Printer Friendly

Shadow precedents and the separation of powers: statutory interpretation of congressional overrides.

In both judicial decisions and critical commentary on statutory interpretation, the possibility of congressional override is generally considered a significant balance to the countermajoritarian reality that courts, through statutory interpretation, make policy. This Article demonstrates that the "check" on judicial power provided by overrides is not as robust as is typically assumed. Despite the importance routinely ascribed to overrides, the actual effect of overrides has received surprisingly little attention within the academic community. This is perhaps because one might assume that overridden precedents are functionally erased or reversed. But because Congress technically cannot overrule a prior decision, courts must determine whether the enactment of an override fully supersedes the prior judicial interpretation. Thus overrides raise unique, and previously largely ignored, questions of statutory interpretation.

Using examples from employment discrimination, an area of the law where Congress frequently overrides Supreme Court decisions, this Article demonstrates that the Supreme Court and lower courts often narrowly construe the significance of congressional overrides and instead rely on the prior judicial interpretation of statutes as expressed in overridden precedents. Thus, for example, although Congress clearly disagreed with a Supreme Court decision holding that pregnancy discrimination is not sex discrimination, lower courts, noting that the statutory language of the override only explicitly references "pregnancy, childbirth, or related medical conditions," continue to apply the reasoning employed by the Court in that overridden case when faced with sex discrimination claims in other contexts. I call this phenomenon reliance on "shadow precedents."

The Article shows how reliance on shadow precedents threatens legislative supremacy and undermines the standard rationales offered for adherence to precedent. It argues that, in drafting overrides, Congress should strive to clarify the extent to which it disagrees with the prior judicial interpretation. It also argues that courts should adopt interpretive conventions that are more respectful of the significance of the enactment of an override: (1) a rebuttable presumption that an override supersedes the judicial interpretation of the pre-existing statutory language, thus requiring "fresh" interpretation of the original statute as well as the override, and (2) a clear rule that overridden interpretations are no longer binding on lower courts.
INTRODUCTION

   I. STATUTORY OVERRIDES AND THEIR ROLE IN THE SEPARATION
      OF POWERS
      A. Overrides As a Check on Judicial Law-Making
      B. Empirical Studies of Overrides

  II. THE INTERACTION OF PRECEDENT AND OVERRIDES
      A. The Standard Rule of Precedent
      B. Overrides and the Creation of "Shadow Precedents"
      C. Overrides and Retroactivity

III. EXAMPLES OF SHADOW PRECEDENTS IN EMPLOYMENT
     DISCRIMINATION JURISPRUDENCE
     A. Federal Employment Discrimination Law and the 1991 Civil
        Rights Act
     B. Lorance v. AT&T Technologies
     C. Price Waterhouse v. Hopkins
     D. General Electric Company v. Gilbert
     E. Different Approach: The "Nullification" of Patterson v.
        McLean Credit Union

IV.  REDUCING RELIANCE ON SHADOW PRECEDENTS
     A. Problems with Reliance on Shadow Precedents
     B. Potential Congressional Responses--and Their Limitations
     C. Proposed Interpretive Reforms
        1. "Fresh" Statutory Analysis with a Rebuttable
           Presumption that the Prior Judicial Interpretation
           Is Superseded
        2. Overridden Precedents Should Not Bind Lower
           Courts
        3. Sample Applications
        4. Scope of the Proposal
     D. Benefits of the Proposed Reforms

CONCLUSION


INTRODUCTION
   When a court says to a legislature: 'You (or your predecessor)
   meant X," it almost invites the legislature to answer: "We did
   not." (1)


In passing legislation to override a judicial interpretation of a statute, Congress does indeed answer, "We did not." An override thus raises fundamental questions regarding the relative power of Congress and the courts. Using examples from employment discrimination jurisprudence, this Article demonstrates that courts often continue to follow statutory interpretation precedents whose holdings have been repudiated by Congress. Accordingly, although overrides are generally lauded as a significant balance to the countermajoritarian reality that courts, through statutory interpretation, make policy, the "check" they offer on the judicial branch is far less robust than is typically assumed.

The role that overrides are presumed to play in promoting legislative supremacy depends on two propositions. The first is that Congress monitors court decisions in statutory cases closely enough to be aware of interpretations with which it disagrees. Although, traditionally, there was skepticism regarding this point, empirical scholarship makes clear that Congress does respond in some fashion to many statutory decisions. (2) For example, a recent study found that about half of the Supreme Court's tax decisions were at least mentioned in Congress, and that Congress enacted statutory language either overriding or explicitly codifying approximately fifteen percent of the interpretations. (3)

The second necessary proposition is that overrides do, in fact, override: that is, that by enacting an override Congress effectively supersedes statutory interpretations by the courts. The validity of this second proposition has typically been assumed rather than established. A recent empirical study, however, found that high levels of judicial "dissensus" often remain after the passage of an override. (4) This Article is the first attempt to craft a theoretical framework to understand the interpretive puzzles inherent in analyzing overrides. (5) It demonstrates that, because Congress technically cannot overrule judicial decisions, the interpretation of overrides poses a particular challenge within a judicial system that is built on adherence to precedent.

One might assume that overridden precedents simply have no more effect. The reality, however, is far more complex. Overrides are often drafted relatively narrowly to respond to a particular judicial holding. Difficult questions therefore arise when a court faces a new factual scenario that is arguably "relevantly similar" to the issue addressed in the precedent case (such that, but for the override, it would be controlled by the precedent) but not clearly within the statutory language of the override itself. Courts must then determine whether the rationales expressed in the overridden decision continue to govern interpretation of the statute and the corollary question of whether the enactment of the override can be understood to reinterpret the preexisting statutory language. Both the Supreme Court and lower courts resolve these questions in ad hoc and inconsistent manners.

Lower courts, in particular, are faced with competing signals: they must apply the law as Congress enacts it, but they must also follow Supreme Court (and circuit court) precedent to the extent that it is not clearly superseded. Some interpret overrides as fully supplanting the precedent case, an understanding that is often suggested by legislative history associated with the override. Other courts interpret overrides as carving out an exception to a general interpretive principle announced in the precedent case and deem any factual applications of such a principle that Congress "fails" to address in the override language to continue to be governed by the precedent case. Thus, for example, even though Congress clearly disagreed with a Supreme Court decision holding that discrimination on the basis of pregnancy is not sex discrimination, many lower courts, noting that the text of the override only specifically addressed "pregnancy, childbirth, or related medical conditions," apply the reasoning employed by the Court in that overridden case when faced with sex discrimination claims in other contexts. (6) Notwithstanding Congress' repudiation of the central holding of a precedent, the precedent continues to hold sway. This is a phenomenon that I call reliance on "shadow precedents."

Statutory precedents typically serve two functions. They can provide an interpretive gloss on statutory language that is applied to future cases decided under that statute. (7) They can also govern interpretation of relevantly similar language in other statutes. (8) Shadow precedents arise in both contexts. And, although adherence to precedent typically promotes fairness, predictability, and efficiency, (9) reliance on shadow precedents often undermines these objectives. Cases that are similar are treated differently because some are governed by the override and others by the overridden precedent. Under the current ad hoc approach, individuals cannot reasonably predict whether conduct will be governed by the override or by the rationales expressed in the shadow precedent. And courts must not only resolve the difficult threshold question of whether to apply the override or the shadow precedent but also develop a parallel body of case law further interpreting the implications of precedents whose primary holdings have been repudiated by Congress. Moreover, to the extent that reliance on shadow precedents is counter to Congress' intent, the practice undermines the central role that overrides are expected to play in ensuring legislative supremacy.

Employment discrimination is an area where this problem often takes center stage because Congress frequently disagrees with Supreme Court interpretations of Title VII and other employment discrimination laws. The 1991 Civil Rights Act, (10) for example, overrode twelve decisions by the Supreme Court concerning employment discrimination and related attorneys' fees issues. (11) In September 2008, Congress enacted an ADA Amendments Act (12) which overrides several employment decisions under the Americans with Disabilities Act. The same Congress also considered a proposed override of a recent pay discrimination decision (a decision that itself relied heavily on a shadow precedent); the override passed the House but fell subject to a controversial filibuster in the Senate. (13)

This Article uses examples from employment discrimination jurisprudence to illustrate the interpretive moves that give rise to shadow precedents. (14) It looks at courts' analysis of the overrides of three significant Supreme Court cases: Lorance v. AT&T Technologies, Inc., (15) concerning when the statute of limitations to challenge a discriminatory act begins to run; Price Waterhouse v. Hopkins, (16) concerning the standard used to assess employment decisions in which both legitimate and discriminatory factors play a role; and General Electric Co. v. Gilbert, (17) concerning whether discrimination on the basis of pregnancy is sex discrimination. The analytic lens offered by this Article reveals a common theme connecting these seemingly unrelated subjects: the general tendency by courts to construe narrowly the significance of Congress' disapproval of prior holdings and instead rely upon the statutory analysis contained in the overridden decisions. Since in this context the court interpretations are consistently less protective of employee rights than the overrides, individual employees are denied protections they arguably should have. Thus, the analysis in the Article offers new approaches for advocacy in unsettled areas of employment discrimination jurisprudence.

More generally, this Article advocates for greater weight to be accorded to the significance of a statutory override of a precedent. It might seem appropriate simply to put the onus on Congress to do a "better" job in drafting overrides. And indeed, Congress should, in statutory language, strive to make clear whether it agrees with the interpretive rationales expressed in the precedent case. However, it would be extremely difficult, and in some cases impossible, for Congress to draft legislative language in an override comprehensively enough to anticipate all future factual scenarios that might be governed by a precedent with which it disagrees. More fundamentally, interpretive conventions should recognize that in enacting an override, Congress has the power to reinterpret the preexisting statutory language as well as to add new statutory language.

The Article therefore proposes a rebuttable presumption that enactment of an override supersedes the prior judicial interpretation of the statute and further that lower courts are no longer bound by such overridden precedents. Like other conventions of statutory interpretation, this would be a general rule of thumb that would aid in properly implementing congressional intent. Of course, to the extent that Congress consciously chooses to enact a narrow override that is simply an exception to the general rule established by a judicial interpretation, it could do so by partially codifying and partially overriding a holding. The approach I propose, however, would shift legislative "silence" related to other applications of the interpretive rationale expressed in the overridden case from presumptive acquiescence to presumptive disapproval. This approach would refocus attention on the actual statutory language rather than on the interpretive gloss provided by the courts. In other words, even if Congress had not expressed a specific intent regarding how the statute would apply to a factual scenario not squarely addressed by the language of the override, courts would then do "fresh" statutory analysis rather than mechanistically applying a prior judicial interpretation that Congress had already rejected in its original context. This proposed rule would better serve the principles of legislative supremacy and the values typically advanced by following precedent. Overrides are expected to play a central role in ensuring that Congress has the authority to shape statutory law. Reliance on shadow precedents often distorts the proper balance among the branches of government. Thus this issue is likely to become even more pressing in the near future with an increasingly conservative Supreme Court and an increasingly liberal Congress.

The Article proceeds as follows. Part I discusses the theoretical significance attached to overrides as an expression of legislative supremacy and reviews empirical research on overrides. Part II summarizes the standard rationales for precedent, identifies the interpretive challenges posed by overrides, and introduces the concept of shadow precedents. Part III provides several examples of shadow precedents in employment discrimination jurisprudence. Part IV advocates replacing the current ad hoc approach with a rule that "fresh" statutory analysis is required and that the rationales expressed in overridden precedents are no longer binding on lower courts.

