 cases, the Article argues that much of the conventional wisdom about substantive canons of statutory construction is wrong, or at least overstated with respect to the modern Supreme Court. Substantive canons—for example, the rule of lenity, the avoidance canon, and the presumption against extraterritorial application of domestic laws—have long been criticized as undemocratic judge-made rules that defeat congressional intent, enable interpreters to massage different meanings out of the same text, and make statutory interpretation unpredictable. Scholars have bemoaned the amount of work that substantive canons perform in statutory interpretation cases, and several have charged that textualist judges in particular overuse such canons. But virtually all of these critiques have occurred in the absence of empirical evidence about how judges invoke substantive canons in practice.
This Article reconsiders the substantive canons in light of new data collected from the Roberts Court. The data show that, contrary to the conventional wisdom, substantive canons are infrequently invoked on the modern Court—and even when invoked, they rarely play an outcome-determinative role in the Court’s statutory constructions. Perhaps most surprisingly, textualist justices—including Justice Antonin Scalia—rarely invoke substantive canons in the opinions they author, and do so less often than most of their purposivist counterparts. Moreover, contrary to the conventional view that substantive canons empower judges to read their personal policy preferences into statutes, the Court’s conservative justices have employed substantive canons to support liberal case outcomes as often, or nearly as often, as they have employed such canons to support conservative outcomes. Further, doctrinal analysis shows that the Roberts Court repeatedly has used substantive canons to honor, rather than frustrate, congressional intent.
The Article also challenges scholars’ gloomy warnings that justices in the modern, textualism-influenced era have replaced legislative history with substantive canons as the go-to resource for deciphering ambiguous statutory text. Rather, the data from the Roberts Court show that most of the justices referenced legislative history at higher rates than they referenced substantive canons. Moreover, the Court’s own precedents—rather than substantive canons or legislative history—seem to be the unsung gap-filling mechanism that the justices turn to when confronted with unclear statutory text. After reporting the data, the Article discusses the implications of its findings for current debates in statutory interpretation, arguing that statutory interpretation theory needs to pay less attention to substantive canons and more attention to how the Court employs precedents when construing statutes.
TABLE OF CONTENTS TABLE OF CONTENTSScholars on both sides of the textualist-purposivist divide have criticized substantive canons, although textualists also have defended some substantive canons as entrenched background conventions that Congress is aware of when it legislates.4 Justice Antonin Scalia, notably, decried substantive canons as “dice-loading” devices;5 but he, in turn, has been accused of employing such canons generously when it suited his ends.6 Scholars have pointed out that substantive canons are countermajoritarian, subject to judicial invention and reinvention, and difficult for Congress to overcome.7 Recent work has characterized substantive canons as the equivalent of federal common law, explored the theoretical tension between textualism and the substantive canons, and concluded that the constitutionally based canons, at least, are defensible on a “faithful agent” theory of statutory interpretation.8
This Article provides the first empirical study of the Roberts Court’s use of substantive canons in its statutory interpretation cases. Based on data from 296 statutory interpretation cases decided by the Roberts Court during its first six and a half terms, the Article reports several surprising findings that call into doubt the conventional account of substantive canons and, particularly, their relationship to textualism. Five points stand out: (1) contrary to popular claims that textualist judges rely on substantive canons frequently, the Court’s textualist justices rarely invoked substantive canons in the opinions they authored (11.0 percent and 11.7 percent for Scalia and Justice Clarence Thomas, respectively), and did so no more often than their nontextualist counterparts;12 (2) despite the ubiquity of substantive canons and charges that judges regularly invent new ones, only a handful of substantive canons appear to be doing meaningful work on the modern Court;13 (3) a majority of the Roberts Court’s references to substantive canons have been in passing, as makeweight or secondary arguments, with only a small number of cases relying significantly on such canons; (4) when the Court did rely significantly on substantive canons, it often exhibited more attentiveness to Congress’s intent than the conventional view of substantive canons accounts for; and (5) contrary to the popular belief that substantive canons empower judges to decide cases based on their personal policy preferences, the Court’s conservative justices invoked substantive canons to support liberal outcomes nearly as often, and in some cases more often, than they invoked such canons to support conservative outcomes.
These findings have important theoretical implications. For example, they raise the question: If textualist judges are not relying on substantive canons to mitigate the rigors of textualism, then what interpretive tools are they using to serve that function? Data from textualist-authored opinions in which substantive canons were not used suggest that the answer to this question is Supreme Court precedent and, to some extent, practical-consequences-based reasoning.14 The findings also suggest that some of the criticisms leveled against substantive canons may be overstated—for example, substantive canons may be less prone to invention and reinvention than previously thought. Indeed, in six and a half terms, I counted only four instances in which the members of the Roberts Court invoked an arguably “new” substantive canon—and over half of the Court’s substantive canon references involved one of just six well-established canons.15 This means that despite the wide array of substantive canons created over the past two centuries, the Roberts Court was remarkably constrained, and somewhat predictable, in the canons it tended to invoke. The data also suggest that substantive canons may be applied in a manner that is more supportive of congressional intent than the conventional account recognizes.16 Doctrinal analysis of the handful of cases in the data set in which the Court did rely significantly on substantive canons reveals that, in several instances, the Court used the avoidance canon to read a statute in a manner that honored a recent congressional override, or to preserve a long-standing statute against constitutional challenge.17 In other cases, the Court employed substantive canons in tandem with, or as an approximation of, congressional intent.18 These cases stand in marked contrast to the conventional wisdom—including recent commentary about the Roberts Court19 —arguing that the Court uses substantive canons to displace legislative preferences with judicial ones.
Further, the data from this study suggest that, in practice, there may not be the stark textualist-purposivist divide that scholars have described regarding whether to consult substantive canons or legislative history to provide contextual clues about statutory meaning.20 As Part II describes in detail, eight of the eleven justices who have served on the Roberts Court for a significant amount of time have referenced legislative history at higher rates—sometimes much higher rates—than they have referenced substantive canons. Even archtextualist Scalia invoked substantive canons at almost the same rate as he invoked legislative history—an interpretive resource he considered illegitimate—and Thomas referenced both of these tools at quite low rates, suggesting a miserly view of both.21 In addition, the data on levels of reliance provide some support for pragmatists’ contentions that the canons of construction do not play much of a role in the judicial interpretation of statutes, and that judges use them as mere “window-dressing” to shore up interpretations reached primarily through other tools.22
This Article proceeds in three parts. Part I describes the conventional understanding of substantive canons. Part II reports the findings from my study of the Roberts Court’s substantive canon use from January 2006, when Justice Samuel Alito joined the Court, to the end of the Court’s 2011 term. It also provides doctrinal analysis of several cases in which the Court relied significantly on substantive canons, and examines the tools the Court invoked when it did not reference substantive canons. Part III explores the theoretical implications of the data and doctrinal observations.
Before exploring the conventional wisdom surrounding substantive canons in detail, it is worth pausing for a moment to define some terms. First, what are “substantive canons”? “The phrase ‘canons of construction’ is understood to encompass a set of background norms and conventions that are used by courts when interpreting statutes.”26 Many scholars further divide the “canons of construction” into two categories: language canons and substantive canons.27 Language canons, as their name suggests, focus on the text of the statute and encompass rules of syntax and grammar, “whole act” rules about how different provisions of the same statute should be read in connection with each other (for example, to minimize internal inconsistency or to avoid superfluity), and Latin maxims such as expressio unius est exclusio alterius and noscitur a sociis.28 Substantive canons, by contrast, are principles and presumptions that judges have created to protect important background norms derived from the Constitution, common-law practices, or policies related to particular subject areas.29 Substantive canons sometimes operate as tiebreakers, or thumbs on the scale, but in recent years they have been accused of playing a much more determinative role in the Court’s statutory cases.30
Perhaps the most famous substantive canon is the rule of lenity, which holds that ambiguities in criminal statutes must be resolved in favor of the defendant.31 A close second is the canon of constitutional avoidance, which holds that if there are two or more plausible readings of a statute, and one of these raises serious constitutional concerns, the Court should adopt the reading that avoids the constitutional problem.32 Other well-known substantive canons include: the rule that waivers of sovereign immunity should be narrowly construed,33 a rule requiring a “clear statement” before a federal statute may be read to intrude in areas traditionally regulated by states,34 a rule that ambiguities in statutes dealing with Indian tribes are to be resolved in favor of the tribes,35 a rule that statutes in derogation of the common law are to be narrowly construed,36 a rule that remedial statutes are to be construed liberally,37 and a rule that interpretive doubts should be resolved in favor of veterans.38
As the above list illustrates, substantive canons come in many varieties. Most take one of three forms: (1) “presumptions,” or “rules of thumb that cut across different types of statutes” and “that the Court will ‘presume’ Congress intends to incorporate into statutes”;39 (2) “liberal” or “strict” construction canons, which direct courts to read statutes dealing with certain subject areas either expansively or narrowly;40 and (3) “clear statement rules,” which require a clear statement on the face of the statute in order to rebut a policy presumption the Court has created.41 The next Section discusses prevailing scholarly assumptions about how these policy-based canons operate.
Statutory interpretation scholars of widely divergent philosophies tend to converge in several respects in their views about substantive canons. They agree, for example, that textualist judges employ substantive canons regularly42 and, at the same time, that there is significant theoretical tension between substantive canons and textualism.43 Implicit in these two points of convergence are two additional assumptions—that substantive canons play a significant role in the interpretation of statutes and that their effect is to displace legislative policy preferences with judicial ones.44
On the opposite side of the interpretive divide, textualist scholar Professor John Manning has made similar observations. Manning has noted, for example, that devices like the avoidance canon and clear statement rules “mitigate the textualists’ strict focus on the conventional meaning of the enacted text.”48 Further, he has acknowledged that these canons are in tension with “the most basic textualist assumptions” and that some textualists or textualist-sympathetic judges “have recognized the necessity for restrained application of such tools of construction.”49 In addition, although Manning has defended certain substantive canons as reflecting “established background conventions,” he has conceded that when textualists create new substantive canons, they should be forced to justify their reliance on such interpretive tools.50
Other scholars with less clearly staked-out jurisprudential philosophies have echoed these assumptions as well. Professor Amanda Frost has noted that “textualist judges are particularly fond of clear statement rules, which aid them in interpreting ambiguous statutes without the need to resort to legislative history.”51 And Professor Amy Coney Barrett has argued that textualists “embrace” many substantive canons, including the canon of constitutional avoidance, the rule of lenity, and several clear statement rules.52
In addition, numerous commentators writing over a wide time span have maintained that judges use substantive canons strategically, to effect judicial policy preferences at the expense of legislative intent.53 Speaking broadly about both substantive and language canons, British jurisprude Sir Frederick Pollock complained more than a century ago that canons “cannot well be accounted for except on the theory that Parliament generally changes the law for the worse, and that the business of the judges is to keep the mischief of its interference within the narrowest possible bounds.”54 More recently, Eskridge has argued that Scalia used substantive canons selectively and arbitrarily, and in a manner that was less constraining than interpretive approaches that consider legislative history and intent.55
Scalia himself characterized at least some substantive canons as “a sheer judicial power-grab”56 and called all substantive canons “a lot of trouble” to “the honest textualist.”57 He described substantive canons as “indeterminate,” leading to the “unpredictability, if not the arbitrariness, of judicial decisions.”58 He also questioned “where the courts get the authority to impose them,” doubting whether courts can “really just decree that [they] will interpret the laws that Congress passes to mean less or more than what they fairly say.”59
Commentators also seem to agree that there is tension between substantive canons and legislative history, and that as conservative justices have come to dominate the Supreme Court, there has been a trend toward increasing reliance on substantive canons and diminishing reliance on legislative history. Judge Patricia Wald, for example, has observed that “legislative history is often rejected in favor of, or at least filtered through, canons, presumptions, or principles considered overriding by a majority of the Court.”60 Professor Charles Tiefer similarly has argued that textualists have “strengthen[ed] [ ] substantive canons along grounds of dispensing with legislative history.”61 More recently, Professors James Brudney and Corey Ditslear’s study of Supreme Court canon use in employment law cases found that, in closely decided cases, there was a “distinctly conservative influence associated with substantive canon reliance.”62 Brudney and Ditslear identified a subset of cases in which the majority opinion relied on canons (language or substantive), while the dissenting opinion relied on legislative history.63 In short, the picture painted by statutory interpretation scholars has been of a Court dominated by conservative justices who regularly employ substantive canons and rarely consult legislative history, and of a liberal minority who infrequently invoke substantive canons but continue to consult legislative history regularly.
At first blush, the prevailing scholarly wisdom seems unimpeachable. Some prominent substantive canons blatantly are designed to contravene congressional intent, or at least to impose default rules that penalize Congress for imprecise drafting. The rule of lenity, for example, dictates that doubts about whether particular conduct is covered by a criminal statute should be resolved in favor of criminal defendants, even though members of Congress tend to take a “tough on crime” stance and likely would prefer to resolve such close calls in favor of the government.64 Similarly, federalism clear statement rules impose a default rule that preserves state power at the expense of federal power, deliberately setting a high bar for congressional efforts to intrude on state rights—even though members of Congress would likely prefer that such doubts be resolved in favor of the federal rule.65 But despite the intent-defeating design of some substantive canons, doctrinal analysis of the cases studied in this Article suggests that substantive canons often are used in tandem with, or to further, legislative intent, rather than to frustrate it.66
There is only one study to date that has examined the substantive canons as a set, rather than focusing on one or a handful of specific such canons. That study, published in 1992 by Eskridge and Professor Philip Frickey, reviewed the Supreme Court’s substantive canon use during the last eleven years of the Burger Court and the first five years of the Rehnquist Court.67 Eskridge and Frickey posited that, rather than deciding cases based on what a substantive canon dictates, the Court shapes and even invents canons to fit the results it reaches in individual cases.68 That is, substantive canons play an important role in the way the Court expresses or justifies the “value choices” it is employing when it construes a statute, but they do not dictate or produce outcomes; rather, outcomes produce canons.69 As evidence in support of their view, Eskridge and Frickey reported that a comparison of the Court’s behavior during the Burger and Rehnquist Courts, and even during different periods within the Burger Court, revealed “striking differences both in the canons the Court invokes and in the way in which the canons are invoked.”70 They argued that this makes sense because, if outcomes produce canons, we can expect that “different Courts will express their different ideologies by emphasizing, or even creating, different substantive canons in the complex work of statutory interpretation.”71 Indeed, Eskridge and Frickey’s article focused on the Court’s creation of several new federalism-protecting clear statement rules during the 1980s.72 Specifically, they observed that “[t]he Court not only created new canons reflecting federalism-based values, but also transformed some of the existing clear statement rules into super-strong clear statement rules.”73 And they took this invention and reinvention of substantive canons to be par for the course, rather than a development unique to the federalism clear statement rules or to the particular period they studied.74
In a later article, Eskridge further observed that during the period between 1987 and 1994, the Rehnquist Court employed “no fewer than seventy-nine different” substantive canons in statutory cases.75 He also pointed out that “Scalia joined or wrote the Court’s opinion in almost all of the cases where the Court invoked or revised these substantive canons.”76 As Part II elaborates, I found considerably less variety in the number of substantive canons employed by the Roberts Court between 2006 and 2012 and low rates of substantive canon reliance in the opinions authored by Scalia.77
Although not focused on substantive canons, Brudney and Ditslear’s study of the Supreme Court’s workplace law cases also provides some important background for this Article. In particular, the Brudney-Ditslear study reinforces several prevailing assumptions about the relationship among substantive canons, ideology, and congressional intent. As noted above, Brudney and Ditslear identified a subset of 10 closely divided cases (out of 148 total closely divided cases) in which the majority opinion relied on substantive canons but not legislative history, while the dissent invoked legislative history to support its position.78 In nine of these cases, the majority opinion applied a substantive canon to reach a conservative result, while the dissenting opinion relied on legislative history to reach a liberal result.79 Based on these and related findings, Brudney and Ditslear concluded that “this polarized pattern suggests that for an identifiable subset of divisive cases, the canons are being used by the Rehnquist Court to help produce a judicially desired set of policies, ignoring or sacrificing legislatively expressed preferences in the process.”80 Although Brudney and Ditslear’s study involved a relatively small number of cases, the conclusions Brudney and Ditslear drew are consistent with prevailing views that substantive canons operate to undermine congressional intent. As discussed in Part II, my study of the Roberts Court suggests that these conclusions may be overstated and that, at least in some cases, the Court seeks to use substantive canons to honor congressional intent.
Both the Eskridge-Frickey and the Brudney-Ditslear studies appear to have assumed that when substantive canons were invoked, they played a meaningful role in the Court’s statutory constructions. The evidence reported in the next Part provides some reason to question that assumption and affords a more complete picture of how individual justices compare in their use of substantive canons.
For purposes of this study, “substantive canons” were defined as background constitutional or policy norms, rules, or presumptions about how statutes should be interpreted, including rules about how statutes in a particular subject area should be construed. Coders were instructed to identify the specific substantive canon being invoked.88
The interpretive resources coded for in this study are consistent with those that have been examined in other empirical studies of the Court’s statutory interpretation practices.89 A few differences in definitions used for the different sources were inevitable and are pointed out when notable. For example, some early empirical studies of the Supreme Court’s statutory cases lumped language and substantive canons together under the heading “canons of construction,” rather than measuring references to these different forms of canons separately, as I did.90
Secondary or corroborative references to an interpretive tool, on the other hand, were counted; thus, when the Court reached an interpretation based primarily on one interpretive source but went on to note that x, y, and z interpretive tools further supported that interpretation, the references to x, y, and z were coded along with the primarily-relied-upon source.93
In addition, the vote margin in each case was recorded, and each case and opinion was recorded as unanimous, close margin, or wide margin (cases with six or more justices in the majority). Each justice’s vote in each case also was recorded, as was the author of each opinion. This methodology was the same as that followed in my previous empirical studies.94
Finally, every opinion in which a substantive canon was invoked was coded as containing “passing reliance,” “some reliance,” or “primary reliance” on the substantive canon. While this coding necessarily involved some judgment calls, I believe that it adds valuable texture to our understanding of how the Court uses substantive canons when it chooses to invoke them. In any event, my data are available for others to review and agree or disagree with.96 The coding parameters for reliance were as follows: An opinion was coded as containing “passing reliance” on a substantive canon if it made minimal reference to the canon, or mentioned it as a fallback or add-on argument supporting a reading already arrived at through other interpretive tools. An opinion was coded as involving “some reliance” on a substantive canon if it made more than minimal reference to the canon, but did not rely on the substantive canon as the main justification for the construction it adopted. Finally, an opinion was coded as containing “primary reliance” if it relied primarily on a substantive canon to justify the result reached.
A few examples should help illustrate how these judgment calls were made. In Exxon Shipping Co v Baker,97 for instance, the Court considered whether the Clean Water Act’s98 water pollution penalties preempt punitive damages awards in maritime spill cases.99 The Court concluded that the Act does not preempt punitive damages awards, based primarily on the statute’s text.100 The Court also commented that it saw no clear indication of congressional intent to preempt such awards and made a passing reference—in a parenthetical—to the rule that a statute must clearly state its intent to abrogate a common-law principle.101 This case was coded as containing “passing” reliance on a substantive canon. By contrast, in Northwest Austin Municipal Utility District Number One v Holder102 (“NAMUDNO”), the Supreme Court employed the avoidance canon to adopt a strained reading of the Voting Rights Act of 1965103 (VRA)—concluding that a Texas utility district counted as a “political subdivision” eligible for bailout from the preclearance requirement of VRA § 5—in order to avoid thorny questions about the constitutionality of § 5’s preclearance formula.104 NAMUDNO was coded as involving “primary reliance” on a substantive canon.105 Finally, in Begay v United States,106 the Court held that the offense of driving under the influence (DUI) does not count as a “violent felony” under the Armed Career Criminal Act of 1984.107 Justice Scalia agreed and authored a concurring opinion that relied on the statutory text, the whole-act rule, and a dictionary definition to conclude that the proper test for determining whether an offense is a “violent felony” is whether the offense “pose[s] at least as serious a risk of physical injury to another as burglary.”108 His concurring opinion then stated that drunk driving could not clearly be said to pose “at least as serious a risk of physical injury to another as burglary” and that the rule of lenity therefore required him to find in favor of the defendant.109 Scalia’s concurrence in Begay was coded as containing “some reliance”—that is, more than “passing reliance” but less than “primary reliance”—on a substantive canon.
Before reporting the data, it is important to note some limitations of this study. First, the study covers only six and a half Supreme Court terms and only 296 statutory interpretation cases, decided by some combination of the same eleven justices. While this data set is large enough to reveal some things about the Supreme Court’s use of substantive canons, the data reported may reflect trends specific to the Roberts Court. Second, great significance should not be placed on the precise percentages reported; the number of cases reviewed is large enough to provide some valuable insights, but the focus should be on the patterns that emerge rather than on specific percentages. Third, in noting the weight, or intensity, of an opinion’s reliance on a substantive canon, I make no claims to have discovered the justices’ underlying, or “true,” motivations for deciding a statutory case; the data do not reveal whether a particular opinion relied heavily on a substantive canon because the opinion’s author was persuaded by the canon, or merely because the author thought the canon was a convincing interpretive tool. The study’s empirical and doctrinal claims are confined to describing how the justices publicly engage the substantive canons as justifications for their statutory constructions and to theorizing about discernable patterns in their public engagement of such canons.
Table 1 reports the frequency with which the members of the Roberts Court referenced substantive canons—and other interpretive canons or tools of construction—in the opinions they authored during the Court’s 2005 through 2011 terms.110 For seven of the eleven justices, the rates of reference to substantive canons are low, roughly at or below 15.0 percent.111 As Table 1 shows, this compares to much higher rates of reference, across justices, for almost every other interpretive tool.112 Even the three most frequent users of substantive canons, Justices John Paul Stevens, John Roberts, and Sonia Sotomayor, employed such canons in no more than 25.0 percent of the opinions they authored.113 Indeed, out of the eleven interpretive tools listed in Table 1, substantive canons were among the least frequently invoked. Only common law exhibited similarly low rates of reference across the board.114 While some of the justices referenced legislative intent at low rates as well, several others invoked intent at much higher rates (above 20.0 percent).115
Canons / Interpretive Tools
Supreme Court Precedent
Text / Plain Meaning*
Combined Language Canons /
Substantive Canons
11.7%
11.0%
13.8%
22.0%
15.2%
11.3%
14.3%
14.3%
15.8%
17.6%
25.0%
Table 2 provides a slightly different comparison of the rates at which the members of the Roberts Court referenced substantive canons, relative to other interpretive tools, as a percentage of all opinions issued during the period studied.
Canons /
Interpretive Tools
Supreme Court Precedent
Text / Plain Meaning
Combined Language Canons /
Substantive Canons
14.4%
15.5%
18.0%
5.2%
5.0%
Like the data in Table 1, the data in Table 2 reveal that substantive canons were infrequently invoked in the Court’s statutory cases (14.4 percent of all opinions). Also as in Table 1, substantive canons were the Court’s second-least-frequently-used interpretive resource overall, behind only the common law.116 This raw rate is not shocking in itself; Professors Brudney and Ditslear’s study of workplace law cases found similarly low overall rates of reference to substantive canons (15.6 percent of Rehnquist Court cases).117 What is new is the finding that such low rates of reference hold true across all statutory subject areas—not just in cases involving employment law statutes—and the comparative data indicating that the Court invoked substantive canons far less often than it invoked almost all other interpretive canons and tools.