I. STATUTORY OVERRIDES AND THEIR ROLE IN THE SEPARATION OF POWERS

A. Overrides As a Check on Judicial Lawmaking

The Constitution rests the responsibility for "[a]ll legislative Powers" in Congress. (18) The judiciary, by contrast, is charged with the interpretation of such law. (19) At least rhetorically, both courts and commentators ascribe to the principle of "legislative supremacy." There are nuances to exactly how the primacy of the legislature should be understood, (20) but the basic structure is simple: courts have the power to strike down statutes that violate the Constitution, but in all other respects, Congress has absolute authority to determine the contours of statutory law. Moreover, if Congress chooses to delegate interpretive functions to executive agencies, modern Supreme Court doctrine requires courts to defer to the agencies' reasonable interpretations. (21)

If it were always apparent how to apply a statute to a given factual dispute, this principle would be easy to apply. In fact, however, it has long been recognized and accepted that "interpretation is inescapably a kind of legislation." (22) Courts, in the process of deciding cases, discover ambiguities in statutes. They must fill in gaps and determine how to apply statutes to changing circumstances. (23) Even committed textualists such as Justice Scalia are willing, in certain circumstances, to ignore the plain text of the statute to correct perceived legislative mistakes or to avoid absurd results. (24) In other words, courts, through interpretation, substantively "make law." Thus, as Jane Schacter has put it, "[e]very judicial interpretation of a statute" is an "interbranch encounter" that "carries with it a lesson about democratic theory." (25)

The acknowledgment that courts, through the process of statutory interpretation, play a lawmaking function threatens basic precepts of legislative supremacy by suggesting that courts may thus usurp the legislative role. This poses a classic countermajoritarian difficulty (26): unelected judges, rather than democratically elected legislators (or agency personnel who are at least theoretically responsive to democratic forces as members of the executive branch), are making policy. The countermajoritarian nature of statutory interpretation, however, typically receives little attention because, as opposed to constitutional decisionmaking, there is a ready check on the judiciary: if Congress disagrees with a judicial interpretation of a law, it may override that interpretation by passing a new statute or amending an existing statute. (27) Congressional overrides, in other words, are a primary mechanism through which the core principle of legislative supremacy is presumed to be maintained. (28)

Thus, not surprisingly, the possibility of overrides informs both doctrinal and theoretical discussions of the process of statutory interpretation and, particularly, the precedential weight to be awarded to statutory interpretation decisions. The Supreme Court, which applies a "super-strong" presumption of stare decisis for statutory decisions, justifies its approach by stating that generally updating or correction of statutes should be done by Congress through the enactment of overrides rather than by the courts. (29) Accordingly, it may continue to follow a precedent even if it admits that the initial interpretation is incorrect or illogical (30) (although in certain instances, the Court will overrule its own statutory precedents). (31) Likewise, courts cite the possibility of override as a justification for literal application of a statute that seems to include a drafting error or result in an unacceptable policy decision (32) (although, as noted above, sometimes courts will take it upon themselves to "correct" such "mistakes").

Commentators who express widely divergent views regarding whether the Supreme Court's approach is appropriate nonetheless agree that overrides are essential to maintain separation of powers and protect legislative supremacy. At one end of a spectrum are commentators, including Lawrence Marshall and Edward Levi, who argue that courts should adopt a policy of absolute stare decisis in statutory cases because this would be the best way to ensure that Congress plays its constitutionally appointed role as primary lawmaker. (33) At the other end of the spectrum are commentators, including William Eskridge, T. Alexander Aleinikoff, and Einer Elhauge, (34) who posit that courts should (and do) feel comfortable updating statutes to reflect changing circumstances or the desires of the current Congress, even if such interpretations depart from the intentions of the enacting Congress. More radically, Guido Calabresi has suggested that courts should be empowered to nullify statutes they deem obsolete. (35) While embracing the lawmaking power inherent in statutory interpretation, such "dynamic" theorists typically also explicitly state that congressional overrides serve as a necessary check against the possibility that courts might take this license too far. (36) Courts may be granted so much freedom because Congress may always override them if they disagree. (37) Thus, commentators across this spectrum present overrides as a productive component of an ongoing "colloquy" between Congress and the courts. (38) Courts engage in a good faith effort to interpret statutes in line with congressional intent and they welcome congressional overrides if necessary to correct errors or update statutes.

A somewhat different picture of the role of overrides emerges from positive political theory literature because it characterizes both the judicial process and the legislative process in more self-interested terms. One strain of public choice literature characterizes the Supreme Court as a political entity in its own right that seeks to implement its own policy objectives. Such theorists recognize that overrides--and the threat of overrides--can serve as a limit on courts. They posit that the Supreme Court interprets statutes with an eye on the political make-up of the sitting Congress and the President, developing models that assume the Court will issue decisions on contested matters that are as close to its preferred policy outcomes as possible without triggering the risk of override. (39) Empirical studies offer at least some support for this characterization of the Court's statutory decisions. (40) Under such a framework, overrides are not evidence of a productive conversation between political branches; rather, they are a smackdown or pushback that the Court will strive to avoid.

On the other hand, while it is common to characterize overrides as a healthy correction to judicial policymaking, overrides may also reflect capture of the political process by an influential interest group. (41) This concern, too, grows out of public choice literature. Simplifying greatly, public choice theorists understand legislators as seeking, above all else, to ensure their reelection. (42) Such theorists caution that statutes may be enacted not to serve the public interest but rather "rent-seeking" special interests. (43) Some overrides may fall within this description, and the possibility of special interest overrides must be considered when evaluating the relationship between overrides and the precedent they address. (44)

All of these theorists assume that overrides play an important role in balancing the powers between the judicial and legislative branches. However, by failing to examine what happens after the override, these various characterizations of overrides tell only half the story. The specific examples in Part III demonstrate that courts retain considerable discretion in how they implement an override. They often interpret an override as establishing a narrow exception to a general rule and continue to rely upon the precedent that was overridden. In some cases, this may be what Congress intends. But to the extent that courts' reliance on overridden precedents is not in accordance with congressional intent, this practice suggests two alternative conclusions. The first is that judges, dutifully trying to implement congressional will, misread the signals sent by Congress in enacting overrides. As described more fully below, this could happen in part because some general rules of statutory interpretation do not work well with overrides and because lower court judges, in particular, may feel obliged to apply Supreme Court precedent unless there is no doubt that it has been completely overridden (and sometimes even then). The second is that judges seeking to implement their own policy objectives can do so notwithstanding the enactment of an override through consciously narrowing the scope of the override and justifying their approach by relying on the precedent that was overridden. Either characterization demonstrates that overrides often fail to play their role as a check on judicial lawmaking.

B. Empirical Studies of Overrides

Overrides are generally understood as statutes or statutory amendments that "supersede" statutory interpretation decisions by the Supreme Court or lower courts. In a classic empirical study of overrides, Professor William Eskridge defines overrides as statutes in which Congress "consciously" modifies a statutory interpretation decision by enacting a statute that:
   (1) completely overrules the holding of a statutory interpretation
   decision, just as a subsequent Court would overrule an
   unsatisfactory precedent; (2) modifies the result of a decision in
   some material way, such that the same case would have been decided
   differently; or (3) modifies the consequences of the decision, such
   that the same case would have been decided in the same way but
   subsequent cases would be decided differently. (45)


This definition has been employed in other empirical studies. (46)

It likewise describes the statutes that this Article treats as overrides, although my analysis focuses on those within the first prong of the definition. Note, however, that this definition includes, without analysis, an assumption that an override may "completely overrule the holding" of a decision "just as a subsequent Court would overrule an unsatisfactory precedent." In fact, as demonstrated in Parts II and III of this Article, a congressional override is not the equivalent of a decision by a "subsequent Court" to "overrule" a precedent. Rather, courts frequently continue to rely on the discredited precedent that was targeted by the override. It is precisely this gap that is identified and analyzed in this Article.

Nonetheless, working with this standard definition of overrides, we can better understand their role in the relationship between the political branches. Overrides were once considered relatively unusual. (47) As the significance of statutory decisions in our "statutorified" (48) world has increased, and as the size of congressional staffs has likewise increased, (49) overrides and other congressional reactions to statutory decisions have become more common. In recent years, a substantial majority of the Supreme Court's decisions have been statutory, (50) and a large number of these decisions were subsequently discussed in some manner in Congress. (51) Lower court decisions, particularly if there is a circuit split, often engender responses as well. (52)

There is a range of potential congressional reactions to statutory interpretation decisions by the courts. The most significant is enactment of legislation that codifies a decision, overrides a decision, or partially codifies and partially overrides a decision. Congress, or, more precisely, a committee of Congress, may also indicate its approval or disapproval of a statutory decision in legislative history, such as committee reports. (53) Congress may hold hearings or consider bills that would either override or codify decisions but that are not ultimately enacted. Finally, individual members of Congress may make statements on the floor regarding court decisions without introducing actual legislation.

In his influential study, Professor Eskridge found that between 1975 and 1990 (when his study was completed), each Congress overrode, on average, about a dozen of the Supreme Court's statutory decisions and about twice as many lower court decisions. (54) More recent studies make similar findings. (55) While this is a relatively small number of laws in absolute numbers, it is important to recognize that our constitutional system purposefully makes it relatively difficult to enact legislation, and the growing size of the congressional "docket" of proposed legislation further reduces the likelihood that any given bill will be enacted. In each session of Congress, there are more than 10,000 bills introduced and fewer than ten percent of these become law. (56) Significant obstacles include the gatekeeping role that committees and committee chairs play, the possibility of filibuster and other dilatory techniques, the necessity of passing both houses, and the possibility of presidential vetoes. (57)

Beyond these common barriers, additional hurdles stand in the way of enacting overrides. Overrides are, by definition, the result of court cases, and they are often court cases in which both parties have invested considerable resources. The winning party in the court case usually lobbies in Congress to protect its win. Thus, not surprisingly, overrides are more often enacted when the more politically powerful party loses in court. (58) That said, a recent empirical study found that in almost all instances, Congress heard testimony from both perspectives on a given issue when deliberating on an override. (59)

Given the barriers to enactment of legislation generally and overrides specifically, a rich body of doctrine and critical literature explores whether congressional failure to enact an override should be considered approval of, or at least acquiescence to, judicial interpretations of a statute. As Justice Scalia has famously observed (albeit in dissent), "congressional inaction is a canard." (60) Common arguments against presumed acquiescence include that it is improper to assume that Congress knows about judicial interpretations; that Congress may not agree with the interpretation but may nonetheless have other higher priorities; and that even an indication that the sitting Congress agreed with an interpretation does not mean that it is in accord with the intent of the enacting Congress, generally deemed the touchstone for proper judicial interpretation. (61) The debate over the relevance of failure to enact an override is unlikely to be resolved, but it should make one thing apparent: when Congress does actually enact an override it should be recognized as a significant expression of congressional dissatisfaction with the interpretation expressed in the precedent.

The actual effects of overrides, however, are much less clear. In 2004, Jeb Barnes, a political scientist at the University of Southern California, published the first relatively large-scale empirical study that attempts to assess the efficacy of overrides. (62) Barnes randomly selected 100 overrides passed between 1974 and 1990 and measured the extent to which substantial judicial "dissensus," defined as either a circuit split or significant intracircuit disagreement, remained after an override was enacted. (63) Barnes distinguished between what he classified as "prescriptive" overrides, defined as those in which Congress attempts to "resolve the override issue," and "partial" overrides, defined as those in which Congress overrides a decision or decisions but leaves development of a replacement standard to the courts either through explicit delegation or through enacting a "vague" statute. (64)

Barnes' findings are striking: almost exactly half of the overrides he studied yielded significant judicial dissensus. (65) Not surprisingly, there was judicial dissensus in one hundred percent of the overrides he coded as "partial overrides," in which Congress explicitly granted courts discretion. (66) The more shocking conclusion is his other finding: in more than one third of the overrides coded as "prescriptive" overrides, that is, those in which Congress attempts to resolve the override issue, substantial dissensus remains. (67) This finding calls into question the role that overrides are presumed to play as an expression of legislative supremacy. Notwithstanding a "prescriptive" override, courts continue to disagree about how to interpret or apply an override.

Barnes further found that the likelihood of judicial dissensus varied considerably based on the subject matter of the override. Tax overrides, for example, almost uniformly generated consensus; on the other hand, not a single civil rights override in his data set brought about consensus. (68) These findings also aligned with the extent to which judges' rulings were partisan: the greater the partisan divide in the pre-override period, the greater the likelihood that dissensus would result from even prescriptive overrides. (69)

Barnes' empirical study catalogues judicial consensus or dissensus--it does not attempt to analyze what interpretive moves permit or encourage significant judicial dissensus after an override. Moreover, while his findings regarding the significance of partisan divides obviously lend some credence to an assertion that ongoing dissensus is a result of willful disregard of congressional directives rather than a good faith confusion regarding the scope of the override, they do not explain how courts, working within the rule of law, can justify such disregard. The analysis that follows is intended to lay the groundwork for such an understanding, as well as to propose a solution that will better permit overrides to live up to their promised role under the separation of powers.

II. THE INTERACTION OF PRECEDENT AND OVERRIDES

Overrides, by definition, respond to a judicial precedent that interprets a statute. Because, unlike common law development, it is Congress rather than the courts that seeks to supersede the precedent, overrides pose unique--and heretofore largely ignored--questions of statutory interpretation and the relationship between the branches of government. This Part introduces the principle of "shadow precedents," which is then illustrated with several examples in the next Part.

A. The Standard Rule of Precedent

The basic structure of an argument from precedent is well known. An argument based on precedent relies on a showing that a court has decided a matter in a given way and that a subsequent dispute, which is relevantly similar, should be decided in the same way. (70) Traditionally, this was understood as a process of inductive legal reasoning to determine the holding and the ratio decidendi (that is, the rationale) of the prior case. (71) Today, however, American appellate courts issue written decisions setting forth in detail the reasoning for the holding and the rationales that support the holding. (72) The text of the decisions themselves becomes important "controlling" precedent, which will be applied in situations that are deemed relevantly similar to the precedent case.

Precedent works along two vectors, and this distinction, while familiar, will become important in the analysis of overrides. The first, typically described as "vertical" precedent, binds lower courts to follow precedents announced by higher courts. (73) Although federal court judges have life tenure, and thus cannot be sanctioned for failing to follow precedent, they generally regard adherence to precedent as a basic responsibility of judging. (74) Accordingly, they typically will not openly disregard precedent on point, although they may (and often do) distinguish precedent that others might consider controlling. Nonetheless, as one commentator has characterized it, distinguishing cannot violate the "red face test." (75)

The second vector, the jurisprudential rule of stare decisis, is typically characterized as "horizontal" precedent. (76) Although, as noted above, statutory precedents are accorded heightened precedential weight because of the possibility of overrides, horizontal precedent is less "binding" than vertical precedent. The Court has indicated that it may overrule its own statutory precedents where there has been a significant "intervening development of the law" through further judicial action or by Congress that has "removed or weakened the conceptual underpinnings from the primary decision" or rendered it "irreconcilable with competing legal doctrines or policies." (77)

The traditional justifications for adherence to precedent are that it promotes fairness (in that like cases will be treated alike), predictability, and efficiency. (78) Given its malleability, the question remains why judges adhere to precedent rather than simply exerting their own policy preferences. Contemporary scholars have developed theories ranging from the concept of judicial good faith, to an instrumental interest in exercising political power through a reciprocal commitment to uphold precedents, to a desire to increase personal leisure time. (79) Whichever explanation or combination of explanations is correct, precedent continues to exert a strong force in the legal system, particularly on lower courts.

As applied to statutes, precedent functions in two ways. The first is more typical of precedent in other contexts. A court interprets a word or phrase in a statute. That interpretation becomes part of what we understand the statute to mean. Future factual scenarios will be governed by the statutory language as interpreted through the precedent case. I refer to this as "substantive" statutory precedent. The second type of statutory precedent is a convention of statutory interpretation that provides that statutory provisions that are similar to each other will be interpreted similarly. (80) Although courts are inconsistent in following this rule, the presumption is particularly strong when the language of a later statute has been modeled on an earlier statute. I refer to this as "related statutes" statutory precedents.

Invoking the rhetoric of legislative supremacy, courts often justify following "related statutes" precedents by suggesting that Congress, in borrowing language from one statute when enacting a new statute, also purposefully intended to borrow authoritative interpretations of that statute. (81) However, the Supreme Court has suggested that even interpretations of the primary (model) statute that post-date enactment of the second statute govern interpretation of that second statute. (82) Accordingly, reliance on "related statute" precedent may also reflect either an implicit or explicit recognition that the court's reasoning with respect to statutory language or concepts in one statute is valid in another related context or a desire to keep the law developing in a relative uniform manner. Like adherence to substantive precedent, the application of consistent interpretations across different statutes typically helps promote fairness, predictability, and efficiency.

B. Overrides and the Creation of Shadow Precedents

Overrides are enacted in response to court decisions, but technically Congress cannot "overrule" a prior statutory decision. It can only enact statutory language that supersedes or "overrides" the prior decision. Up to and until the time when a court subsequently considers the effect of the override, the prior case is not officially overruled. As the examples in the next Part make clear, overrides often track relatively closely the specific holding or factual application of the decision they seek to supersede. Thus, a subsequent factual scenario that raises precisely the same questions under precisely the same statute as the case that led to the override will be easily resolved by the now-controlling statutory language enacted in the override. (Of course, if there is an aspect of an opinion that is not addressed in the override, courts properly continue to follow this aspect of the holding). (83) The difficult question is how to resolve factual disputes that are arguably similar to the issue addressed in the precedent case--such that they would naturally, but for the override, be governed by the rationales expressed in the precedent--but not directly addressed by the language of the statutory override. In resolving such matters, should a subsequent court rely upon the precedent that triggered the override?

The answer depends on what effect the override is deemed to have on the preexisting precedent and, relatedly, to what extent it may legitimately be understood as not only adding or removing statutory language but also as reinterpreting the preexisting statutory language. Conceptually, an override, particularly a relatively narrowly written override, could stand for several different propositions: that Congress generally agreed with the Court's interpretation and reasoning but carved out a specific exception that would supersede the specific holding of the case; that Congress disagreed with the specific holding of a case and specifically disagreed with the interpretive reasoning applied by the court as well, suggesting that the "opposite" (or at least different) reasoning should apply and thus that the preexisting statutory language should be understood differently; or that Congress, in passing the override, "nullified" the prior case, both its holding and underlying rationale, so that it is as if it had never been decided and should have no more precedential weight. Furthermore, Congress' override of a single holding--the end of a line of precedent--does not necessarily answer whether prior precedent, on which the discredited opinion rested, is also called into question.

The Supreme Court has never clearly articulated which of these conceptual approaches is correct or what standards should be used to choose among them. Rather, as the examples in the next Part make clear, courts' resolution of the issue tends to be ad hoc and inconsistent. But, strikingly, courts often choose the first proposition: that is, that notwithstanding Congress' clear override of the holding in a case--that is, the judicial interpretation of the statutory language as applied in the case at issue--the interpretation remains controlling with respect to any application not explicitly addressed by the override. This creates a phenomenon that I call "shadow precedents." A precedent, whose central holding has clearly been repudiated by Congress, continues to hold sway. As in Peter Pan, the shadow has an existence entirely divorced from its primary subject. (84)

If this result were clearly intended by Congress, reliance upon shadow precedents would be appropriate. In fact, however, Congress typically characterizes an override as "overruling" the prior decision. Even when statutory language is drafted narrowly, committee reports and other persuasive indicators of legislative intent often suggest that Congress disagrees with the rationale (85) of the decision as well as its specific result and expects the override to end reliance on the repudiated precedent. (86) Indeed, most strikingly, in at least some instances lower courts rely upon shadow precedents even when the Supreme Court itself has indicated that Congress, in passing an override, clearly rejected the rationale as well as the holding of the precedent case. (87)

Rather than reflecting congressional intent, the reliance on shadow precedents often stems from the central role of precedent in judicial interpretation. As noted above, lower courts are bound by Supreme Court precedent. They are also required to apply the law as Congress enacts it. To the extent that Congress is not clear in--or is not clearly understood as--overriding the rationale as well as the holding in a Supreme Court case, a lower court interpreting an override must do so in the face of conflicting signals: the precedent on the one hand and the override on the other. The easiest way to resolve these is to understand the override narrowly and apply the rationale of the precedent to the case at hand.

Courts also often suggest that reliance on shadow precedents is appropriate by referencing the standard proposition of statutory interpretation expressio unius est exclusio alterius, a doctrine that holds that expression of one thing implies exclusion of another. (88) Congress, by specifically addressing one application of a rationale but failing to specifically address others is presumed to have specifically excluded them. Likewise, Congress, by specifically amending one statute (that is, the statute that the Court was interpreting in the precedent that was overridden), but failing to amend others, is presumed to have specifically excluded them. The problem with these assumptions, as addressed more fully in Part IV, is that whatever merit they may hold when analyzing a statute as initially enacted, they do not accord very well with the way Congress considers issues when it drafts overrides.

Moreover, to the extent that the preexisting statutory language could have been interpreted in a different way (in accordance with Congress' intent as expressed in an override), the override itself is necessary to supersede the judicial interpretation, but it is also in some sense redundant if Congress, by enacting the override, intends to signal an intention to change the interpretation of the preexisting statutory language. But courts often deem Congress, by clarifying one application, to have implicitly excluded other equally plausible applications of the statutory language. In other words, in an example we will return to, if pregnancy discrimination is naturally understood as a form of sex discrimination, a clarifying amendment that overrides a Supreme Court decision to specify that pregnancy discrimination is in fact a form of sex discrimination should not be used as evidence that Congress did not intend discrimination based on breasffeeding, which also might naturally be understood as a form of sex discrimination, to not be recognized as a form of sex discrimination. But it is. (89)

Part III contains specific examples that illustrate these rather abstract principles and Part IV sketches a proposal for a different approach. Before turning to the examples, however, it is helpful to consider briefly the doctrine of retroactivity as applied to overrides because the Supreme Court's pronouncements in that context have sometimes (I think inappropriately) been used to justify reliance on shadow precedents.

C. Overrides and Retroactivity

Similar questions regarding the scope of an override and its effects on precedent arise when courts consider whether an override is to be applied retroactively. In that context, the Supreme Court has clearly indicated that absent clear congressional directives, overrides will, like other legislation, be applied prospectively only. The leading decisions are a pair of cases, Landgraf v. USI Film Product (90) and Rivers v. Roadway Express, Inc., (91) addressing the 1991 Civil Rights Act (the same Act that provides several of the examples for shadow precedents discussed in the next Part). Landgraf addressed whether new remedial provisions permitting damages, including punitive damages, would be replied retroactively; Rivers addressed whether a substantive override of an interpretation of a civil rights statute would be applied retroactively. In each case, the Court noted that there were competing concerns. On the one hand, retroactive application could require retrial of cases completed before the Act was enacted and could expose defendants to punitive damages for conduct that predated the Act. (92) On the other hand, "[p]urely prospective application ... would prolong the life of a remedial scheme, and of judicial constructions of civil rights statutes, that Congress obviously found wanting." (93) Noting that fairness generally dictates that individuals "know what the law is" so that they can conform their conduct to it and so that "settled expectations" are not disrupted, the Court declined to apply the statutes retroactively because it felt that Congress had not clearly indicated its intention to pass retroactive statutes. (94)

The petitioners in Rivers argued that retroactive application was more appropriate for the portions of the statute that overrode Supreme Court interpretations of substantive provisions than for those that expanded availability of damages as a remedy. (95) They reasoned that since most circuits, including the Sixth Circuit where the conduct at issue in Rivers had occurred, had interpreted the civil rights statute in accordance with the congressional override (rather than how the Court had interpreted it in the case that was overridden), there was no reason for concern regarding unfair surprise because the override merely implemented the law as the parties believed it to be at the time that they acted. (96) Notwithstanding what the Court characterized as the "equitable appeal" of the argument, (97) the Court rejected the claim, again simply on the grounds that Congress had not clearly made the statute retroactive. (98) The Supreme Court also noted that, under the hierarchical principles that guide statutory interpretation in the federal court system, a Supreme Court characterization of what a statute means is an "authoritative statement of what the statute meant before as well as after the decision of the case giving rise to that construction." (99) Thus, although the Supreme Court's interpretation ran counter to the unanimous interpretations of the courts of appeals prior to the decision, it did not "change" the law but rather "finally decided" what it had "always meant and explained why the Courts of Appeals had misinterpreted the will of the enacting Congress." (100)

Notably, however, there is no reason that the Court's determination that overrides will be presumptively prospective, and thus that disfavored precedents should continue to play a role in interpretation for a limited period of time, should narrow the effect of overrides after enactment. Indeed, in Rivers, the Court was clear that it understood that Congress indicated disapproval of its holding and cited, apparently positively, to petitioner's characterization of the override as "legislatively overrul[ing]" the Court's interpretation. (101) More importantly, the various competing concerns that the Court identified weigh very differently when one considers the prospective scope of overrides. The concern with "proper notice" and avoiding "unfair surprise" disappears (at least to the same extent that it does with any new legislation); individuals can plan their actions accordingly under the newly amended law. Meanwhile, the concern over perpetuation of a judicial construction that Congress obviously found wanting is sharply increased. The Court, in discussing the issue of retroactivity of overrides, seemed to assume that the disfavored interpretation would naturally disappear--the only question was how soon. Reliance on shadow precedents directly contradicts that assumption.

III. EXAMPLES OF SHADOW PRECEDENTS IN EMPLOYMENT DISCRIMINATION JURISPRUDENCE

Up to this point in the Article, the analysis has been general; in any area of statutory law, Congress may pass overrides and courts will be forced to interpret the relationship of the override statute and the discredited precedent. Specific examples, however, can help elucidate how these relatively abstract principles play out in practice. I have chosen to illustrate these precepts with examples drawn from employment discrimination law because it is an area of the law where Congress often objects to Supreme Court interpretations and thus passes overrides. Additionally, employment discrimination jurisprudence presents an ideal opportunity to examine the "related statute" issue because there are many statutory provisions that the Supreme Court has specified are to be interpreted in tandem.