More striking, however, is what the data show about how infrequently textualist justices employed substantive canons in the statutory opinions they authored. Contrary to popular belief, the Court’s most prominent textualist, Scalia, referenced substantive canons in only 11.0 percent (nine of eighty-two) of the opinions he authored, and fellow textualist Justice Thomas did so in just 11.7 percent (nine of seventy-seven) of the opinions that he authored.118 Textualist-leaning Justice Alito119 exhibited similarly low rates of reference.120 Equally surprising is that the three most frequent users of substantive canons turned out to be Stevens, Roberts, and Sotomayor.121 This is unexpected, and turns the conventional wisdom about substantive canons on its head, because Stevens and Sotomayor are intentionalists, who regularly use legislative history to inform their statutory constructions—not textualist judges forced to turn to substantive canons to provide the context they lack because they refuse to consult legislative history.122 Part III discusses the implications of these findings in detail.
The data also reveal some surprising information about the weight that the justices placed on substantive canons when they did invoke them. Table 3 reports how often the members of the Roberts Court placed “passing,” “some,” or “primary” reliance on substantive canons in the opinions in which they employed such canons.
Substantive Canons (n=87)
Notably, the data show that the justices placed only “passing” or “some” reliance on substantive canons in 71.3 percent of the opinions that invoked such canons.123 Many of these opinions mentioned substantive canons only at the tail end, in a fleeting, throwaway fashion or as an alternative, “even if,” argument.124 Others employed substantive canons as a secondary resource or interpretive aid.125 Ultimately, less than 30.0 percent of the substantive canon–invoking opinions in the data set relied “primarily,” or in a dispositive manner, on such canons.126 Thus, it seems that not only are substantive canons infrequently invoked, but even when invoked, they are used only to reinforce other considerations in the vast majority of cases.127
Finally, the data show that the modern Court is not invoking substantive canons at dramatically higher rates than legislative history, as some scholars have speculated. On the contrary, eight of the eleven justices who served on the Roberts Court during the period studied—including conservative, textualist-leaning Alito and Justice Anthony Kennedy—invoked legislative history more frequently than they invoked substantive canons, and Scalia invoked these two interpretive resources at almost equal rates.128
Overall, these data call into question scholarly assumptions that, as the Court has grown more conservative in recent years, it has moved toward frequent reliance on substantive canons at the expense of legislative history.132 Indeed, I found only 4 conservative-outcome divided-vote cases (out of 158 divided-vote cases and 60 closely divided cases in the data set) in which the majority opinion invoked a substantive canon but not legislative history, while the dissenting opinion referenced legislative history.133 Recall that Brudney and Ditslear found 9 cases fitting this pattern, out of 148 closely divided cases total.134 Brudney and Ditslear interpreted this subset of cases to represent a disturbing snapshot of how the Court uses substantive canons to effect judicially desired policies at the expense of clearly expressed legislative preferences.135 I view the (smaller) subset of similar cases in my data set differently—that is, as evidence that substantive canons rarely are employed instead of, or to contradict, legislative history. In other words, while it may be true that the Court’s use of legislative history has declined since the Burger Court era,136 its reliance on legislative history as an interpretive resource has not been replaced, in most cases, with reliance on substantive canons.
As discussed earlier, Professors Eskridge and Frickey’s prior study of substantive canon use by the Burger and early Rehnquist Courts posited that we should expect to see wide differences in the specific substantive canons employed by different Courts, or even different iterations of the same Court—and that the cases they studied revealed such differences.137 Eskridge also suggested that the universe of substantive canons is vast and evolving, noting that, in the period from 1987 to 1994, the Court invoked “no fewer than seventy-nine” different substantive canons, many of them new or substantially modified canons.138 Other scholars have argued that there are “too many” of certain types of substantive canons, and that the variety of such canons makes it “difficult . . . to predict the courts’ interpretive path.”139
But the data from the Roberts Court’s first six and a half terms call into doubt this conventional account. Out of 296 statutory cases, and 87 opinions that invoked a substantive canon, I found only 4 instances in which the justices arguably could be said to have created a “new” substantive canon or to have modified an existing one.140 Nor was there significant variety in the canons the justices invoked; most opinions referenced one of just six well-established canons—the avoidance canon (sixteen), the rule of lenity (eleven), the presumption against preemption (eight), a federalism clear statement rule (six), the presumption against retroactivity (five), or the narrow construction of waivers of sovereign immunity (four).141 Overall, forty-eight of the eighty-seven (55.2 percent) Roberts Court opinions that invoked a substantive canon employed one of just a handful of canons.142 The remaining opinions referenced thirty-three other canons—meaning that in the period between 2006 and 2012, the Court employed only thirty-nine different substantive canons, total.143 Even if we disaggregate the federalism clear statement rules and count each one separately, the total number would come to only forty-four different canons.144 This is a far cry from the seventy-nine substantive canons that Eskridge and Frickey recorded during the Court’s 1987 through 1994 terms.145
There could be many reasons for these stark differences in the variety of canons invoked over different time periods. Notably, the Court’s docket has shrunk over the years, so the number of statutory cases decided likely was larger during the period of Eskridge and Frickey’s study.146 In addition, the Court’s composition stayed conservative over the entire period from 1994 to 2012, so the Roberts Court may have been closer, ideologically, to the Rehnquist Court than the Rehnquist Court was to the Burger Court. As a result, the Roberts Court may have found less need to create new canons or to reinvent canons created by the Rehnquist Court in order to reach its preferred results. Alternately, perhaps the Roberts Court is inherently more cautious than was the Rehnquist Court.147
Another possibility is that Eskridge and Frickey may have defined what constitutes a substantive canon more broadly than I did. At a theoretical level, there does not seem to be much difference between Eskridge and Frickey’s conception of what constitutes a substantive canon and mine—we agree that such canons establish rules or presumptions of statutory construction that are based on substantive policy norms derived from the Constitution, common law, or general Anglo-American legal traditions.148 In practice, however, they may have been more capacious in what they counted as a substantive canon. One crude way to test this possibility is to compare the list of substantive canons provided in the appendix to Eskridge and Frickey’s statutory interpretation casebook with the list of substantive canons identified in this study.149 Such a comparison reveals that Eskridge and Frickey identified eight substantive canons from the 2005–2011 terms that my study did not count as substantive canons.150 This crude comparison suggests that Eskridge and Frickey were, indeed, somewhat more generous in their determinations of what counts as a substantive canon. But it is uncertain whether such definitional or coding differences account for the entire gap between Eskridge and Frickey’s findings of extensive substantive canon use and my more modest findings. Indeed, even if we add these eight canons to the list compiled in my Appendix, the total number of different canons invoked during the 2005 through 2011 terms would come to fifty-two—still a long way from the seventy-nine different canons identified by Eskridge in his earlier study.151
Ultimately, it is unclear which of the two studies captures the norm regarding the Court’s use of substantive canons across time, or if either one does—it is possible that both studies are anomalies or sui generis. Either way, the differences in the variety of substantive canons invoked by the Roberts versus Rehnquist and Burger Courts—and especially the Roberts Court’s lack of new substantive canon creation—are intriguing. Moreover, whatever the reason for the differences between the two periods studied, it is undeniable that the Roberts Court has been surprisingly predictable, as well as somewhat restrained, with respect to the universe of substantive canons it has invoked: again, just six well-established canons account for more than half of the Court’s total substantive canon references.152 These data contradict the conventional view that there are “too many” substantive canons and that their ubiquity makes it difficult to predict the courts’ interpretive path.153
As noted in Part I, many commentators have argued that judges use substantive canons strategically, to achieve outcomes that accord with their policy preferences. Brudney and Ditslear’s study of workplace law cases reinforced this view, finding that liberal justices tend to use canons (both language and substantive) to reach liberal outcomes, while conservative justices use canons to reach conservative outcomes.154 The data from the Roberts Court’s first six and a half terms are only partly consistent with these findings. Table 4 reports the ideological direction for each justice for every opinion he or she authored or joined that referenced a substantive canon.155 As Table 4 shows, all of the liberal justices employed substantive canons to support liberal outcomes at far higher rates than they employed such canons to support conservative outcomes.156 However, four of the five conservative justices—Scalia, Thomas, Roberts, and Kennedy—employed substantive canons to reach liberal results in almost the same number of cases as they used substantive canons to reach conservative results.157
Table 4. Ideology by Justice in Opinions Authored or Joined Invoking Substantive Canons (n=87)
n = the number of cases in which the justice authored or joined an opinion that referenced a substantive canon. For example, cases in which a majority opinion invoked a substantive canon, but in which the justice authored or joined a dissenting opinion that did not invoke a substantive canon, are not counted in the denominator.
Table 5 similarly reports the ideological direction of the opinions authored by each justice that invoked a substantive canon.
Caution should be used in interpreting the precise percentages reported, because the raw number of cases is quite low for each justice. However, the pattern demonstrated is clear. The Table shows that four of the justices employed substantive canons to reach outcomes consistent with their ideological preferences in the majority of opinions they authored.158 Four justices—Scalia, Thomas, Alito, and Justice David Souter—wrote opinions that used substantive canons to support liberal and conservative outcomes at equal or nearly equal rates.159 Roberts and Sotomayor were the only justices who employed substantive canons inconsistently with their ideological preferences in a substantial majority of the cases in which they authored an opinion that used substantive canons.160 The next Section discusses Roberts’s use of substantive canons in greater detail, arguing that he often invoked such canons because of institutional concerns, including a desire to avoid upsetting congressional intent.
Substantive Canon
Federalism clear statement rule
Burden on the taxpayer seeking a deduction
Non obstante (allowing implied repeal)
APA presumption of judicial review of all final agency actions
Presumption against retroactivity
Narrow construction against the government
Presumption against a major departure from the “long tradition” of equity practice
Clear statement rule for attorney’s fee awards
Ambiguity resolved in favor of veterans
Clear statement rule for disallowing state law claims
Presumption that the Bankruptcy Code does not erode past practice absent clear intent
Presumption favoring the retention of “long-established and familiar principles”
Equal sovereignty principle
Presumption favoring severability
Presumption that waivers of sovereign immunity must be clearly stated
As Table 5a shows, there was no clear pattern in the specific substantive canons that the conservative justices invoked in support of liberal outcomes. While Scalia did invoke the rule of lenity often—in four of nine opinions he authored (representing four of ten substantive canons referenced in those opinions)—both he and the other conservative justices invoked an array of other substantive canons as well when authoring opinions that reached liberal results.
The data regarding the justices’ ideological use of substantive canons—or, perhaps more accurately, demonstrating that substantive canons do not seem to be constraining the justices to vote against their ideological preferences in most cases—are not surprising. Several earlier studies have found that the members of the Court tend to vote consistently with their ideological preferences in statutory cases.161 But the data showing that Roberts and Scalia used substantive canons to support liberal outcomes more often than conservative outcomes in the opinions they authored, and that several of the other conservative justices, including Thomas, authored or joined opinions that invoked substantive canons to reach liberal outcomes in as many (or nearly as many) cases as they did to reach conservative outcomes, are unexpected and noteworthy. This is especially so in light of the widespread scholarly assumption, detailed in Part I, that judges—particularly textualist judges—use substantive canons manipulatively, to subvert legislative intent and to justify outcomes that accord with their policy preferences.