In some respects, however, the significant role played by shadow precedents within employment discrimination cases may not be representative of overrides generally. As noted above, Jeb Barnes found in his empirical study that overrides in civil rights statutes, including employment ones, resulted in particularly high levels of judicial dissensus. (102) He suggests, and I agree, that this may be in part because the questions of equality and discrimination addressed by employment discrimination statutes are similar to, and sometimes overlap, constitutional claims that may be brought under the Equal Protection Clause. (103) The Supreme Court and lower courts may be particularly resistant to congressional interpretations of these precepts that depart from constitutional interpretation; other areas of statutory law without constitutional analogues may be less divisive. More generally, employment discrimination is often a sharply partisan issue, and in recent decades the Supreme Court's interpretations in this area have tended to be far more conservative than those of Congress. Thus, judges may use shadow precedents as something of a fig leaf for advancing their own policy preferences.

Nonetheless, even if not fully representative, it is important to understand and appreciate the considerable role that shadow precedents play in this area of the law and the challenge they pose to the premise of legislative supremacy. Hopefully, future research will help us understand whether shadow precedents play a similarly central role in other areas of statutory law.

Nor am I suggesting that every employment discrimination decision that is overridden lives on as a shadow precedent. I have chosen the examples that follow because in each case, courts have expressed substantial disagreement regarding the scope and significance of the specific override. They thus illustrate the extent to which doctrinal confusion may stem from a lack of analytic clarity regarding the statutory interpretation of overrides and further the possibility that courts, in following shadow precedents, may thwart congressional intent. Of course, each of the individual substantive areas of employment discrimination law addressed is itself a subject of extensive critical commentary, and I am not attempting to fully explore the practical and theoretical ramifications of each subject. That said, the analysis below may provide a fruitful new approach for understanding the individual specific substantive issues.

This Part begins with a brief overview of federal employment discrimination law and the circumstances that led to the 1991 Civil Rights Act. It then turns to specific examples of overrides to illustrate the significant role played by shadow precedents. The next Part offers a proposal for reducing reliance on shadow precedents and illustrates the proposal with a reconsideration of the examples introduced in this Part.

A. Federal Employment Discrimination Law and the 1991 Civil Rights Act

Federal employment discrimination law is governed by several different statutes. Title VII of the Civil Rights Act of 1964 (l04) is the broadest in scope. It prohibits discrimination on the basis of race, color, religion, sex, and national origin. (105) Title VII also explicitly prohibits retaliatory actions against individuals who seek to invoke their rights under the statute. (106) Three years after Title VII was enacted, Congress enacted the Age Discrimination in Employment Act (ADEA), (107) which prohibits discrimination on the basis of age. (108) The substantive provisions of the ADEA largely track the language in Title VII. In 1990, Congress enacted the Americans with Disabilities Act (ADA), (109) which includes a title prohibiting employment discrimination against persons with a qualifying disability. (110) The substantive provisions of the ADA also largely track the language of Title VII (with an additional substantive requirement that discrimination includes failure to make a reasonable accommodation for an individual's disability). (111) In addition to these relatively modern employment discrimination statutes, 42 U.S.C. [section] 1981, (112) a civil rights statute adopted shortly after the Civil War, is also interpreted to prohibit employment discrimination on the basis of race. (113) These various statutes, and the interactions between them, come into play in the statutory decisions discussed in this Part. (There are many other federal employment discrimination laws that I am not outlining here because they do not play a role in these cases.)

Employment discrimination cases make up a very large portion of the lower courts' dockets, (114) and they represent a sizable share of the Supreme Court's statutory docket. (115) It is an area of the law where sharply splintered decisions are common and where Congress frequently overrides Supreme Court interpretations. The 1991 Civil Rights Act (1991 CRA), which responded to twelve Supreme Court decisions, was a particularly striking example. Because it included many of the overrides discussed below, a summary of the context for the Act and the compromises that were made to get it enacted is helpful. (116)

In 1988, George H.W. Bush appointed Justice Anthony Kennedy to fill Justice Lewis Powell's seat on the Supreme Court. Justice Kennedy's appointment shifted the Court to the right, and in 1989, the Court announced a series of employment discrimination cases (as well as other civil rights cases) that sharply curtailed individual employees' rights. NAACP Director Benjamin Hooks compared the decisions to cross-burnings by the Ku Klux Klan and promised "'civil disobedience on a mass scale that has never been seen in this country before."' (117) Leaders in Congress quickly responded. James Jeffords, then a Republican Senator, characterized the decisions as a "wholesale retreat from civil rights and equal employment commitments made by this nation" and Democratic Senator Howard Metzenbaum similarly complained that they were a "stunning example of the Court's retreat from equal employment opportunity." (118)

Congress quickly began to work on an override bill. After long and rather torturous negotiations among the two houses of Congress and the administration, Congress passed the Civil Rights Act of 1990, but it was vetoed by President George H.W. Bush. Congress narrowly failed to overcome the veto, and in 1991, Congress again addressed the issue. The Democrats in the House introduced a bill that built on, but in some respects went further than, the compromise version that had passed both houses in 1990. This bill passed the Democrat-controlled House, but largely along partisan lines. Republicans, reportedly frustrated that compromise language had been abandoned, introduced alternative versions. In the Senate, the Republicans' bill was the basis for negotiations. After some significant modifications in the Senate and then in conference committee, the Senate bill was enacted. The 1991 CRA, like the 1990 Act, overrode several Supreme Court decisions but some of the language in the overrides was less expansive than it had been in the 1990 version. Congress also placed caps on new damages provisions and made several other changes designed to make the bill more palatable. George Bush signed the 1991 Act and it became law.

One change had particular significance for the override discussion. The 1990 bill included explicit provisions that would have made all of the overrides retroactive, generally back to the day before the Supreme Court issued the decision that was being overridden. (119) Thus, the Supreme Court's interpretations would never have had legal effect. These retroactivity provisions were controversial and they were not in the 1991 law. In a related change, the 1990 bill, and the version of the 1991 Act that originally passed the House, characterized the Supreme Court decisions as "cut[ting] back dramatically on the scope and effectiveness of civil rights protections," and stated that a primary purpose of the bill was to "respond to the Supreme Court's recent decisions by restoring the civil rights protections." (120) The final law, by contrast, took language instead from the Republican counterproposal. It characterized the law's purpose as "to respond to recent decisions of the Supreme Court by expanding the scope of relevant civil rights statutes in order to provide adequate protection to victims of discrimination." (121) The Supreme Court relied in part on this shift in language in determining that Congress had not clearly indicated an intention that the overrides apply retroactively. (122) This language, as discussed below, also (sometimes) has been deemed significant in determining the substantive contours of the overrides. (123)

Understanding the convoluted history of the 1990 and 1991 bills is important. In 1991, there was a detailed committee report for the House Democratic bill. (124) There was not a committee report for the Senate bill; however, the Republican sponsors developed "interpretive memoranda" that they placed in the Congressional Record and which are generally deemed significant expressions of their understanding of the meaning of various provisions. (125) Additionally, there were various House, Senate, and conference reports for the 1990 Act. (126) Thus, when considering this legislative history, it is essential to identify which bill commentary relates to, and whether there were substantive differences between, the language discussed in the report or memoranda and the language that was ultimately enacted. Notwithstanding all of these caveats, the legislative history, particularly when House and Senate sources agree, can provide helpful context for understanding likely congressional intent regarding the Act's provisions. (127)

B. Lorance v. AT&T Technologies

As described above in Part II.B, an override can be interpreted as a narrow superseding of a holding that does not discredit the rationales supporting that holding or the interpretation of the preexisting statutory language. This is what I have called creating a "shadow precedent." The majority opinion in Ledbetter v. Goodyear Tire & Rubber Co. (128) analyzing Congress' override of Lorance v. AT&T Technologies, Inc. is a good example of this phenomenon. The vigorous dissent shows how sharply divided the Supreme Court is on whether reliance on a shadow precedent is appropriate.

Lorance concerned a question of when the statute of limitations to challenge an employment decision begins to run. To preserve the possibility of filing a claim under Title VII, an employee must file a charge with the Equal Employment Opportunity Commission within 180 (or in some states, 300) days "after the alleged unlawful employment practice occurred." (129) Lorance was a challenge brought by a group of female employees to a seniority system. The system disadvantaged women and, although it was facially neutral, the employees alleged that it had been adopted for discriminatory reasons. (130) The system had been in place for several years but a round of layoffs, based on the seniority system, had occurred immediately prior to the plaintiffs filing their charge. (131)

The Court found that their charge was not timely. It relied on a pair of decisions (132) holding that a discriminatory act must be challenged at the point where it occurred, rather than when its effects are felt: Delaware State College v. Ricks, (133) which had held that when a college made an allegedly discriminatory decision to deny an employee tenure but granted the employee a final one year nonrenewable contract, the charging period ran from the denial of tenure rather than the termination, (134) and United Air Lines, Inc. v. Evans, (135) which had held that when an airline implemented a discriminatory requirement that female flight attendants resign upon marriage, the charging period ran from the date of an employee's discharge rather than the subsequent effect of this discharge on her seniority status when she was rehired several years later. (136) It distinguished a pay discrimination case, Bazemore v. Friday, (137) which had held that each new paycheck was a new unlawful event restarting the statute of limitations, (138) by characterizing the pay structure in Bazemore as "facially" discriminatory. (139)

Justice Marshall, joined by Justice Brennan and Justice Blackmun, dissented. They contended that Evans and Ricks were not controlling, arguing that unlike the system at issue in Evans, the seniority system challenged in Lorance was alleged to have been adopted with discriminatory intent, and that unlike the injury alleged in Ricks, the injury suffered by the plaintiffs was not an "inevitable consequence" of the initial adoption of the seniority system. (140) They contended that it was irrelevant whether a seniority system or pay system that was adopted with discriminatory intent was also facially discriminatory (141) and accordingly would have followed Bazemore in holding that each new application of such a discriminatory system was a newly chargeable event. (142) The dissent further argued that the majority's approach was impractical because any harm to the employees upon the implementation of the seniority system was entirely speculative. (143)

The Lorance decision was overridden in the 1991 CRA. The Act amended the general rule regarding time limits to add:
      For the purposes of this section, an unlawful employment practice
   occurs, with respect to a seniority system that has been adopted
   for an intentionally discriminatory purpose in violation of this
   title (whether or not that discriminatory purpose is apparent on
   the face of the seniority provision), when the seniority system is
   adopted, when an individual becomes subject to the seniority
   system, or when a person aggrieved is injured by the application of
   the seniority system or provision of that system. (144)


The text of the override tracked closely the holding in Lorance, but it also made clear that at least with respect to seniority systems there was no relevant distinction between facially and non-facially discriminatory employment systems.

The legislative history of the bill from both the Senate and the House suggested an awareness--and disapproval--of use of Lorance in other contexts. The Senate sponsors' memorandum explained that "[u]nfortunately, some lower courts have begun to apply the 'Lorance rationale' outside the context of seniority systems," such as a challenge to allegedly discriminatory promotion policies or under the ADEA. (145) Referring to the legislative language ultimately enacted (reproduced above), it stated that the "legislation should be interpreted as disapproving the extension of [Lorance] to contexts outside of seniority systems." (146) The House Committee Report, discussing a proposed version of the law that included the language ultimately enacted as well as broader language that was not adopted, was equally clear in disapproving application of Lorance to other situations. (147)

Despite its override, Lorance continues to be applied as a shadow precedent. The most prominent example of its continued application was the 2007 Supreme Court decision in Ledbetter concerning whether a claim of pay discrimination was timely. The plaintiff in the case, Lilly Ledbetter, had won at trial based on evidence that she had been given poor employment evaluations because of her sex and that her pay was therefore inappropriately--and substantially--lower than that of men at the company with similar experience and qualifications. (148) When she filed her charge, she was still receiving paychecks that she contended were discriminatorily low, but her last discriminatory performance review had occurred several years before. (149) The Eleventh Circuit set aside the jury verdict on the grounds that her charge had not been timely filed. (150) The Supreme Court, in a 5-4 decision, affirmed. (151)

In its decision, the Court relied heavily on Lorance, as well as the Ricks and Evans cases that had themselves been cited in the Lorance decision. Although the Court acknowledged (in a footnote) that the 1991 Civil Rights Act could be understood to have some relevance to its discussion of Lorance, (152) its rhetoric minimized the significance of the override. Indeed, rather than calling it an override at all, the Court merely observed that "[a]fter Lorance, Congress amended Title VII to cover the specific situation involved in that case." (153) It further opined that the override relates "only" to seniority systems, (154) and accordingly that the reasoning that underlaid the holding in Lorance was not called into question, noting that the purposes clause introducing the 1991 CRA characterized the amendments as "'expand[ing] the scope of relevant civil rights statutes.'" (155) The majority opinion accordingly relied upon the reasoning in Lorance and particularly Lorance's distinguishing of Bazemore on the grounds that the system challenged in Bazemore was facially discriminatory. (156)

Justice Ginsburg, writing for four Justices in dissent, vigorously disputed the majority's analysis of the significance of the override, characterizing it instead as having "superseded" the decision in Lorance and made it "no longer effective." (157) She contended that Congress "thus agreed with the dissenters in Lorance that 'the harsh reality of [that] decision,' was 'glaringly at odds with the purposes of Title VII'" (158) and cited to legislative history that she suggested demonstrates that the override was intended to end reliance on the rule announced in Lorance and to generalize the rule announced in Bazemore. (159)

Ledbetter is the most prominent example of Lorance as a shadow precedent, but it is far from the only example. Although the Supreme Court had not previously relied upon Lorance since the 1991 CRA was enacted, lower courts cite the precedent routinely (sometimes without acknowledging in any way that it was overridden) for general propositions regarding when a statute of limitations should be considered to start running. (160)

C. Price Waterhouse v. Hopkins

A shadow precedent can also be created when an override is understood to apply only to a single statute and a court, interpreting a "related statute," follows the discredited precedent. The situation arises when a court interprets a given statute, for example, Title VII, and Congress overrides that interpretation by amending the same statute, that is, just Title VII. In interpreting the meaning of similar statutory provisions in related statutes, do courts follow the discredited precedent or the statutory amendments? Often, they choose the former. This leads to the anomalous result that a court's interpretation of a statute no longer governs the statute that was in fact interpreted but nevertheless controls the interpretation of other statutes.