The cases are remarkable not just because they challenge the conventional account that substantive canons tend to be used to trump legislative intent, but also because several invoke substantive canons that are by nature designed to check Congress, or even to defeat congressional intent in the name of some larger—typically constitutional—concern. The avoidance canon, for example, deliberately seeks to curtail the reach of federal statutes in order to cure potential constitutional defects—even when Congress may have intended a much broader reach and wished to push the envelope as far as possible regarding what is constitutionally permissible. Similarly, federalism clear statement rules, such as the presumption against preemption, are designed to cut back the scope of congressional legislation in order to preserve state laws and power. Yet as the cases discussed in this Section demonstrate, the Roberts Court has used both of these canons to fulfill likely congressional intent, despite their intent-negating design. By contrast, other substantive canons seem designed to effectuate Congress’s larger policy goals regarding particular subject areas—for example, the maxim that preferences in the Bankruptcy Code are to be narrowly construed in order to fulfill the Code’s equal distribution goal—and the Roberts Court has invoked such canons in a manner that furthers congressional intent as well.
In providing these case examples, I do not mean to suggest that cases do not exist in which the members of the Roberts Court invoked substantive canons without paying attention to congressional intent, or even used substantive canons to defeat congressional intent. Rather, my aim is to bring to light the fact that substantive canon use is not necessarily inconsistent with attentiveness to congressional intent, and that substantive canons were not used to defeat congressional intent in most cases in which they were invoked. Indeed, my doctrinal analysis of the cases in this study revealed only fifteen opinions (out of eighty-seven total) in which the members of the Roberts Court invoked substantive canons in a manner that arguably served to defeat Congress’s intent.163
The canon of constitutional avoidance has long been criticized as sanctioning the judicial revision of statutes in a manner that undercuts Congress’s design.164 The canon operates on the assumption that Congress would not want to press constitutional limits and would rather have its laws upheld in some form than invalidated as unconstitutional.165 But as several scholars have pointed out, it might make more sense to presume that Congress prefers to legislate to the maximum extent of its power and to have its work product given full effect—and for the Court to openly say so when it believes that work product to be unconstitutional, rather than use the possibility of a constitutional violation as an excuse to rewrite the statute.166
Given these prevailing assumptions, it was surprising to find a number of Roberts Court cases in which the justices seemed to be using the avoidance canon to honor congressional intent—not just paying lip service to what Congress presumably wanted, but engaging in a serious effort to give effect to Congress’s overall aim in amending or enacting a statute. The cases took two forms. First, a number of cases involved opinions that invoked the avoidance canon to support a construction that honored a recent congressional override. Second, a handful of cases involved opinions that used the canon to preserve a long-standing statute from what the opinions explicitly found would be otherwise-certain constitutional invalidity—rather than to rewrite the statute based on speculated or possible invalidity, as the conventional critique predicts.
The first set of cases tended to follow this pattern: an earlier Supreme Court case invalidated a statute or limited its reach on constitutional grounds, Congress sought to override the decision by amending the statute to “fix” the constitutional problem, and then a new lawsuit arose in which the Court was required to interpret the amended statute—and to determine whether the amended statute contained the same deficiencies as the original version. In reviewing the congressional override, the Roberts Court invoked the avoidance canon and argued that the Court should read the amended statute in a way that recognized Congress’s clear intent to remedy the problems that led to invalidation of the original statute, rather than construe the new statute to repeat the deficiencies contained in the original.
A couple of examples should help illustrate. First, in Skilling v United States,167 the Court considered whether a recently enacted criminal statute covered honest-services fraud, including investment fraud.168 Like other cases involving overrides, Skilling had a backstory. For decades, the federal courts of appeals had construed the federal mail and wire fraud statutes to cover acts that deprive the public of the “intangible” right to receive “honest services,” such as bribery and kickback schemes.169 In McNally v United States,170 the Supreme Court put a stop to this construction, holding that the mail fraud statute does not cover honest-services fraud.171 Congress responded quickly, enacting a new statute that was designed “specifically to cover” the “intangible right of honest services” that lower courts had protected prior to McNally.172
Jeffrey Skilling was an Enron executive charged with investment fraud under the new statute; he challenged the new law as unconstitutionally vague.173 The Roberts Court acknowledged that “Skilling’s vagueness challenge has force,” but invoked the avoidance canon, noting that “[i]t has long been our practice, [ ] before striking a federal statute as impermissibly vague, to consider whether the prescription is amenable to a limiting construction.”174 The Court observed that the history of Court-Congress interactions demonstrated that Congress intended for the new statute “to refer to and incorporate the honest-services doctrine recognized in Court of Appeals’ decisions before McNally derailed the intangible-rights theory of fraud.”175 It then found that while the pre-McNally case law was inconsistent about the outer reaches of the honest-services doctrine, the “vast majority” of honest-services cases decided before McNally involved offenders who participated in bribery or kickback schemes, in violation of a fiduciary duty.176 In order to “preserve the statute without transgressing constitutional limitations,” the Court thus read the new law to criminalize “only the bribe-and-kickback core of the pre-McNally case law” and not to apply to investment frauds such as that committed by Skilling.177
Although the majority opinion in Skilling limited the reach of the new honest-services statute, it did so in a manner that very consciously sought to preserve coverage of the primary conduct associated with the doctrine—and the only conduct indisputably intended to be covered by Congress. Indeed, because pre-McNally case law was unclear about what behavior beyond bribes and kickbacks counted as honest-services fraud, it is uncertain whether the behavior left out by the Court’s construction fell within Congress’s consideration and whether legislators had any specific intent to include or not include it within the new statute’s reach.
Further, in applying the avoidance canon to limit the statute’s coverage, the Court made clear that the alternative was near-certain constitutional invalidation on vagueness grounds.178 In other words, the limiting construction was the only way to give effect to Congress’s clearly expressed intent to override McNally. Thus, the Court was not using a merely possible or speculative constitutional concern as an excuse to rewrite the statute, as some scholars have argued that the avoidance canon encourages it to do.179 In short, although the Skilling majority employed the avoidance canon in a manner that cut back the scope of the honest-services statute, it seemed highly attuned to congressional intent in so doing. For this reason, the case does not seem to fit within the conventional account of the relationship between the avoidance canon and congressional intent.
A second example of this form of substantive canon attentiveness to congressional intent occurred in Gonzales v Carhart,180 in which the Court considered the validity of the federal Partial-Birth Abortion Ban Act of 2003.181 That statute was enacted following the Court’s decision in Stenberg v Carhart,182 which held that a Nebraska “partial-birth abortion” statute violated the federal Constitution because its ban applied too broadly—covering the commonly used dilation and evacuation procedure (“D&E”) as well as the dilation and extraction procedure known as “partial-birth” abortion (“D&X”).183 In enacting the statute, Congress was responding directly to the Stenberg ruling and sought to avoid the Nebraska statute’s constitutional difficulties by being more precise about the procedures to which it applied.184 The federal Act was very explicit about Congress’s intent to remedy the deficiencies the Court had found in the Nebraska statute; in fact, its opening provisions expressly referenced the Court’s opinion in Stenberg.185 Several doctors and the pro-choice organization Planned Parenthood challenged the new federal Act, arguing that despite Congress’s attempts at precision, it too was unconstitutionally vague.186
Before addressing the Act’s constitutionality, the Court sought to determine the scope of the Act’s operation and effect. The Act’s challengers claimed that it was void for vagueness because, like the Nebraska statute, it was unclear about whether it reached D&E procedures other than the “partial-birth” procedure.187 The Court disagreed, construing the Act to cover only the “partial-birth” procedure and not to prohibit other D&E procedures.188 In so doing, the Court highlighted numerous places in which the Act deliberately differed from the Nebraska statute, observing that Congress sought “to meet the [ ] objections” the Court articulated in Stenberg and that “[t]he Act makes the distinction the Nebraska statute failed to draw.”189 The Court then referenced the canon of constitutional avoidance, noting that the canon “extinguishes any lingering doubt as to whether the Act covers the prototypical D&E procedure” because it requires that “every reasonable construction must be resorted to, in order to save a statute from unconstitutionality.”190
In two prominent cases and a dissenting opinion in a third case, the members of the Roberts Court also used (or advocated using) the avoidance canon to preserve a politically salient liberal statute from invalidation. The majority opinions in the first two cases were written by Roberts, and constituted notable instances of a conservative justice using a substantive canon to reach a liberal result.193 The third case contained a dissenting opinion penned by Stevens that invoked the avoidance canon to argue for the preservation of a high-stakes liberal statute.194
Perhaps the most famous of the cases is National Federation of Independent Business v Sebelius195 (“NFIB”), in which a divided Roberts Court upheld most provisions of the Obama administration’s Affordable Care Act.196 The key provision at issue was the individual mandate, which requires individuals to purchase a health insurance policy providing a minimum level of coverage or be forced to pay a penalty to the IRS when they file their income taxes.197 Challengers claimed that the individual mandate exceeded Congress’s constitutional authority.198 The Obama administration argued that Congress had the power to mandate the purchase of health insurance under the Commerce Clause, under the Necessary and Proper Clause, and, as a fallback, under its power to “lay and collect Taxes.”199
A majority of the Court ruled that the health-care law exceeded Congress’s constitutional authority under the Commerce Clause and the Necessary and Proper Clause.200 However, a different majority of the Court upheld the mandate under Congress’s taxing power.201 Roberts’s opinion acknowledged that “[t]he most straightforward reading of the mandate is that it commands individuals to purchase insurance.”202 But it went on to construe the mandate to merely “establish[ ] a condition—not owning health insurance—that triggers a tax” and to hold that, under this reading, the mandate was “not a legal command to buy insurance.”203 In so doing, the opinion relied heavily on the avoidance canon. “The question,” Roberts stated, “is not whether [the tax reading] is the most natural interpretation of the mandate, but only whether it is a ‘fairly possible’ one.”204 As he explained, “every reasonable construction must be resorted to, in order to save a statute from unconstitutionality.”205 Notably, Roberts’s opinion laid bare the institutional stakes involved, commenting that “[t]he Government asks us to interpret the mandate as imposing a tax, if it would otherwise violate the Constitution. Granting the Act the full measure of deference owed to federal statutes, it can be so read.”206
These statements make clear that the Court’s interpretation was driven by a strong commitment to preserving the statute before it if at all possible. Indeed, Roberts openly was influenced by institutional comity considerations—as evidenced by his explicit references to the “deference owed to federal statutes” and the fact that he was reading the statute as “[t]he Government ask[ed].”207 Contrary to the conventional avoidance canon criticism, NFIB simply does not read as a case in which the Court used the canon to limit the scope of a statute to which it was hostile, or to fulfill judicial policy preferences at the expense of legislative ones. Rather, it reads as a case in which the Court used the avoidance canon to embrace an alternate reading that the government—that is, the president with the presumptive support of the enacting Congress—asked it to embrace.208
It is worth noting that NFIB involved an extraordinarily visible, high-stakes statute enacted by a recent Congress. This may explain why Roberts—hardly an intentionalist by philosophy209 —voted to uphold the statute. As the chief justice, Roberts may have been particularly concerned about institutional considerations, including the need to respect the legislative branch, preserve Congress’s work product, and protect the legitimacy of the Court, in a case that was closely watched and certain to generate significant public attention. Some have read his avoidance canon opinions less charitably, suggesting that he has used the canon to perform radical surgery on statutory text, and that he has done so in order to pave the way for future constitutional invalidation.210 I find this sinister view to be a bit extreme—indeed, there is nothing in NFIB that suggests the Court will invalidate the Affordable Care Act on constitutional grounds in the future.211 I discuss Roberts’s relatively high rate of reference to substantive canons in general in Part II.B.1.