This subpart provides a detailed example concerning the override of Price Waterhouse v. Hopkins, but courts have similarly struggled with whether to apply other overrides in the 1991 CRA, such as the override of Wards Cove v. Atonio Packing Co. (161) (regarding the burden-shifting to be applied in disparate impact cases) and of Lorance, to related statutes such as the ADEA or the ADA. (162) A few commentators have explored some of the particular applications of such shadow precedents, including Price Waterhouse. (163) The analysis below builds on their work. It substantially broadens the conversation, however, by showing similarities between the interpretive conventions that lead to shadow precedents in "related statutes" and substantive shadow precedents. This helps identify larger questions posed by statutory interpretation of overrides generally.

In Price Waterhouse v. Hopkins, the Supreme Court addressed the liability standard for "mixed-motive" claims, that is, those in which the plaintiff alleges that a challenged decision was motivated both by discriminatory bias and legitimate nondiscriminatory factors. (164) Title VII defines an "unlawful employment practice" as any of a variety of discriminatory actions taken "because of [an] individual's race, color, religion, sex, or national origin." (165) In Price Waterhouse, a four-Justice plurality opinion written by Justice Brennan interpreted the "because of" standard as requiring that a plaintiff show that an illegitimate factor played a "motivating part in [the] employment decision." (166) By contrast, a concurrence by Justice O'Connor, which is generally considered to provide the holding for the case, (167) interpreted Title VII to require a plaintiff to show with "direct" evidence that an illegitimate factor played a "substantial" role in the decision. (168) Both the plurality and Justice O'Connor held that even if a plaintiff made the requisite showing, a defendant could escape liability by showing it would have taken the same action without considering the illegitimate criteria. (169)

The 1991 CRA overrode this portion of Price Waterhouse. Congress did so by adding two new subsections to Title VII. The first stated that unlawful discrimination could be established if a plaintiff showed that any of the prohibited criteria (race, sex, etc.) were "a motivating factor for any employment practice, even though other factors also motivated the practice." (170) This standard was similar to that announced by the plurality opinion in Price Waterhouse and definitely easier to satisfy than Justice O'Connor's substantial factor standard. The second new subsection replaced the affirmative defense on liability that both the plurality and Justice O'Connor interpreted Title VII to provide. A showing by a defendant that it would have taken the "same action in the absence of the impermissible motivating factor" no longer eliminated its liability, but only limited the remedies available to the plaintiff. (171)

The House Report emphasizes the importance the Committee ascribed to the override:
      If Title VII's ban on discrimination in employment is to be
   meaningful, victims of proven discrimination must be able to obtain
   relief, and perpetrators of discrimination must be held liable for
   their actions. Price Waterhouse jeopardizes that fundamental
   principle. (172)


The Senate sponsors' memorandum does not address the Price Waterhouse override but the language in the Senate bill was identical to that in the House bill. Notably, the House Report also demonstrated that the Committee was sensitive to, and concerned by, the "related statute" problem by specifying that "other laws modeled after Title VII [should] be interpreted consistently in a manner consistent with Title VII as amended by this Act," although the specific example it provided concerned disparate impact and the ADA. (173)

Notwithstanding Congress' clear renunciation of the mixed-motive standard announced in Price Waterhouse, and the Report language indicating that it undermined the "fundamental" principle that perpetrators of discrimination must be held liable for their actions, courts struggle with whether Price Waterhouse controls interpretation of the ADA and the ADEA, as well as retaliation claims under Title VII. In all three areas, there is mixed case law regarding applicability of the override, with the trend moving towards applying Price Waterhouse as a shadow precedent. (174)

One approach, almost uniformly followed with respect to the ADEA and retaliation claims under Title VII, and sometimes followed in analyzing the ADA, is to apply Price Waterhouse as a shadow precedent rather than the "motivating factor" language of the override. (175) Courts doing so observe that Congress failed to amend these other statutes or the retaliation provisions in Title VII when it overrode Price Waterhouse. They conclude that that omission was significant and then, applying the general rule that they will follow the Supreme Court's interpretation of Title VII in interpreting these statutes, they follow Justice O'Connor's concurrence in Price Waterhouse. The other approach, which seems to have been used primarily (and even there not consistently) in interpreting the ADA, is to interpret the override of Price Waterhouse as reinterpreting the meaning of "because of" discrimination under Title VII and then follow this reinterpretation in analyzing the analogous language under the ADA. (176) For example, the Second Circuit reasoned that "[a]lthough that amendment does not, by its terms, apply to violations of the ADA, nothing in either the language or purpose of either statute suggests that Congress intended different causation standards to apply to the different forms of discrimination," and therefore, since the ADA and Title VII have otherwise substantially identical language regarding causation, the 1991 amendment should apply. (177) But, emphasizing the confusion attendant with this issue, a more recent Second Circuit case suggests that Price Waterhouse applies to mixed-motive claims under the ADA (and fails to cite to this earlier decision and its analysis at all). (178)

The Third Circuit, while ultimately following Price Waterhouse in a retaliation claim, offers one of the clearest explications of the interpretive conundrum. (179) The court indicated that it was "given pause" by the fact that "we and other courts have generally borrowed from discrimination law in determining the burdens and order of proof in retaliation cases," and thus that one "could say that Congress knew of the practice of borrowing in retaliation cases, and presumed that courts would continue this practice after the 1991 Act." (180) This of course would lead to the opposite conclusion than the one ultimately reached by the court:
   Considering the question with this assumption in mind, Congress'
   failure to reference [section] 2000e-3 [the retaliation provisions]
   specifically in [section] 107 [the section of the 1991 CRA
   overriding Price Waterhouse] would not mean that [section] 107 does
   not apply in retaliation cases; rather, it would mean that Congress
   assumed that it was unnecessary for it to do so because courts
   would borrow the "motivating factor" language in deciding
   retaliation claims. (181)


The court further noted some legislative history that might be thought to support this interpretation. (182) Nonetheless, it ultimately concluded that it was insufficient to counter what it characterized as the "language of the statute" in the override being clearly limited to Title VII's discrimination provisions and thus followed Price Waterhouse as a shadow precedent instead. (183)

D. General Electric Co. v. Gilbert

The overrides of Lorance and Price Waterhouse may lend themselves to creating shadow precedents because their language was drafted relatively narrowly: the statutory language overriding Lorance only explicitly addressed seniority systems and the statutory language overriding Price Waterhouse only explicitly addressed Title VII. Although legislative history in both cases suggested that Congress expected the override to forestall reliance on the discredited precedent in other contexts, it is understandable that courts conclude that the rationales expressed in the precedents remain controlling outside the literal language of the override. Notably, however, courts continue to rely upon shadow precedents even when the Supreme Court itself has held that an override should be understood as a clear rejection of the Court's rationale as well as its holding.

In General Electric Co. v. Gilbert, (184) the Supreme Court considered whether a disability policy that provided benefits to employees for all short-term disabilities except pregnancy discriminated on the basis of sex. The Court, in a 6-3 opinion, followed the analysis in an earlier constitutional precedent, Geduldig v. Aillo, (185) to hold that it did not. The majority reasoned that the plan did not discriminate on the "basis" of "sex," in that "sex" was not the explicit distinguishing factor. Rather, there were two groups of potential recipients of disability benefits, "'pregnant women and nonpregnant persons,'" and while the "'first group is exclusively female, the second includes members of both sexes.'" (186) Since there was no class of disabilities for which men were covered and women were not, the Court found there was no sex discrimination; there simply was a certain kind of disability--pregnancy--which was "unique to women" and was not covered. (187)

There were two dissents. Justice Stevens rejected the Court's characterization of the classification, arguing that a rule that discriminates on the basis of pregnancy "[b]y definition ... discriminates on account of sex; for it is the capacity to become pregnant which primarily differentiates the female from the male." (188) Justice Brennan, joined by Justice Marshall, observed that while Geduldig might be read to establish conclusively that a "pregnancy classification standing alone cannot be said to fall into the category of classifications that rest explicitly on 'gender as such.' ... [I]t offends common sense to suggest that a classification revolving around pregnancy is not, at the minimum, strongly 'sex related'" and that the district court's finding that the employer had discriminatory motives in adopting the plan should have been sufficient to state a claim. (189) Justice Brennan also noted that the Court's decision "repudiate[d]" an applicable administrative guideline that had been promulgated by the Equal Employment Opportunity Commission as well as the "unanimous conclusion of all six Courts of Appeals that ha[d] addressed this question." (190)

Shortly after the decision in Gilbert, Congress easily passed the Pregnancy Discrimination Act (PDA) (191) to override it. The PDA added for the first time a definition of "sex" as used in Title VII, providing: "[t]he terms 'because of sex' or 'on the basis of sex' include, but are not limited to, because of or on the basis of pregnancy, childbirth, or related medical conditions...." (192) In hearings on the bill and statements on the floor, speakers repeatedly cited to the language in the dissenting opinions quoted above and expressed a desire to "return" the law to what many believed that it had meant prior to the decision in Gilbert. The House Report stated baldly: "It is the committee's view that the dissenting Justices correctly interpreted the Act." (193) The Committee also endorsed the EEOC's prior position as "rightly implement[ing] ... the 1964 act." (194) The Senate Report likewise cited to the EEOC guidelines and to lower federal court decisions, as well as to state court decisions interpreting state laws prohibiting sex discrimination, as support for the general proposition that pregnancy discrimination was sex discrimination; it then quoted passages from the dissenting opinions and indicated that they "correctly express[ed] both the principle and the meaning of Title VII." (195)

The override was intended not only to overturn the holding of Gilbert but also to reject the interpretive rationale applied by the Court in that case. In fact, the Supreme Court, ruling shortly after the PDA was enacted, analyzed the legislative history discussed above and held explicitly that Congress "not only overturned the specific holding in General Electric Co. v. Gilbert, but also rejected the test of discrimination employed by the Court in that case." (196) Accordingly, under Supreme Court precedent it should be clear that the override fully supersedes Gilbert: not only is its holding no longer valid but its reasoning is no longer valid either. Moreover, the Court recognized that Congress suggested that the Gilbert dissents properly interpreted Title VII's prohibition on discrimination because of sex. (197) Thus the enactment of the PDA should stand not only for the proposition that "pregnancy, childbirth, and related medical conditions" are discrimination "because of sex," but also as a repudiation of the Court's holding in Gilbert that a policy that discriminates based on a condition (in that case, pregnancy) unique to women is not a form of sex discrimination.

In fact, however, Gilbert continues to live on as a shadow precedent. Examples from two factual scenarios--claims relating to discrimination against women for breastfeeding and denial of prescription contraceptive benefits--illustrate this phenomenon. Each describes issues that clearly have some similarity to the policy addressed in Gilbert. Breastfeeding, like pregnancy, is a medical condition "unique to women." Likewise, although theoretically prescription contraceptives could be available to men and to women (and certain surgical contraceptives are available to men and women), currently prescription contraceptives are only available to women. Moreover, the risk posed by the denial of contraceptives--that is, an unwanted or unplanned pregnancy--is borne by women. It is not surprising that courts reference Gilbert, as well as the PDA, in deciding these cases; rather, the surprising thing is that they follow the repudiated reasoning in Gilbert.