The Roberts Court also has applied substantive canons in a manner that seeks to approximate Congress’s intent in run-of-the-mill cases that are not politically salient. Consider, for example, Howard Delivery Service, Inc v Zurich American Insurance Co,212 in which the Court invoked a substantive canon dictating that provisions in the Bankruptcy Code should be narrowly construed.213 Howard involved a Bankruptcy Code section that “accords a priority, among unsecured creditors’ claims, for unpaid ‘wages, salaries, or commissions’ and for unpaid contributions to ‘an employee benefit plan.’”214 It was uncontested that the provision covered “fringe benefits that complete a pay package”—for example, retirement plans and “group health, life, and disability insurance.”215 Howard raised the question whether the priority provision “also encompasse[d] claims for unpaid premiums on a policy purchased by an employer to cover its workers’ compensation liability.”216 The Court held that the priority provision did not cover such unpaid premiums.217 In so ruling, it discussed the evolutionary history of the priority provision, citing legislative history explaining that the provision covers “health insurance programs, life insurance plans, pension funds, and all other forms of employee compensation that [are] not in the form of wages.”218 The Court then observed that workers’ compensation does not compensate employees for work performed but, rather, for on-the-job injuries, and that workers’ compensation regimes substitute for tort liability, not for wage payments.219 Finally, the Court emphasized that one of the Bankruptcy Code’s underlying objectives is to ensure the “equal distribution” of payments to all creditors and invoked the “corollary” presumption that provisions in the Bankruptcy Code allowing preferences must be narrowly construed.220
Throughout its discussion, the Court’s focus was on Congress’s intent and design with respect to the scope of the priority provision. Unsatisfied that that intent and design encompassed the premium payments at issue in the case, the Court fell back on the Bankruptcy Code’s underlying equal distribution purpose, and a substantive canon based on that underlying purpose. The Court thus seemed to use the “narrow construction of Bankruptcy Code preferences” canon as a means to effectuate Congress’s intent and purpose—not as a vehicle for judicial abrogation of legislative policy preferences.
A number of the Court’s preemption cases also contain judicial efforts to divine congressional intent, often while invoking the presumption against preemption. In Wyeth v Levine,221 for example, a majority of the Court concluded that the Federal Food, Drug, and Cosmetic Act222 (FDCA) does not preempt state law tort claims alleging that manufacturers placed inadequate warnings on drug labels.223 In so ruling, the Court invoked the presumption against preemption in conjunction with the Act’s purpose, legislative history, congressional intent, and other canons.224 Notably, the Court emphasized that Congress enacted the FDCA in order to “bolster consumer protection against harmful products”—quoting congressional hearing testimony and floor statements to show that Congress intended to continue to allow state law claims of this kind when it enacted the FDCA.225 The Court also stressed Congress’s “awareness of the prevalence of state tort litigation” and argued that Congress’s failure to expressly preempt state tort law on inadequate warnings in labels despite such awareness “is powerful evidence that Congress did not intend FDA oversight to be the exclusive means of ensuring drug safety and effectiveness.”226 Thus, far from using the preemption canon as a blunt instrument to curtail congressional intent in furtherance of judicially preferred policies, the Court employed the canon to help reinforce its sense of Congress’s intent. Moreover, it did not use the canon to supersede statutory purpose, legislative history, or other indicia of legislative intent; instead, it paid significant attention to these intent-focused interpretive resources and employed a substantive canon to achieve a result consistent with the reading dictated by these other resources.
Similarly, in Riegel v Medtronic, Inc,227 the Court considered whether the Medical Device Amendments of 1976228 (MDA)’s preemption clause bars common-law claims challenging the design or labeling of a medical device once the design or label has gained premarket approval from the FDA.229 This time, the Court concluded that the federal statute did preempt the state law claims.230 Justice Ruth Bader Ginsburg’s dissenting opinion argued that the MDA should not be read to preempt state common-law claims seeking compensation for injuries caused by defectively designed or labeled medical devices because Congress did not intend for the MDA to radically curtail state common-law suits; rather, in enacting the MDA, Congress was trying to preempt only state premarket regulation of medical devices.231 The dissent began its analysis with the presumption against preemption and then invoked House and Senate reports describing several medical device failures and resulting injuries that motivated Congress to enact the MDA,232 as well as comments by the Act’s sponsor indicating that “[this] legislation is written so that the benefit of the doubt is always given to the consumer.”233 The dissent also made a “dog that did not bark” legislative intent argument, insisting that “[i]t is difficult to believe that Congress would, without comment, remove all means of judicial recourse for large numbers of consumers injured by defective medical devices.”234 Thus, the dissent used the preemption canon in tandem with and as a means to fulfill congressional intent, not as a tool for trumping legislative intent with judicial preferences.
As noted above, in offering these examples of the justices’ attentiveness to congressional intent, I do not mean to suggest that the Court does not at times use substantive canons in a manner that defeats or contradicts congressional intent. My argument is based purely on examples, and opposing examples surely exist. Indeed, some substantive canons are by nature designed to check Congress and, therefore, almost always are applied in a manner that restrains congressional intent. Federalism clear statement rules impose deliberate obstacles to congressional interference with state authority, and the Roberts Court sometimes applied these without focusing on congressional intent.235 The rule of lenity, likewise, gives the benefit of the doubt to criminal defendants when Congress has not been clear enough about the reach of a criminal statute, even though it is almost always the case that Congress would prefer for the benefit of the doubt to go to the government.236 My point in providing the case examples in this Section is not to argue that the conventional view that judges use substantive canons to defeat legislative intent is wrong per se but, rather, that it is overstated. Substantive canons can be used, and sometimes are used, to defeat congressional intent—but they also regularly are used to further congressional intent. In other words, there is nothing inherent about substantive canons as an interpretive resource that necessitates that they will be used to contradict legislative preferences. Moreover, this is true even with respect to substantive canons that are designed to check Congress, such as the avoidance canon and the presumption against federal preemption of state laws, as the cases discussed in this Section show. Finally, it is significant that two of the Roberts Court’s most frequent employers of substantive canons—Stevens and Sotomayor—are intentionalist judges who view congressional intent as the lodestar of statutory interpretation. The fact that these intent-seeking justices invoked substantive canons at higher rates than did their textualist counterparts is strong evidence that the conventional account misses something important about textualism’s relationship to substantive canons237 and about the relationship between substantive canons and congressional intent.
The data reported in Part II have important theoretical implications, calling into question several conventional assumptions about substantive canons and the work that they perform in statutory interpretation. This Part explores the theoretical upshot of the data regarding the Roberts Court’s use of substantive canons. It argues that the data suggest that substantive canons have received a “bum rap” from academics, who have overestimated the role that such canons play in the Court’s statutory interpretation cases—and that such canons may not be as dangerous or dastardly as commentators have painted them out to be. The Article posits that one reason for the mismatch between academic perception and reality likely is an excessive scholarly focus on the federalism clear statement rules, which represent only a fraction of all the substantive canons employed by the Court.238 The data also suggest that Justice Scalia has received a bit of a bum rap for ostensibly using substantive canons aggressively, while criticizing them in theory; in reality, he invoked such canons in only nine of eighty-two opinions authored during the period studied and relied primarily on them in only three opinions.
The data also have important implications for two leading theories of statutory interpretation. First, they undermine long-standing scholarly assumptions that substantive canons provide a necessary “escape valve” for textualist judges who find a statute’s plain meaning unpalatable. It turns out that, in practice, textualist judges invoke substantive canons rarely, and less often than some of their intentionalist colleagues. Further, during the period studied, textualist judges invoked substantive canons to support liberal outcomes in a substantial minority of cases—sometimes at rates equal to the rates at which they invoked such canons to support conservative outcomes.239
At the same time, the data lend some support to pragmatic theories of statutory interpretation—demonstrating that practical reasoning, along with the Court’s own statutory precedents, plays a gap-filling or escape valve role similar to that which scholars have attributed to substantive canons. This Part concludes by discussing how the substantive canons employed most often by the Roberts Court align with state and federal legislators’ preferences regarding interpretive methodology and what the substantive canon data might add to recent discussions about methodological stare decisis—the suggestion that federal courts should adopt a binding statutory interpretation methodology.
As discussed in Part I.A, scholars of different theoretical stripes tend to share the views that textualist judges employ substantive canons regularly, that there is theoretical tension between substantive canons and textualism, and that substantive canons play a significant role in the interpretation of statutes. Further, many scholars view substantive canons as trump cards that judges illegitimately whip out to justify a construction that conflicts with the one dictated by other interpretive resources, including legislative intent.
But the data and doctrinal analysis of the Roberts Court’s first six and a half terms suggest that these scholarly assumptions may be overstated—and that, as an interpretive tool, substantive canons may not be as nefarious, or the Court may not be as wily in its use of them, as scholars have imagined. Substantive canons are infrequently invoked, and even when invoked, they are relied on only in passing or as a secondary interpretive resource in the vast majority of opinions (71.3 percent).240 Indeed, the much-maligned classic case in which the Court pulls a substantive canon out of a hat and uses it to reject a statutory reading dictated by the statute’s text, purpose, and other traditional tools may be the exception rather than the norm.
Of course, there is a difference between measuring the Court’s quantitative use of substantive canons and measuring its qualitative use of such canons. The fact that the justices infrequently invoke substantive canons does not mean that such canons are not powerful when used; nor do the data showing that the justices rely “primarily” on such canons in only 28.7 percent of opinions and that they appear to have used substantive canons to defeat congressional intent in only fifteen cases mean that those particular cases are not worth worrying about. Indeed, some may argue that even one nefarious use of a substantive canon to circumvent the statute’s plain meaning or congressional intent is odious—and can work sufficient harm to justify the academic outcry against such canons. Some might even view the Court’s infrequent and inconsistent use of substantive canons as more problematic than the widespread use of such canons would be, because infrequent use fails to put Congress on notice about the background norms that might suddenly pop up to do significant work in the odd case, and thereby increases the likelihood that Congress’s intent will be undermined. These are legitimate worries. I do not necessarily disagree with individualized criticisms of the Court’s use of particular substantive canons in particular cases, or mean to suggest that substantive canons are harmless merely because they are infrequently used. My point, rather, is that the data suggest that there is nothing inherently harmful, or intent defeating, or plain meaning eschewing, about substantive canons as an interpretive resource—and that the prevailing scholarly wisdom about substantive canons as a category has swept too broadly.
In addition, textualist judges and the substantive canons as a group may deserve greater credit for neutrality than scholars or the conventional account has given them. The ideological data reported in Tables 4 and 5, showing that the Roberts Court’s textualist and textualist-leaning justices—who also are its most conservative justices—have employed substantive canons to reach liberal results in a substantial percentage of opinions,241 are stunning. The data suggest that substantive canons may have some, at least mild, constraining effect on the conservative justices, and that the standard scholarly account that the justices wield substantive canons in a purely ideological manner, to justify judicial policy preferences, is not accurate in many cases. Indeed, scholars may owe both the conservative justices and the substantive canons, as an interpretive tool, an apology—and the latter a more charitable place in the statutory interpretation annals.
Scalia in particular may deserve more credit than scholars have given him for being consistent, in practice, with his theoretical criticism of substantive canons. As noted in Part II.B, during the Roberts Court’s first six and a half terms, Scalia rarely invoked substantive canons in the opinions he authored (11.0 percent).242 If we remove cases in which Scalia authored an opinion invoking the rule of lenity—a substantive canon he has argued is justified based on its antiquity243 —then his rate of use for this interpretive resource falls to 7.3 percent.244 Further, Scalia joined or authored an opinion that invoked substantive canons in only 16.9 percent (50 of 296) of the cases in the data set,245 and only 3 (out of 9) of the opinions he authored that invoked substantive canons relied on a substantive canon as a primary interpretive resource.246 Thus, Scalia seems to have hewed more closely to his theoretical criticism of substantive canons, at least in recent years, than scholars have given him credit for.
The puzzle thus arises: What accounts for the mismatch between the conventional view of substantive canons and actual judicial practice during the Roberts Court’s first six and a half years? Or for the differences in how many substantive canons the Court invoked during the Burger/Rehnquist Court years versus the Roberts Court years? One explanation may be that much of the conventional account—and criticism—regarding substantive canons seems to have been formulated in response to the Court’s creation and aggressive application of several new “super-strong” clear statement rules during the 1980s and 1990s.247 As Professors Eskridge and Frickey have pointed out, these clear statement rules were a “remarkable” and “striking innovation,” transforming background norms about federalism and sovereign immunity into nearly insurmountable obstacles to federal legislation that seeks to supersede state laws or waive the government’s immunity from suit.248 Clear statement rules also are extraordinarily countermajoritarian and dismissive of congressional intent in that they “permit the Court to override probable congressional preferences in statutory interpretation in favor of norms and values favored by the Court.”249 Worse, super-strong clear statement rules sanction this judicial displacement of congressional preferences with virtually no notice to Congress, which may be aware of the background norm the Court seeks to enforce but receives no warning that it must be exceptionally clear about its intent to override that norm in the text of the statute—rather than, say, in the legislative history—until the Court issues its decision several years later. As Eskridge and Frickey have observed, this notice problem is particularly disturbing when the Court applies a new super-strong clear statement rule to an older statute, enacted during an era when different statutory interpretation techniques predominated and courts rarely required clear statements of any kind in the statute’s text.250
Given the radical effects worked by clear statement rules, the Rehnquist Court’s invention of several new and “super-strong” versions of these rules garnered significant academic attention251 and criticism and may—uncoincidentally—have distorted scholars’ views of substantive canons as a category. That is, scholars may have been a little too quick to attribute the characteristics associated with clear statement rules—for example, their centrality to textualist jurisprudence and their inattentiveness to congressional intent—to all substantive canons. Moreover, the sheer shock value of the Court’s creation and use of these “super-strong” clear statement canons during the early Rehnquist Court years may have created a heightened perception of the amount of work that substantive canons as a whole perform in the Court’s statutory interpretations. I say “heightened perception” because, in reality, the Rehnquist Court may not have used substantive canons more often than the Roberts Court; recall that the Brudney-Ditslear study found low overall rates of judicial references to substantive canons in employment law cases decided by the Rehnquist Court during its 1986 through 2002 terms (15.6 percent).252
Of course, there also are other factors that may explain why substantive canons seem to be doing less work on the Roberts Court than scholars have assumed, and why such canons are being used less frequently and with less variety by the Roberts Court than during the early years of the Rehnquist Court.253 One such factor is that federalism clear statement rules may have been in their heyday during the early Rehnquist Court—the time period that Eskridge and Frickey happened to study—so that such canons may have been used more frequently and in greater variety during the period they studied than during other periods. That is, the early Rehnquist Court may have perceived a need to stake out strong protections for underenforced constitutional principles254 related to federalism and sovereign immunity and may have acted with vigor and urgency to articulate canons that helped it do so in its early years. By the time of the early Roberts Court, twenty years later, those battles may have been fought and won, so that the members of the Court—even those who employed substantive canons energetically during the early Rehnquist years—have found less need to create or invoke such canons than they once did. Relatedly, because federalism clear statement rules are now well established, cases involving such rules may no longer need to be decided by the Supreme Court. That is, lower courts may be applying such clear statement rules to resolve cases before they even reach the Court; if this is true, then focusing on the Supreme Court’s docket could understate the practical importance of substantive canons—particularly ones that establish bright-line rules that are easy for lower courts to implement.
Another, related, explanation could be that the Supreme Court has not undergone a significant political or ideological shift from the Rehnquist Court to the Roberts Court—and, indeed, seems to have become more solidly conservative since the early Rehnquist years (1986 through 1991).255 Thus, the canons created, or modified, by the Rehnquist Court may suit the justices on the Roberts Court—or at least a majority of those justices—quite well, and the justices accordingly may see little need to invent new canons as the Rehnquist Court once did. This could explain the low incidence of “new canon” creation observed in my study, and perhaps also the decrease in the variety of different canons invoked—that is, the current Court may be well settled into a comfort zone with respect to the canons it prefers to invoke, whereas the Rehnquist Court may have been in a substantive canon transition period of sorts. Finally, part of the explanation for the differences in findings may lie in the fact that, since 2010, the US Supreme Court and Congress have been more closely aligned, politically and ideologically (skewing conservative), than they were during the first decade of the Rehnquist Court—when both houses of Congress were controlled by liberal majorities, while the Court was moderately conservative.256 To the extent that the Court uses substantive canons to cut back the scope of legislation, it may see less need to use such canons in times of political alignment with Congress. Note, however, that while this theory is consistent with the Roberts Court’s use of substantive canons in cases such as Carhart,257 it cannot explain Chief Justice Roberts’s use of substantive canons in cases like NAMUDNO258 or NFIB,259 which upheld statutes enacted by Democrat-controlled Congresses.
As discussed in Part I.A, leading legislation scholars have theorized that substantive canons operate as an “escape valve” for textualism, providing a justification—couched in time-honored rules—for construing a statute in a manner that conflicts with its plain meaning.260 However, the findings reported in Part II showing that textualist and textualist-leaning justices infrequently invoked substantive canons call this assumption into question.261 Moreover, the findings raise the following queries: If substantive canons do not regularly serve as an escape valve for textualist justices, does this mean that textualism in fact has no escape valve? Or is some other interpretive tool or technique doing the work of filling in gaps and providing an escape mechanism for textualism?
More granular analysis of the Roberts Court’s cases suggests that textualism may in fact have an alternate escape valve, or valves—namely, Supreme Court precedent, practical consequences, and perhaps also the common law and other statutory tools. In order to explore the answers to the above questions, I examined the thirty-six out of eighty-two cases in the data set authored by Scalia that did not reference statutory text or the plain meaning rule and the twenty-nine out of seventy-seven cases in the data set authored by Justice Thomas that did not reference statutory text or plain meaning and sought to identify what interpretive resources these textualist justices relied on when they did not base their statutory constructions on text or plain meaning. In these sixty-five cases, Scalia and Thomas overwhelmingly relied on Supreme Court precedent in lieu of statutory text or plain meaning; in a handful of cases, they also referenced practical consequences and, to a lesser extent, the common law and other statutes.262 If we expand this closer study to examine the 333 opinions in the entire data set that did not reference text or plain meaning (authored by all of the justices), Supreme Court precedent and practical consequences again stand out as the two most frequently referenced alternate interpretive resources.263 Thus, Supreme Court precedent seems to be the real gap-filling interpretive tool in the Court’s jurisprudence, followed by practical-consequences-based reasoning.
The Supreme Court’s frequent use of practical consequences in statutory interpretation cases has been explored in several earlier empirical studies.264 But the Court’s use of its own precedent as an interpretive resource has gone largely unstudied.265 Scholars may thus far have ignored the Court’s use of its own precedents to construe statutes because precedent is such an intrinsic part of judging and judicial opinion writing that the Court’s reliance on it has hardly seemed remarkable. But the data from the Roberts Court’s first several terms—showing Supreme Court precedent to be the interpretive resource most frequently invoked across justices266 —suggest that it is important for scholars to examine critically how the Court uses such precedents when tackling a new statutory construction. Does the Court use its own prior interpretations to trump other interpretive tools, or does the Court use such precedents to corroborate the reading dictated by other interpretive resources? How often does the Court employ statutory stare decisis to decide a case? Are arguments from precedent typically used to “escape” the statute’s plain meaning, to reinforce it, or simply to fill in gaps when there is no plain meaning? These are just a few of the many questions that bear further exploration. At bottom, statutory interpretation theory lacks a good account of whether the Court uses its own precedents merely to supply authority for other canons and interpretive tools (for example, as authority for the rule of lenity or to explain what noscitur a sociis means) or as an independent interpretive tool that itself dictates a particular statutory construction.
The discovery that textualist judges employ Supreme Court precedent more often than substantive canons when construing statutes that do not seem to have a plain meaning also could have important implications for textualism’s claims to predictability and determinacy. Here is what I mean: Textualists long have argued that their approach to statutory interpretation is the most legitimate because it constrains judges, leading straightforwardly to the one correct reading of a statute.267 Critics have countered that textualism’s dependence on substantive canons undermines such claims to determinacy because substantive canons are policy based and unpredictable and because they empower judges to manipulate outcomes268 —points that prominent textualists, including Scalia, have conceded.269
The surprising finding that the Roberts Court relies on the Court’s own precedents to fill in gaps in statutory context far more often than it relies on substantive canons could change the contours of this debate in important ways. On the one hand, Supreme Court precedent cannot be invented and reinvented the way that substantive canons sometimes can—for example, the federalism clear statement rules in the 1980s and 1990s. In this sense, the Court’s heavy reliance on its own precedents may be less problematic than similar reliance on substantive canons. On the other hand, reliance on prior judicial decisions is a very common-law, judge-centered approach to statutory interpretation—not an approach that puts the legislature at its center, as statutory interpreters are supposed to do. Indeed, most foundational theories of statutory interpretation rest, to some degree, on a faithful agent model of the Court-Congress relationship—one that envisions the Court as the agent in a master-servant relationship with Congress.270 Textualism, notably, claims that it is more faithful to the legislature than approaches that look to statutory purpose or legislative history, because it pays close attention to the text that is the product of compromise among differently motivated legislators, rather than elevating the views of some legislators over others’ as suits the judge’s fancy.271 But textualism’s form of faithfulness, it turns out, relies significantly on a judicially generated interpretive resource—the Court’s own past precedents. This is significant because there is, of course, substantial room for different interpretations of what precedents mean when applied to new circumstances—rendering the use of past precedent as a guide to statutory interpretation a notoriously unpredictable and judge-empowering exercise.272 Indeed, legal realist Professor Karl Llewellyn delineated no fewer than sixty-four different ways in which judges can apply a prior precedent.273
Further, even before judges get to the point of arguing over a precedent’s meaning, they have significant discretion to decide whether a precedent is even applicable in a particular case, just as they have discretion to decide whether a substantive canon is applicable in a particular case. Recent studies are just beginning to explore the ways in which the members of the Supreme Court disagree, or “duel,” over the application of the same precedent in the same case.274 But the lesson from this Article may simply be that more analysis of this kind is necessary—that is, that scholars need to pay closer attention to how textualists (and other jurists) employ the Court’s prior precedents, no less than they have paid attention in the past to textualists’ use of substantive canons. Ultimately, whether textualists supplement textual analysis with substantive canons or with precedent, they are supplementing with a judicial source rather than a legislative source—and that, in itself, is both enormously interesting and potentially worrisome.
This Article’s findings also have important theoretical implications for other leading theories of statutory interpretation. Purposivism is an interpretive approach that encourages jurists to interpret a statute by identifying the statute’s purpose and selecting the meaning that best effectuates that purpose.275 Purposive statutory interpretation typically involves inquiries into legislative history, the societal problem that prompted the legislature to enact the statute, legislative intent, and other sources that might shed light on a statute’s objectives. It can entail guesswork and judicial discretion, but is often defended on the ground that reliance on legislative history and purpose helps restrict judicial discretion and fulfill congressional intent.276
Purposivism once was the dominant approach to statutory interpretation, but over the past few decades it has come under significant attack.277 The criticism has come both from textualism and from public-choice theorists, who have condemned it as providing a rose-colored-glasses view of the legislative process. Both sets of critics have emphasized the open-endedness of the search for statutory purpose and have argued that purpose-based interpretation enables judges to import their personal policy preferences into the statute.278
The data regarding the Roberts Court’s substantive canon use have at least two important implications for purposivism. First and foremost, the data show that despite textualism’s thirty-year-old campaign against legislative history—and despite scholars’ warnings that substantive canon use has increased while legislative history use has decreased279 —substantive canons have not displaced legislative history on the modern Supreme Court. As noted earlier, eight of the eleven justices who have served on the Roberts Court—including conservative, textualist-leaning Justices Alito and Kennedy—referenced legislative history more often than they referenced substantive canons in the opinions they authored.280 Moreover, a comparison of my data to Professors Brudney and Ditslear’s data from the Burger and Rehnquist Courts shows that the conventional scholarly wisdom is only half correct: legislative history use did decline significantly from the Burger Court (46.6 percent) to the Rehnquist (27.7 percent) and Roberts Courts (25.0 percent),281 but it has not been replaced by substantive canon use. Indeed, while substantive canon use increased from 8.3 percent during the Burger Court to 15.6 percent during the Rehnquist Court and 14.4 percent during the Roberts Court,282 this 6.1 percentage point increase hardly makes up for the 21.6 percentage point drop in legislative history references since the Burger Court, nor does it show that substantive canon use has surged ahead of legislative history use.