In the breastfeeding context, courts typically first hold that the specific issue posed is not addressed within the language of the PDA (that is, that it is not discrimination based on "pregnancy, child birth, or related medical conditions"). Then, in determining whether it could nonetheless be considered "sex" discrimination, they hold that it could not by following the reasoning of the majority opinion in Gilbert. (198) Thus, for example, one district court rejected such a claim on the grounds that "[t]he drawing of distinctions among persons of one gender on the basis of criteria that are immaterial to the other ... is not the sort of behavior covered by Title VII. This was made clear more than twenty years ago in General Electric Co. v. Gilbert." (199) Similarly, another began by reasoning that there was "no significant difference between the situation in Gilbert and the case here" and that accordingly "under the principles set forth in Gilbert" the plaintiff could not succeed. (200) Suggesting therefore that "[i]f plaintiff has a cause of action, then it must be by virtue of some change in the law after Gilbert," (201) the court found that breastfeeding was not within the literal language of the PDA and thus that alleged discrimination based on breastfeeding could not state a claim under Title VII. (202) In other words, Gilbert is invoked as a shadow precedent.

Similar questions arise in considering whether health insurance benefit plans that disallow coverage for prescription contraceptives or all contraceptives violate Title VII. In re Union Pacific Railroad Employment Practices Litigation is the only federal circuit court decision on the matter. (203) The majority opinion in the case rejected plaintiffs' claim that denial of insurance coverage for contraceptives violates Title VII. Like the courts considering breastfeeding claims, the court read the language of the PDA narrowly and held that contraceptives are not "related" to pregnancy for purposes of analysis under the PDA. (204) It then did a "separate" sex discrimination analysis and held there was no sex discrimination on the grounds that Union Pacific's plan denied coverage for both men and women. (205) It deemed the fact that prescription coverage was currently available only for women, as well as the fact that women were the ones who bore the risk of becoming pregnant, irrelevant. (206) Other district courts have likewise found there to be no sex discrimination. (207)

But some judges faced with determining whether a denial of contraceptive coverage constitutes discrimination analogize to Gilbert not to argue that the result is controlled by the "narrow" language of the PDA but rather that it should be controlled by the Gilbert dissenters' broader understanding of "sex" under Title VII. Both a dissent in Union Pacific (208) and the decision in Erickson v. Bartell Drug Co., (209) the first federal decision issued on the contraceptive question, take this approach. The decision in Erickson is particularly clear on this point. The court observed that although the language of the PDA "was chosen in response to the factual situation presented in Gilbert," it contended that "enacting the PDA, Congress embraced the dissent's broader interpretation of Title VII." (210) The court therefore reasoned that although contraceptive coverage was not clearly included in the language of the PDA, denial of contraceptive coverage violated Title VII because by enacting the PDA Congress had made clear that "mere facial parity of coverage does not excuse or justify an exclusion which carves out benefits that are uniquely designed for women." (211) Additionally, the EEOC has issued an opinion letter that likewise follows the reasoning employed by the dissenters in Gilbert to conclude that denial of contraceptive coverage violates Title VII. (212)

The split described above regarding the applicability of the PDA and Gilbert to cases of breastfeeding and contraceptives helps lay out the contours of unsettled law. It demonstrates, as discussed in more detail in the next Part, that even when it is apparent that Congress, in enacting an override, repudiates not only the holding of a decision but also its reasoning, and even when the Supreme Court has confirmed that understanding clearly in an opinion, shadow precedents may continue to hold sway. On the other hand, the Union Pacific dissent and the Erickson decision also show how a clear understanding that an override rejects the rationale of a precedent as well as its specific holding can be understood to reinterpret the preexisting language of the statute.

E. A Different Approach: The "Nullification" of Patterson v. McLean Credit Union

The Supreme Court's recent interpretation, in CBOCS West, Inc. v. Humphries, (213) of the effect of an override of Patterson v. McLean Credit Union, (214) another decision overridden in the 1991 CRA, stands in sharp contrast to these shadow precedents. In Patterson, the Court held that [section] 1981, a statute enacted shortly after the Civil War which provides that "[a]ll persons ... shall have the same right ... to make and enforce contracts ... as is enjoyed by white citizens," (215) did not encompass postformation contract conditions such as harassment. (216) The Court justified its interpretation in part by arguing that a broader interpretation of [section] 1981 would "undermine the detailed and well-crafted procedures for conciliations and resolution of Title VII claims." (217)

Justice Brennan, joined by Justice Marshall, Justice Blackmun, and Justice Stevens (in the sections of the opinion discussing the harassment issue) dissented. (218) They argued that the legislative history of [section] 1981 made clear that it was intended to reach discriminatory working conditions and other postformation conduct. (219) They also argued that any overlap with Title VII was irrelevant and further that, in the putative interests of balancing Title VII and [section] 1981, the majority opinion restricted the availability of [section] 1981 as a remedy to victims of discrimination in nonemployment contexts to which Title VII does not extend. (220)

Patterson was overridden in the 1991 CRA. The law added a new subsection to [section] 1981, providing that "[f]or the purposes of this section, the term 'make and enforce contracts' includes the making, performance, modification, and termination of contracts, and the enjoyment of all benefits, privileges, terms, and conditions of the contractual relationship." (221) Both the Republican Senators' sponsor memo (222) and the House Report emphasized that this list was to be "illustrative" only, with the House Report further specifying that the prohibition "would include, but not be limited to, claims of harassment, discharge, demotion, promotion, transfer, retaliation, and hiring." (223)

In CBOCS, the Court considered whether [section] 1981, as amended by the Patterson override in the 1991 CRA, encompassed retaliation claims as well as racial discrimination claims. (224) Modern employment statutes, such as Title VII, the ADEA, and the ADA, prohibit retaliation in separate statutory language from the antidiscrimination prohibitions related to specific protected characteristics. Further, modern doctrinal understandings of retaliation emphasize that it is a different "kind" of discrimination based on conduct rather than status. (225) Although the statutory language of [section] 1981 seems to address only racial discrimination, the Court, in a 7-2 decision, held that retaliation claims could be brought under it. (226)

In so holding, the Court rejected CBOCS's argument that Congress' failure to include explicit antiretaliation language in its override of Patterson should be recognized as intent by Congress to exclude retaliation claims. (227) The Court claimed that its holding was required by stare decisis, (228) but it was, as Justice Thomas pointed out in dissent, a rather unusual invocation of the doctrine because the Court had never ruled specifically on whether [section] 1981 permitted retaliation claims. (229) The Court instead pointed to Sullivan v. Little Hunting Park, Inc., (230) a 1969 decision interpreting 42 U.S.C. [section] 1982 (a civil rights statute regarding property rights) to permit retaliation claims, and to Supreme Court decisions making clear that [section][section] 1981 and 1982 are generally interpreted in tandem. (231) Prior to Patterson, circuit courts had relied upon the decision in Sullivan to support determinations that [section] 1981 permitted retaliation claims. (232) In the period after Patterson and prior to the enactment of the 1991 CRA, circuit courts relied upon Patterson to overrule their earlier decisions permitting retaliation claims under [section] 1981. (233) The Court explained the effect of the override as follows: "[G]iven Sullivan and the new statutory language nullifying Patterson, there was no need for Congress to include explicit language about retaliation. After all, the 1991 amendments themselves make clear that Congress intended to supersede the result in Patterson and embrace pre-Patterson law. And pre-Patterson law included Sullivan." (234) In other words, if Congress were enacting a new statute in 1991 governing postformation discrimination in contracts, and if it intended to also prohibit retaliation related to complaints of postformation discrimination, it almost certainly would have had to specify retaliation in the statutory text. But if, instead, its 1991 amendment of [section] 1981 is perceived as "nullifying" Patterson and returning to pre-Patterson law, it can also return to the precedents interpreting that law as if Patterson had never been decided.

The Court bolstered this analysis with reference to legislative history of the 1990 and 1991 Acts indicating that Congress expected the Patterson override to reinstate retaliation claims. (235) But this approach is arguably inconsistent with the analysis in Ledbetter where the Court found significant the statutory purposes clause which introduces the Patterson override as well as the Lorance override--that characterized the overrides in the 1991 Act as "expanding" rather than "restoring" law. (236) Moreover, in deciding the retroactivity cases, the Court had also already held specifically with respect to [section] 1981 that Patterson established what [section] 1981 had always meant and why prior circuit court interpretations were incorrect. (237) In CBOCS, the Court ignores both of these decisions, holding instead that Congress could, in passing the override, simply reinstate all of these interpretations and implicitly broaden the scope of the understanding of the statute beyond the plain text of the override language. (238)

The resolution of CBOCS was not a foregone conclusion. Justice Thomas, joined by Justice Scalia, dissented, arguing that the plain text of the statute should control: [section] 1981 is a ban on racial discrimination and "[r]etaliation is not discrimination based on race." (239) Even more strikingly, Chief Judge Easterbrook, dissenting in the Seventh Circuit decision on the case, would have held also that [section] 1981 did not encompass retaliation claims. He justified this interpretation by relying on Patterson itself, arguing (based on the Supreme Court's holdings in the retroactivity context) that since "[l]egislation does not 'overrule' decisions of the Supreme Court," the rationale of Patterson, which he classified as a "holding" that interpretation of [section] 1981 should be done in such a manner as to minimize the overlap with Title VII, remains good law. (240) Applying this rule, he argued that [section] 1981 should be interpreted not to encompass retaliation claims because that would create incompatibility with Title VII in the absence of clear intent by Congress to create overlapping provisions. (241) In other words, Judge Easterbrook would have applied Patterson as a shadow precedent.

IV. REDUCING RELIANCE ON SHADOW PRECEDENTS

As described in detail in the previous Part, courts use ad hoc and inconsistent analysis in assessing the scope of overrides, and often shadow precedents continue to exert sway. This Part begins by discussing some of the costs associated with reliance on shadow precedents. It then discusses how Congress could do a "better" job in drafting overrides, but argues that placing the onus entirely on Congress is both unrealistic as a practical matter and inappropriate as an analytic matter. It advocates instead for rules of statutory interpretation that would largely end reliance on shadow precedents.

A. Problems with Reliance on Shadow Precedents

The most evident problem--or at least potential problem--with reliance on shadow precedents is that it can thwart congressional will. As noted in detail in Part I, overrides are presumed to ensure that Congress has ultimate authority over the shape of statutory law. In each example in Part III, there was relatively persuasive legislative history that suggested that Congress clearly disagreed with the rationales as well as the holdings announced in the various cases and that it intended, by enacting overrides, to end reliance on the cases. (Of course, this raises the question of why Congress did not make its disagreement clear within the statutory text, a question addressed in the following two subparts.) Nonetheless, courts, relying on shadow precedents, continue to apply the repudiated reasoning in these cases. Jeb Barnes' findings that judicial dissensus after overrides is more common in areas of the law that tend to be sharply partisan suggests that shadow precedents can be used by judges to flout congressional directives without obviously violating basic principles of the rule of law. (242)

A less obvious, perhaps, but equally significant result of reliance on shadow precedent is the failure to serve the objectives of precedent. Recall that reliance on precedent is a jurisprudential convention designed to further fairness, efficiency, and predictability in law. (243) Reliance on shadow precedents--particularly the ad hoc and unpredictable application currently in place fails to serve any of these objectives.

First consider fairness, the general principle that similar cases should be resolved similarly. By definition, the application of shadow precedents depends on a showing that a given factual scenario is, at least in some respects, relevantly similar to the overridden case (albeit at least arguably not addressed by the language of the override itself). That is, either the facts or the statutory structure must be deemed to be relevantly similar such that there is a reason to apply the shadow precedent at all. But since the primary holding of the case has been overridden, reliance on shadow precedents means by definition that "relevantly similar" cases are being treated differently, notwithstanding the absence of any affirmative indication by Congress that it intends those differences. Thus, for example, pregnancy is recognized as sex discrimination but the relevantly similar "unique" female characteristic of breastfeeding is not recognized as sex discrimination. To succeed in a mixed-motive claim under Title VII, a plaintiff must simply show that an illegitimate criterion was a "motivating" factor in the decision but to succeed in a relatively similar mixed-motive claim under the ADA or the ADEA, a plaintiff must show it was a "substantial" factor. Fairness and equality thus are not served.

Nor is the interest in efficiency served. The current ad hoc approach to addressing shadow precedents imposes significant costs on the judicial system and on litigants. Courts expend considerable time and energy analyzing overrides and shadow precedents. Indeed, under the related statute phenomenon, they must develop an entirely divorced parallel path of case law, tracing out the meaning of Wards Cove, Price Waterhouse, and other discredited precedents that no longer control Title VII but are deemed to control the interpretation of related statutes.

Likewise, the interest in predictability is undermined. Shadow precedents are applied in an ad hoc manner, meaning that individuals cannot reasonably guess whether and how they will be applied. Any kind of interpretive clarity--for example, either a rule that overrides will always be read narrowly and leave in place any aspect of a precedent not clearly discredited or a rule, such as I advocate below, that creates a rebuttable presumption against such applications--could help with predictability. Significantly, however, the former rule would not address equality concerns (in that it would almost certainly result in similar cases being treated differently) or efficiency concerns, particularly with respect to related statutes. Nor, at least arguably, would it adequately protect the interest in Congress' being able to effectuate its will, for reasons addressed in the next subpart.

B. Potential Congressional Responses--and Their Limitations

One approach to address the problem of shadow precedents (assuming one agrees they are a problem) would put the onus on Congress by adopting a rule that overrides would always be interpreted narrowly. This appears at first an attractive solution: it would ensure that Congress, rather than the courts, makes any necessary judgment calls and that it does so in statutory text that satisfies the Constitution's bicameralism and presentment requirements. In other words, such a rule could be construed as democracy-enhancing, similarly to how theorists claim that a clear policy of absolute refusal to rely upon legislative history (244) or adherence to statutory stare decisis (245) would be democracy-enhancing by "forcing" Congress to articulate precisely the extent to which it intends to supersede prior judicial interpretations.

Certainly, Congress would do well to consider the particular interpretive challenges posed by overrides when drafting them. To the extent that Congress intends to supersede fully a given judicial interpretation, it could and should draft in such a way as to make reliance on shadow precedents less likely. Most importantly, Congress should draft substantive statutory language broadly enough to respond to the primary rationale expressed in the case, not merely the factual application of that rationale. Congress could also state in a purposes clause that it disagrees with the court's interpretation or reasoning as expressed in a specific precedent or precedents, even if it were unwieldy to put such language in the substantive statutory text. (246) Likewise, drafting an override as a definitional amendment to the existing language helps signal that it is intended to change the understanding of the preexisting language. Making clear that any such definitions are illustrative rather than exhaustive can also help. Moreover, Congress should also be careful that introductory purposes and findings clauses that characterize an override as "expanding" the scope of law appropriately capture its understanding of the legislation, recognizing that the Supreme Court has (sometimes) pointed to such clauses as indicative of the substantive scope of an override as well as questions of whether it is retroactive or prospective. (247)

More generally, Congress should draft overrides, indeed all new legislation, with awareness of the limited significance that may be ascribed to statements in legislative history. For in part, the story of the overrides related in Part III is a story of the larger struggle in contemporary jurisprudence over the role that legislative history should play. (248) In many of the examples above, language in the committee reports and the sponsors' interpretive memoranda suggested Congress strongly disagreed with the rationales employed in the precedents it was overriding and that it expected that the overrides would be understood as making clear that the precedents should not be applied in other contexts. Reliance on shadow precedents ignores these signals in legislative history by focusing instead on the (ostensible) narrowness of the statutory text itself. (249) Although at the time these particular overrides were enacted (1978 and 1991) Congress might reasonably have expected that courts would consult this legislative history, contemporary Congresses certainly should not make such assumptions.

Finally, there is always the fall-back solution that if Congress disagrees with a narrow interpretation of an override, it may override that interpretation as well, although the same concerns may arise with respect to the new override. (250)

But there are clear limitations to simply expecting that if Congress could draft "more clearly," the problem of shadow precedents would go away. Let's begin by considering the "related statute" issue. As a practical matter, Congress is unlikely to be able to fully forestall the problem. As Jamie Prenkert observes, it would be exceptionally difficult for Congress "to canvass the entire statutory landscape for potential statutes to which the Court might extend the interpretation." (251) Title VII, for example, is regularly used as a model in interpreting not only other employment and labor laws, but also, for example, housing discrimination (252) and discrimination in education. (253) If Congress amended the ADEA and the ADA to import the mixed-motive standard into those laws but did not amend a housing law, would it be deemed under the principle of expressio unius to have consciously excluded fair housing law from the import of the override? How is Congress to know each and every statute that might be deemed "related"? And even if it did, is it really practical to suggest that Congress needs to amend them all to override an interpretation in a single, primary statute?

This concern helps identify an analogous risk in suggesting that Congress should likewise ensure that it drafts broadly enough to make clear that it disagrees with the rationale as well as the specific holding of a case to end the problem of substantive shadow precedents. In the abstract, this seems reasonable. In practice, it would often prove difficult. Frequently, there are multiple, sometimes conflicting, rationales offered to support a given holding. This is all the more true in an override responding to a circuit split rather than a Supreme Court decision. Addressing all of the proffered rationales would lead to excessively cluttered and confusing statutes. A choice not to address them all would lead to the same expressio unius problem. Would a clear repudiation of the primary rationale that failed to address a subsidiary rationale be interpreted as an intentional adoption of that subsidiary rationale?

Even if Congress could properly identify any and all potential statutes and rationales (an unlikely proposition at best), efforts to do so would use up congressional time and energy that would otherwise be spent on other political objectives. Thus, any such rule would impose significant costs on Congress. This would either take time away from other legislation that Congress might otherwise address, or simply mean that Congress is far less willing to pass overrides, even when it disagreed with judicial interpretations. (254) It would also almost certainly provide more points of controversy in the debate over an override, even if ultimately a majority of Congress agrees that the rationales employed by the court should be superseded. This would in itself have the effect of making overrides a less available or less effective "check" on judicial lawmaking. Moreover, if this were a clear interpretive rule rather than the current ad hoc approach, then a decision by Congress to address only the most significant rationales or related statutes would be deemed an even more telling indication that it purposefully excluded others, although the reality still might be that "narrow" drafting was simply the result of inattention to possible applications or priority setting.

Beyond these practical concerns, there is a deeper flaw in this approach because it accepts, unquestioningly, that only the statutory language of the override can be considered significant. This raises the question of whether and how Congress could draft an override so as to be understood to change the interpretation of the preexisting statutory text. What does Congress say when it thinks that the Court misinterpreted the statute and that the words that were already there should mean something different? For example, when Gilbert was decided and when the PDA was debated in Congress, many lower federal courts and state courts, the EEOC, and several Supreme Court Justices thought that the preexisting language of Title VII, prohibiting discrimination "because of sex," should have been interpreted as sufficient to address pregnancy discrimination. (255) In passing the PDA, Congress clearly indicated that it agreed. (256) The override itself thus should be understood to change the interpretation of what discrimination "because of sex" means, such that the separate mention of pregnancy itself is, in some sense, redundant. Relying on the words of the PDA ("pregnancy, childbirth, or related medical conditions") (257) to exclude other applications that might also be deemed already "adequately" addressed by a general prohibition on the basis of sex (as reinterpreted by Congress) suggests that it is impossible for Congress to actually change the interpretation of the preexisting statutory language. But the whole premise of legislative supremacy makes clear that Congress must have this power. Accordingly, interpretive principles should reflect this.

C. Proposed Interpretive Reforms

Rather than simply relying on Congress to draft specific overrides "more clearly," courts interpreting overrides should do so in a manner that is more respectful of the significance of a congressional override. I propose two interrelated interpretive conventions that would achieve this objective: (1) a rebuttable presumption that Congress, in overriding a nonconstitutional judicial decision interpreting a statute, rejects the court's interpretation of the preexisting statutory language and thus that "fresh" statutory analysis is required; and (2) a rule that those aspects of the overridden precedent are no longer binding on lower courts. These rules of statutory interpretation could be implemented through Supreme Court decisions that provide interpretive instructions to lower courts. Alternatively, Congress could enact a statute that set forth interpretive principles to be used by courts when considering overrides. (258) Under this proposed approach, when considering a future dispute not clearly covered by the statutory language of an override, courts would consider the extent to which the language of the override (and, if they are willing to consider legislative history, the legislative history) can help determine whether Congress intended to carve out a narrow exception to the preexisting interpretation of the statute or rather to reinterpret the existing statutory language in some manner. By creating a rebuttable presumption that it intended the latter rather than the former, Congress would largely end reliance on shadow precedents.

1. "Fresh" Statutory Analysis with a Rebuttable Presumption That the Prior Judicial Interpretation Is Superseded

The first aspect of the proposal--that courts should do "fresh" analysis of the statutory text that triggered the override with a rebuttable presumption that the court's prior interpretation is superseded--builds on the discussion of congressional drafting in the previous subpart that showed it often can be difficult for Congress to comprehensively address every potential application of an overridden precedent. This suggests that a "narrow" override can be characterized two very different ways. The first, which gives rise to reliance on shadow precedents, stems from the general principle of expressio unius. This approach assumes that by enacting a "narrow" override, Congress intentionally excluded everything else that might have legitimately been addressed within such an override. Thus, courts reason, Congress intended, or at least accepted, that these other scenarios would continue to be controlled by the precedent that was "partially" overridden. This is the analysis employed by courts that follow Lorance or Price Waterhouse as shadow precedents. (259) The doctrine of expressio unius makes sense when used to analyze a statute as initially enacted; it has considerably less merit when the specific references are necessary to override a court decision and when the preexisting more general language might have been considered sufficient to address the situation.

The second--quite different--characterization of a "narrow" override is that Congress affirmatively acted on one issue (for example, seniority systems and the standard to be applied to mixed-motive claims under Title VII) and was "silent" on the others. Legislative silence has long been deemed an unreliable indication of acquiescence to court interpretations. (260) Although courts sometimes deem it significant, common critiques of the practice include that Congress may not have noticed the relevant decision, that Congress may not agree with the interpretation but may nonetheless have other higher priorities, and that the reaction or lack thereof of the sitting Congress is irrelevant when considering whether an interpretation properly implements the will of the earlier enacting Congress. (261)

With an override, obviously, Congress does know about the relevant decision. But Congress may not have anticipated or expected that a given precedent would be applied to a different statute or a different kind of factual scenario. It might, as the Third Circuit noted in its analysis of the Price Waterhouse question, (262) rather have assumed that the override would be sufficient to make clear that the precedent should no longer have persuasive significance or be applied in other contexts. Adoption of any override--a clear indication that Congress disagrees with an interpretation in at least some contexts, most notably the factual situation that the precedent actually addresses--should certainly indicate no greater approval of a court's interpretation or interpretive rationales than the failure to enact anything at all. Thus, a helpful reform would simply be to characterize applications of an overridden precedent that are not explicitly addressed by the statutory override language as congressional "silence" rather than acquiescence.

However, given that Congress, in enacting an override, expresses clear disapproval of the primary holding of a case, and recognizing the particular drafting challenges outlined in the previous subpart, I propose going further to shift the understanding of congressional "silence" in the specific context of overrides from a rebuttable presumption of approval or acquiescence to a rebuttable presumption that Congress disapproves of that judicial interpretation (even if a potential application of the interpretation or rationale is not clearly addressed in the override text). (263) Like other conventions of statutory interpretation, this would provide courts a general rule of thumb that would aid in properly assessing congressional intent. My point is not that Congress, in enacting an override, always intends to fully repudiate the prior judicial interpretation but that, on balance, it is far more likely to intend to do so than not. A presumption against relying on overridden precedent would require courts to do "fresh" statutory analysis, not only of the language of the override but also of the preexisting statutory language. Courts would consider other reasonable interpretations of that language and the extent to which the override suggests such other interpretations (particularly any interpretations urged by a dissent in the overridden case) are proper. (264) This proposal would thus have the salutary effect of refocusing attention on the statutory language itself, rather than the judicial gloss on the statutory language. That is, even if Congress has not expressed a specific "intent" regarding how a statute should apply in a given situation, courts would resolve the matter by considering the preexisting statutory language and the override, rather than mechanically applying a judicial interpretation that Congress has already rejected in its original context.

Some limitations on this proposal are important to note. First, this proposed rule would only apply to nonconstitutional rulings. To the extent that a court either struck down or narrowly interpreted a statute on constitutional grounds, the court's constitutional analysis would continue to be applied because it is the courts, rather than Congress, that have ultimate authority for constitutional interpretation. Second, it would only apply to aspects of the statutory interpretation that are closely related to the override itself. Aspects of the prior precedent that address other issues in the case would retain their precedential significance. While undoubtedly this would occasionally call for some difficult line-drawing, the examples discussed in Part III suggest that in most cases, this rule would be reasonably easy to apply. Third, this proposed rule would coexist with other standard approaches to statutory interpretation. Thus, for example, if applied in the criminal context, its effects would likely often be tempered by the rule of lenity, which prescribes that ambiguities in criminal statutes be construed narrowly. (265)

Of course, the proposed rule would not preclude Congress from consciously drafting a narrow override. If Congress really intends to generally endorse a court's interpretation of a statute while creating a narrow exception, it just needs to draft statutory language that makes that clear. The surest way to do so would be to partially codify the holding and partially override it. (266) Likewise, if Congress really intends to override a standard with respect to one statute but not other statutes, it could codify a precedent in the "other" statute while overriding the precedent in the "primary" statute. The difference under the rule I propose is that failure to do so would not be deemed either dispositive or presumptive acquiescence. Rather, there would be a presumption against relying on the prior judicial interpretation.