This is significant because, as scholars have noted, an important part of what divides textualists from purposivists on the modern US Supreme Court supposedly is disagreement about what interpretive tools should be consulted when there is textual ambiguity (with textualists putting substantive canons second, while purposivists list legislative history and purpose in that slot).283 The Roberts Court data suggest that scholars have been too quick to sound the alarm bells about the death of legislative history at the hand of substantive canons, or about the inversion of these two tools in the hierarchy of interpretive resources to be consulted. Judicial reliance on legislative history may have declined, but substantive canons have not filled the resulting void.
Second, substantive canons may have something to offer purposivism as an interpretive tool. That is, given the doctrinal evidence discussed in Part II.C showing that substantive canons such as the avoidance canon at least sometimes are used to fulfill congressional intent, and given purposivist/intentionalist Justices Stevens’s and Sotomayor’s willingness to invoke such canons in a nontrivial percentage of the opinions they author, purposivism may stand to benefit, as an interpretive theory, from expanding its interpretive tool kit to encompass certain substantive canons or to acknowledge ways in which substantive canons can be used to reinforce statutory purpose.
Scholars have long operated on the assumption that judges rely significantly on the canons of construction when interpreting statutes.284 This study shows that assumption to be false with respect to substantive canons, in terms of both the frequency of the Court’s references and the weight it places on substantive canons when it references them. But the study also, surprisingly, shows that the Court relies on traditional tools of legal analysis, including Supreme Court precedent, practical consequences, and its own sense of what the words in a statute mean, at much higher rates than it relies on most statute-specific interpretive tools.285 That is, the Court invokes its own prior precedents, practical consequences, and text / plain meaning with far more frequency than it references statutory purpose, legislative history, congressional intent, or dictionary definitions.286 The only statute-specific interpretive tool the Court invokes at rates somewhat comparable to the traditional legal tools of analysis is the combined language canons + whole-act rule + grammar canons tool.287
This observation highlights one of the central questions in statutory interpretation theory—that is, what role do the canons play? In light of the data, it is worth asking whether the canons are really playing much of a role at all in the Roberts Court’s statutory cases—or whether they are, indeed, “window-dressing,” as legal realist Llewellyn once argued.288 Leading legislation scholars have debated these questions for decades, with Eskridge, Frickey, and Professor Cass Sunstein countering Llewellyn’s charge and arguing that the canons play an important role, at least as default rules, in guiding courts to certain statutory constructions.289 But the data from the Roberts Court suggest that the truth may lie somewhere in between the two opposing positions scholars have staked out. That is, the answer may depend on whether we are talking about substantive canons or about language canons (defined to include the whole-act rule and grammar rules). Based on the frequency of references to particular interpretive resources, it appears that the Court does employ language canons often when construing statutes (in 32.3 percent of all opinions and 67.9 percent of all majority opinions in the data set, compared to a rate of reference of only 14.4 percent of all opinions and 15.5 percent of majority opinions for substantive canons).290
Further work and analysis is necessary to fully understand the differences between the Court’s use of language and its use of substantive canons, but a few possibilities exist. Whole-act rule comparisons or lists susceptible to various Latin maxims may, for example, be more readily available in a larger number of cases than are on-point substantive canons; if the Court is straightforward in its use of substantive canons, employing them as helpful guides rather than inventing and reinventing them as after-the-fact justifications for interpretations reached based on policy preferences, it may simply be finding fewer occasions in which substantive canons, as compared to language canons, are helpful. Alternately, the members of the Roberts Court may consider language canons and whole-act rule comparisons to be more persuasive interpretive tools than substantive canons—and may thus employ the former more often in an effort to persuade colleagues to join an opinion. Future studies might find it useful to explore these and other possible reasons for the Court’s dramatically different rates of reference to these two forms of interpretive canons.
The data regarding the Roberts Court’s substantive canon use also shed new light on recent calls for methodological stare decisis—the suggestion that federal courts should adopt a binding statutory interpretation methodology. Recent work has revealed that several state courts and state legislatures have attempted to dictate a binding hierarchy of interpretive tools that courts in their jurisdictions must follow, in a prescribed order—what has been called “methodological stare decisis.”291 A leading example is the three-step interpretive framework adopted by the Oregon Supreme Court in the case Portland General Electric Co v Bureau of Labor and Industries.292 That framework imposed a three-tier hierarchy of interpretive tools that required courts to consult textual canons at step one, legislative history at step two if and only if the textual canons prove inconclusive, and substantive canons at step three, only as a last resort if both textual tools and legislative history prove indeterminate.293 One scholar has labeled this approach “modified textualism” and has suggested that it may constitute a compromise solution that provides textualism’s best hope of convincing purposivist judges to adopt a text-centric interpretive methodology.294
Relatedly, every state legislature in the country has enacted legislation codifying certain canons of construction that state courts are expected to follow when interpreting statutes.295 A recent study of these legislated interpretive rules found that only a handful of judicially created substantive canons make it onto state legislatures’ lists of preferred interpretive tools.296 Among these favored substantive canons are the avoidance canon, the presumption against retroactive application of later-enacted statutes, and the sovereign immunity canon.297
What is striking about both the state courts’ “modified textualism” and state legislatures’ lists of codified canons is how much they overlap with the Roberts Court’s current approach to substantive canons. Notably, modified textualism puts substantive canons last in the hierarchy of interpretive tools, well behind plain meaning, text-based language canons and the whole-act rule, other statutes, and legislative history.298 The members of the Roberts Court, as Part II showed, likewise invoke substantive canons far less often than they invoke these other interpretive resources.299
The big difference between the Court’s practices and the methodological approach adopted by several state courts, then, is the state courts’ commitment not to consult substantive canons at all when textual canons or legislative history provides a clear answer. The data from the Roberts Court’s first six and a half terms are not consistent with such a commitment—the justices frequently employed substantive canons alongside text / plain meaning, language canons / the whole-act rule, or other statutes (in sixty-two of eighty-seven opinions).
There are also several canons that many state legislatures have codified that never show up in the Roberts Court’s cases, such as the presumption that statutes should be construed so as to “promote justice” or the presumption that remedial statutes should be liberally construed.305 Still, the fact that three out of six of the substantive canons most often referenced by the Roberts Court appear on the short list of substantive canons favored by state legislatures suggests a noteworthy degree of consensus between the US Supreme Court and legislators about which policy-based interpretive rules are most helpful and authoritative in the judicial interpretation of statutes. It also reinforces the point emphasized in Part II.C—that the judicial use of substantive canons may not be nearly as frustrating to legislative intent in practice as scholars have assumed. Both of these points of convergence, in turn, suggest that substantive canons may be less threatening, in terms of democratic legitimacy, to the judicial review of statutes than scholars have feared—on the theory that legislatures are not likely to endorse interpretive rules that empower courts to undermine or usurp legislative authority.
At the same time, however, the data from the Roberts Court’s first six and a half terms suggest that neither state courts’ interpretive hierarchies nor state legislatures’ codified rules account adequately for the role that judicial precedent plays in statutory interpretation. Some state legislative codes of construction do mention the common law, but none lists judicial precedent or the courts’ constructions of similar statutes among the interpretive tools courts should consult. On the one hand, state legislatures’ failure to include precedent in their legislated rules may not be shocking, because legislatures may consider references to precedent to be an inherent part of judicial decision-making, rather than an independent resource or rule/maxim/canon that itself dictates a particular interpretation. State courts, similarly, may consider precedent to be something that courts use as authority for the various textual rules or other interpretive canons they employ, rather than an independent interpretive tool that directs courts to adopt a particular construction and requires its own place in an interpretive hierarchy. As discussed in Part III.B, statutory interpretation theory lacks a clear understanding of whether precedents are used mostly as authority for other interpretive tools or as interpretive guides that themselves dictate a particular statutory reading. Again, future work examining how precisely the Court uses its own precedents in statutory cases would be illuminating.306
At the federal level, Congress has not prescribed any codes or rules of statutory construction for courts to follow.307 But recent empirical work based on interviews with congressional staffers provides some evidence about Congress’s views regarding several of the substantive canons invoked most often by the Roberts Court.308 Most interestingly, such work reveals that while congressional staffers do not know the avoidance canon by name, they want and expect courts to interpret statutes in a manner that is consistent with the canon’s underlying assumptions.309 That is, the staffers responsible for drafting legislation indicate that they try hard to legislate within constitutional bounds, pay attention to prior case law on questions of constitutionality, and expect the Court to err on the side of “upholding federal statutes” whenever possible.310 This, again, is consistent with the doctrinal analysis in Part II.C demonstrating that the Roberts Court has often used the avoidance canon to uphold federal statutes against constitutional challenges and to give effect to congressional overrides. In other words, despite academic criticism of the avoidance canon, there seems to be widespread legislative support for the canon as a saving mechanism, and the Court, perhaps surprisingly, seems to be employing the canon consistently with how legislative staffers expect the canon to be used.
The Roberts Court’s regular use of federalism canons and the rule of lenity, by contrast, seems more problematic when measured against congressional expectations. When interviewed, congressional staff demonstrated little knowledge of the federalism canons or the rule of lenity,311 and many reported an expectation that when a conflict arises between state and federal laws, courts should give greater effect to the federal law—the precise opposite of the presumption embodied in the federalism canons.312
Both the federal and state evidence regarding legislator preferences, then, suggest that scholars have gotten the story half right regarding substantive canons. That is, with respect to the federalism canons and the rule of lenity, the academic consensus that substantive canons operate in a manner that contradicts legislative preferences seems accurate. But for several other canons regularly employed by the Roberts Court, including the much-criticized avoidance canon, academic concerns that the Court invents or prioritizes substantive canons that preserve judicial rather than legislative preferences may be unwarranted. Indeed, the canons most often invoked by the Court seem to align with those that state legislators and, to some extent, federal legislators endorse.
At bottom, the data regarding the Roberts Court’s use of substantive canons teach both that (1) a methodological stare decisis that ranks substantive canons low in its interpretive hierarchy might be closer to the modern Court’s practices and less difficult to garner support for—even from textualists—than scholars thus far have assumed; and (2) with two notable exceptions, there may be greater consistency between legislative preferences about substantive canons and the modern Court’s on-the-ground use of such canons than scholars have recognized.
Substantive canons long have received a bum rap from legal scholars. Commentators have argued that there are too many substantive canons, that courts invent and reinvent them to suit their whims, that textualist judges use substantive canons as an escape valve to end-run plain meaning, and that courts employ substantive canons in a manner that elevates judicial policy concerns over legislative intent. Most of these claims have been based on casual observations, with little empirical testing. This Article offers the first detailed empirical study of the modern Court’s substantive canon use—with several surprising results. Its findings suggest that much of the conventional academic account of substantive canons may be wrong, or at least overstated. Throughout, the Article’s aim has been to illuminate the Court’s actual practices regarding substantive canons—and to evaluate the implications that those practices have for statutory interpretation theory.
| Case Name | Canon | Opinion Author | Ideology |
|---|---|---|---|
| *National Federation of Independent Business v Sebelius, 132 S Ct 2566 (2012) | Avoidance Presumption favoring severability | Roberts (m) | Liberal |
| ØReynolds v United States, 132 S Ct 975 (2012) | Avoidance Rule of lenity | Scalia (d) (avoidance) Breyer (m) (rule of lenity) | Conservative Liberal |
| ΩBrown v Plata, 563 US 493 (2011) | Avoidance | Kennedy (m) | Liberal |
| *Citizens United v Federal Election Commission, 558 US 310 (2010) | Avoidance | Stevens (c/d) | Liberal |
| ΩJerman v Carlisle, McNellie, Rini, Kramer Ulrich, LPA, 559 US 573 (2010) | Avoidance | Kennedy (d) | Conservative |
| *Holder v Humanitarian Law Project, 561 US 1 (2010) | Avoidance | Breyer (d) | Liberal |
| *Skilling v United States, 561 US 358 (2010) (override) | Avoidance Rule of lenity | Ginsburg (m) | Liberal |
| ΩBartlett v Strickland, 556 US 1 (2009) | Avoidance | Kennedy (p) | Conservative |
| ΩHawaii v Office of Hawaiian Affairs, 556 US 163 (2009) | Avoidance | Alito (m) (u) | Unspecifiable |
| *Northwest Austin Municipal Utility District Number One v Holder, 557 US 193 (2009) (“NAMUDNO”) | Avoidance Equal sovereignty principle (presumption against unequal treatment of the states) | Roberts (m) | Liberal |
| ØFederal Communications Commission v Fox Television Stations, Inc, 556 US 502 (2009) | Avoidance | Breyer (d) | Liberal |
| *Boumediene v Bush, 553 US 723 (2008) (override) | Avoidance | Roberts (d) | Conservative |
| ØGonzalez v United States, 553 US 242 (2008) | Avoidance | Thomas (d) | Liberal |
| *Gonzales v Carhart, 550 US 124 (2007) (override) | Avoidance | Kennedy (m) | Conservative |
| ΩOffice of Senator Mark Dayton v Hanson, 550 US 511 (2007) | Avoidance | Stevens (m) (u) | Conservative |
| ØRapanos v United States, 547 US 715 (2006) | Avoidance Federalism clear statement (state has primary power over land and water use) | Scalia (p) | Conservative |
| *Arizona v United States, 132 S Ct 2492 (2012) | Presumption against preemption | Alito (c/d) | Conservative |
| ΩChamber of Commerce of the United States of America v Whiting, 563 US 582 (2011) | Presumption against preemption | Roberts (m) | Conservative |
| ΩPLIVA, Inc v Mensing, 564 US 604 (2011) | Presumption against preemption Non obstante (allowing implied repeal) | Sotomayor (d) (presumption against preemption) Thomas (p) (non obstante) | Conservative Liberal |
| ΩWyeth v Levine, 555 US 555 (2009) | Presumption against preemption | Stevens (m) | Conservative |
| ΩAltria Group, Inc v Good, 555 US 70 (2008) | Presumption against preemption | Stevens (m) | Conservative |
| ΩRiegel v Medtronic, Inc, 552 US 312 (2008) | Presumption against preemption | Ginsburg (d) | Conservative |
| ΩWatters v Wachovia Bank, NA, 550 US 1 (2007) | Presumption against preemption | Stevens (d) | Conservative |
| ØEmpire HealthChoice Assurance, Inc v McVeigh, 547 US 677 (2006) | Presumption against preemption | Ginsburg (m) | Conservative |
| ØFowler v United States, 563 US 668 (2011) | Federalism clear statement (presumption against changing the federal-state balance of crime prosecution) Rule of lenity | Breyer (m) (federalism clear statement) Scalia (c) (rule of lenity and federalism clear statement) | Liberal Liberal |
| ØBruesewitz v Wyeth LLC, 562 US 223 (2011) | Federalism clear statement (police powers not superseded by a federal act) | Sotomayor (d) | Conservative |
| ΩSmith v Bayer Corp, 564 US 299 (2011) | Doubts about federal court injunctions resolved in favor of allowing state proceedings | Kagan (m) | Conservative |
| *Florida Department of Revenue v Piccadilly Cafeterias, Inc, 554 US 33 (2008) | Federalism clear statement (exemption from state taxation) | Thomas (m) | Conservative |
| *Arlington Central School District Board of Education v Murphy, 548 US 291 (2006) | Federalism clear statement (notice Spending Clause condition) | Alito (m) | Conservative |
| ΩBarber v Thomas, 560 US 474 (2010) | Rule of lenity | Kennedy (d) | Liberal |
| ØUnited States v Hayes, 555 US 415 (2009) | Rule of lenity | Roberts (d) | Liberal |
| ΩØDean v United States, 556 US 568 (2009) | Rule of lenity | ØStevens (d) ΩBreyer (d) | Liberal Liberal |
| *ØUnited States v Santos, 553 US 507 (2008) | Rule of lenity | *Scalia (p) ØStevens (c) | Liberal Liberal |
| ΩUnited States v Rodriquez, 553 US 377 (2008) | Rule of lenity | Souter (d) | Liberal |
| ΩBegay v United States, 553 US 137 (2008) | Rule of lenity | Scalia (c) | Liberal |
| ØJames v United States, 550 US 192 (2007) | Rule of lenity | Scalia (d) | Liberal |
| *Federal Aviation Administration v Cooper, 132 S Ct 1441 (2012) | Sovereign immunity (waiver) | Alito (m) | Conservative |
| ØSossamon v Texas, 563 US 277 (2011) | Sovereign immunity (waiver) | Thomas (m) | Conservative |
| *Permanent Mission of India to the United Nations v City of New York, 551 US 193 (2007) | Sovereign immunity (waiver) | Stevens (d) | Conservative |
| ØDolan v United States Postal Service, 546 US 481 (2006) | Sovereign immunity (waiver) | Thomas (d) | Liberal |
| ØCarcieri v Salazar, 555 US 379 (2009) | Liberal construction in favor of Indian tribes | Stevens (d) | Liberal |
| ØBilski v Kappos, 561 US 593 (2010) | Elephants in mouseholes | Stevens (c) | Unspecifiable |
| ØAli v Federal Bureau of Prisons, 552 US 214 (2008) | Elephants in mouseholes | Breyer (d) | Conservative |
| *Morrison v National Australia Bank Ltd, 561 US 247 (2010) | Extraterritorial application of law | Scalia (m) | Conservative |
| ΩMicrosoft Corp v AT&T Corp, 550 US 437 (2007) | Extraterritorial application of law | Ginsburg (m) | Liberal |
| ØExxon Shipping Co v Baker, 554 US 471 (2008) | No derogation of the common law | Souter (m) | Conservative |
| ØBP America Production Co v Burton, 549 US 84 (2006) | Narrow construction against the government | Alito (m) (u) | Liberal |
| ØKucana v Holder, 558 US 233 (2010) | Presumption favoring judicial review of administrative action | Ginsburg (m) | Liberal |
| *Sackett v Environmental Protection Agency, 132 S Ct 1367 (2012) | APA presumption of judicial review of all final agency action | Scalia (m) (u) | Liberal |
| ΩKnight v Commissioner of Internal Revenue, 552 US 181 (2008) | Burden on taxpayer seeking deduction | Roberts (m) (u) | Liberal |
| ØHamdan v Rumsfeld, 548 US 557 (2006) | Common law of war must clearly authorize trial by military commission Deference to the executive on matters of war | Stevens (m) (common law must clearly authorize) Thomas (d) (deference to the executive on war) | Liberal Conservative |
| ØCBOCS West, Inc v Humphries, 553 US 442 (2008) | Affirmative evidence of congressional intent necessary for an implied remedy | Thomas (d) | Conservative |
| *ΩVartelas v Holder, 132 S Ct 1479 (2012) | Presumption against retroactivity | *Ginsburg (m) ΩScalia (d) | Liberal Conservative |
| ØCarr v United States, 560 US 438 (2010) | Presumption against retroactivity | Sotomayor (m) | Liberal |
| ØAT&T Corp v Hulteen, 556 US 701 (2009) | Presumption against retroactivity | Souter (m) | Conservative |
| *Fernandez-Vargas v Gonzales, 548 US 30 (2006) | Presumption against retroactivity | Souter (m) | Conservative |
| *eBay Inc v MercExchange, LLC, 547 US 388 (2006) | Presumption against major departures from the “long tradition” of equity practice | Thomas (m) (u) Roberts (c) | Liberal Liberal |
| ØStoneridge Investment Partners, LLC v Scientific-Atlanta, Inc, 552 US 148 (2008) | No expansion of private rights of action beyond congressional intent | Kennedy (m) | Conservative |
| *Holland v Florida, 560 US 631 (2010) | Presumption in favor of equitable tolling (nonjurisdictional statutes) Constitutional right to access federal courts | Breyer (m) | Liberal |
| ØWoodford v Ngo, 548 US 81 (2006) | Constitutional right to access federal courts | Stevens (d) | Liberal |
| ΩGonzalez v Thaler, 132 S Ct 641 (2012) | Clear statement rule for statutory limits on jurisdiction | Sotomayor (m) | Conservative |
| *Mims v Arrow Financial Services, LLC, 132 S Ct 740 (2012) | Presumption in favor of concurrent state court jurisdiction | Ginsburg (m) (u) | Liberal |
| ΩHardt v Reliance Standard Life Insurance Co, 560 US 242 (2010) | Clear statement rule for attorney’s fee award | Thomas (m) | Liberal |
| Subject-Matter-Specific Canons | |||
| ØHenderson v Shinseki, 562 US 428 (2011) | Ambiguity resolved in favor of veterans | Alito (m) (u) | Liberal |
| ØShinseki v Sanders, 556 US 396 (2009) | Ambiguity resolved in favor of veterans | Souter (d) | Liberal |
| ΩKawashima v Holder, 132 S Ct 1166 (2012) | Ambiguity construed in favor of aliens | Ginsburg (d) | Liberal |
| ØCarachuri-Rosendo v Holder, 560 US 563 (2010) | Ambiguity construed in favor of aliens | Stevens (m) | Liberal |
| *Felkner v Jackson, 562 US 594 (2011) | AEDPA imposes a highly deferential standard (state court decisions are entitled to the benefit of doubt) | Per curiam (m) (u) | Conservative |
| ΩRiley v Kennedy, 553 US 406 (2008) | Liberal construction of the Voting Rights Act | Stevens (d) | Liberal |
| ΩHoward Delivery Service, Inc v Zurich American Insurance Co, 547 US 651 (2006) | Narrow construction of preferences in the Bankruptcy Code “Equality of distribution for similar creditors” principle | Ginsburg (m) (narrow construction and equality of distribution) Kennedy (d) (equality of distribution) | Liberal Conservative |
| ΩTravelers Casualty & Surety Co of America v Pacific Gas & Electric Co, 549 US 443 (2007) | Clear statement rule for disallowing state law claims | Alito (m) (u) | Liberal |
| ØHall Street Associates, LLC v Mattel, Inc, 552 US 576 (2008) | Presumption in favor of freedom to contract (to add to what FAA says) | Stevens (d) | Liberal |
| ØHall v United States, 132 S Ct 1882 (2012) | Presumption that the Bankruptcy Code does not erode past practice absent clear intent | Sotomayor (m) | Liberal |
| ΩHamilton v Lanning, 560 US 505 (2010) | Presumption that the Bankruptcy Code does not erode past practice absent clear intent | Alito (m) | Liberal |
| *CompuCredit Corp v Greenwood, 132 S Ct 665 (2012) | Presumption in favor of arbitration | Sotomayor (c) | Conservative |
| ØNken v Holder, 556 US 418 (2009) | Presumption “favoring the retention of long-established and familiar principles” | Roberts (m) | Liberal |
| *Board of Trustees of the Leland Stanford Junior University v Roche Molecular Systems, Inc, 563 US 776 (2011) | Presumption that an inventor owns a property right in a patentable invention Presumption denying inventors patent rights growing out of research for which the public has already paid | Roberts (m) (inventor owns property right) Breyer (d) (deny if research paid for by public) | Conservative Liberal |