Even if Congress did not explicitly codify a portion of the prior ruling, a court could revisit the interpretation of the preexisting language and conclude that the presumption against relying on the prior judicial interpretation is, indeed, rebutted. Many factors could be important, and, significantly, the viability of this approach would not turn on the willingness of a judge or justice to consult legislative history. Rather, analysis would be done pursuant to whatever interpretive methodology the relevant judge or justice employs. Thus, a textualist would focus on the preexisting language and the override and the implications that could reasonably be drawn from it. If there are two reasonable interpretations of the preexisting language and Congress drafts a relatively broad override that seems to signal generally its disagreement with the prior judicial interpretation and agreement with an alternative plausible interpretation, the presumption would be unlikely to be rebutted. By contrast, if the override was drafted extremely narrowly (and certainly if it were set off by textual signals, such as "provided however," that signal an exception to a general rule), a court might well reasonably conclude that the presumption was rebutted and that the prior judicial interpretation should control in any application not squarely addressed by the text of the override.

Context could also be important. If the overridden decision itself was a departure from relatively uniform prior judicial or agency interpretations of the statutory language, Congress might reasonably be understood as intending to return to such earlier interpretations. Likewise, if the overridden decision was an issue of first impression and Congress immediately signaled its disagreement with the judicial interpretation by enacting an override, continued reliance on the judicial interpretation would generally be inappropriate. By contrast, if the overridden decision conformed with a longstanding interpretation of the statute, and Congress drafted a narrow override, it might be more reasonable to consider the presumption rebutted and reaffirm the prior judicial interpretation in any context not directly addressed by the override. The area of statutory law being interpreted could likewise have relevance. Some statutes, such as the Sherman Antitrust Act, (267) are typically referred to as "common law" statutes; the statutory language is broad and Congress is generally presumed to have acquiesced to having courts develop standards under the statute in a common law fashion. A court might properly conclude that reliance on a shadow precedent was more reasonably warranted in such a context. And finally, a court willing to consider legislative history could look for signals in the legislative history regarding congressional intent. (268)

Despite the possibility that a court could reaffirm the prior judicial interpretation, this proposal obviously makes such a result less likely. Thus, a potential objection might be that it would have the effect of overriding "more" than Congress could get a clear majority to approve in statutory text. (269) This is a risk--it is in some sense the inverse of the common risk recognized in relying on congressional "silence" as acquiescence. In this case, however, it would be tempered by the fact that courts would still be applying the language of the statute, as amended by the override. Narrow override language still would invite narrow interpretations. Creating a presumption of disapproval does not absolve Congress of a need to affirmatively legislate. Statutory language is still the only way that Congress can ensure that courts will interpret the preexisting language differently.

The opposite concern is that a rebuttable presumption is unlikely to make any difference at all. Courts could do "fresh" interpretation and decide that the same interpretation should control even if there were no strong factors to suggest that it was warranted. Undoubtedly, this would occur at times. However, a rule that required courts to at least go through the motions of fresh analysis--and to articulate some justification (beyond presumed congressional acquiescence) that could plausibly rebut a presumption against reliance on the prior interpretation--would certainly help ensure that overrides can play their expected role in ensuring Congress has the ultimate authority to shape statutory law.

Suggesting not only a rebuttable presumption of disapproval but rather a clear rule that an override completely supersedes the prior interpretation would have some pros and cons. To the extent that reliance on shadow precedents is conscious judicial flouting of congressional will, this approach would have merit in helping ensure legislative supremacy. It would also enhance predictability in some respects. However, pragmatically speaking, it is far less likely that the Supreme Court would ever implement such an interpretive principle, or that it would abide by it if it were to do so. (270) Additionally, it would greatly increase the risk that a relatively narrow override would have a significantly broader effect than a majority in Congress would have agreed to. Moreover, suggesting that courts, after doing "fresh" statutory analysis, could potentially reaffirm interpretations or rationales in prior precedents has the benefit of serving as a "counter-check" against magnifying the scope of a "special interest" override. Accordingly, a rebuttable presumption appears to be an appropriate way of balancing these various concerns.

2. Overridden Precedents Should Not Bind Lower Courts

Second, I propose that enactment of an override should be interpreted as sufficient to supersede relevant aspects of the overridden precedent such that it is no longer binding authority on lower courts. In other words, the interpretation of the preexisting statutory language that triggered the override would no longer carry precedential weight. This second step of the proposal is necessary to permit full implementation of the first step, but it has an independent value. Under the principle of vertical precedent, lower courts cannot legitimately ignore precedent from a higher court that they deem on point. Thus, as the examples in Part III make clear, overrides often place lower courts in a particularly difficult position. On the one hand, they know that Congress has superseded the primary holding of a case. On the other hand, they know that an interpretive rationale expressed in the precedent, which but for the override would clearly be binding, is relevant to a given factual situation. If there is any ambiguity regarding the extent to which an override discredits the rationales of the prior precedent, or the extent to which it should or should not be applied to a "related statute," lower courts may believe they are required to follow the precedent. Or, to put a more instrumental spin on it, if a lower court's policy preference is more closely aligned with the judicial interpretation than with the congressional action that overrode it, the court may choose to interpret the override narrowly and hold that it is, therefore, bound by the prior precedent. Moreover, whatever their policy predilections, lower courts may naturally err on the side of following precedent rather than an override because they operate within a judicial hierarchy and are far more likely to be concerned about being reversed by higher courts than overridden by Congress.

Making clear that interpretation of the statutes as expressed in overridden precedents are not binding authority (271) gives lower courts the freedom to do the analysis suggested in step one of this proposal. It is important to recognize, however, that this proposal is actually relatively narrow. Lower courts would still be constrained by the statutory text itself. The only difference would be that a prior judicial interpretation of that text would no longer be binding. Lower courts would do fresh analysis of the preexisting statutory language as well as of the override so that they could appropriately determine the extent to which the override should be understood as a rejection of the prior judicial interpretation. And, of course, decisions by a lower court to disregard an overridden precedent would be subject to appeal, thus providing an additional check against unwarranted judicial activism.

The Supreme Court, which has the power to overrule its own precedents, is not faced with the same competing signals. Indeed, the proposed rule in some sense simply permits lower courts to apply, in the limited context of statutory overrides, the Supreme Court doctrine that permits overruling of statutory decisions due to significant "intervening development of the law" from court decisions or Congress. (272) Lower courts would not officially "overrule" higher court precedents, but they would be able to interpret the override as having done so.

An alternative proposal would simply counsel lower courts to "signal" their discomfort with application of an overridden precedent but nonetheless abide by it up to and until the time when the Supreme Court officially announces the extent to which the override should be interpreted as overruling the Court's prior interpretation. The hope would be that flagging the issue would make it more likely that the Supreme Court would accept a case raising the issue. This approach would fit more comfortably with the standard rule of precedent. But it would also have real disadvantages. Some issues would never make it to the Supreme Court; others could be decided "wrongly" for several years while waiting for a Supreme Court pronouncement. Moreover, as the breastfeeding cases make clear, even when the Supreme Court has definitively opined that an override clearly rejects the rationale of a prior precedent, lower courts still rely upon the discredited precedent. If overrides are to have their intended effect, lower courts need to be assured that they may properly disregard an overridden precedent.

Finally, it is appropriate to consider what role agency regulations or guidance might play in the interpretation of overrides. Agencies obviously could help resolve the relationship between an override and a precedent; they are perhaps particularly well-suited to address the related statute question (so long as, as in the employment discrimination context, a single agency has jurisdiction over the various statutes most likely to be interpreted together). Congress could invite or require an explicit agency role in implementing an override. An agency then could explore the extent to which enactment of an override should be understood as re-interpreting the preexisting statutory language such that reliance on a shadow precedent is unwarranted, notwithstanding Congress's failure to explicitly address some particular application of the precedent. However, agency interpretations would not be an adequate substitute for the new judicial interpretive conventions I propose. Courts are generally required to defer to agency interpretations when Congress has "not directly addressed the precise question at issue" and when the agency interpretation is "reasonable." (273) Thus, as a threshold matter, before a court would be likely to defer to an agency interpretation of an override that advocated ending reliance on a shadow precedent, a court would need to at least determine that the precedent was no longer binding and that Congress' "failure" to address a particular application of a rationale expressed in a given case was "silence" rather than presumed acquiescence.

3. Sample Applications

How would this proposal work in practice? And how much would it change current outcomes? Consider the examples discussed in Part III.

The easiest is the application of the rule in the context of related statutes. A court would do fresh statutory analysis of the meaning of "discriminate because of" in the ADEA, the ADA, or the retaliation provisions of Title VII, understanding that the mixed-motive analysis in Justice O'Connor's concurrence in Price Waterhouse, which, as discussed above, provided the holding for the case, was presumptively rejected. Of course, however, the new statutory provisions of Title VII would not be applicable in these other contexts either since, by their terms, they apply only to the particular antidiscrimination provisions of Title VII. However, courts interpreting one of those other statutes would do so with the knowledge that, in Title VII, Congress definitively stated that showing a prohibited factor plays a "motivating" role in a decision is enough to establish unlawful discrimination. (274) Recall also that four Justices in Price Waterhouse, prior to the 1991 CRA's "clarifying" language, already understood this as the appropriate interpretation of "because of." (275) Thus, courts could legitimately interpret "discriminate because of" in these other statutes in accordance with this new expression by Congress. In other words, the override would be understood as reinterpreting the preexisting language and that "reinterpretation" could be borrowed by courts interpreting similar language in other statutes.

Second, consider the question of contraceptives and breastfeeding coverage under Title VII as amended by the PDA. Under the rule, courts would do fresh statutory analysis of the meaning of discrimination "because of sex" in Title VII. The analysis in Gilbert would no longer be binding, and there would be a rebuttable presumption that Congress had disagreed with the judicial interpretation of the language in that case. Additionally, there was clear congressional signaling that the rationale adopted by the dissenters--that is, that discrimination based on a factor unique to one sex is sex discrimination--is correct. Thus, a court would probably conclude that the enactment of the PDA should be understood as reinterpreting the preexisting statutory text, changing the meaning of what "because of sex" means to include not only discrimination based on "pregnancy, childbirth, or related medical conditions" but also any other discrimination based on a factor unique to one sex (that is, a factor relevantly similar to pregnancy and childbirth). Under this analysis, discrimination against a woman because she is breastfeeding would be recognized as sex discrimination. (276) A denial of contraceptive coverage would be less clear. If understood in terms of its ultimate effect, potential pregnancy, it would be prohibited as sex discrimination. If, on the other hand, understood in terms of coverage denied both men and women, it would be arguably an open question. But, significantly, courts considering the question would consider the analysis in Gilbert as no longer binding on lower courts and further as presumptively rejected.

Third, consider the issue posed in Ledbetter. when does the statute of limitations begin to run in a pay discrimination case challenging ongoing lower pay based on past discriminatory decisions? The first step in an answer is simple. A court would do fresh statutory analysis of the provisions regarding timely filing under Title VII with the understanding that the analysis in Lorance is presumptively rejected. A court would also take note that Congress had specifically made clear that a seniority system may be challenged whenever a person is injured by it, and further that, again with respect to intentionally discriminatory seniority systems, Congress had clearly indicated that it was irrelevant whether the challenged system was facially discriminatory or not.

It would be more difficult to determine whether Congress, in mentioning only seniority plans in its override, intended seniority plans to be an exception to a general rule, and what in fact the "general rule" is. Courts still would need to consider two arguably competing lines of precedent, neither of which is clearly discredited by the override. (277) On the one hand are Evans and Ricks, each of which concerns discrete acts (a termination which later affected application of a seniority system and an indication of a future termination) with later consequences. These cases make clear that claims regarding such acts must be challenged when they occur, even if the more serious effects are felt later. On the other hand are the pay discrimination case Bazemore and a more recent decision, National Railroad Passenger Corp. v. Morgan, (278) concerning continuing violations in a hostile work environment, which establish that at least certain kinds of cumulative injuries may be challenged long after they begin, so long as some discrete event occurs within the charging period. Moreover, Bazemore holds that under a facially discriminatory pay system, each new paycheck is a new chargeable event. (279)

Lorance and its override, as well as the issue posed in Ledbetter, sit somewhere in the middle of the spectrum established by these precedents. Lilly Ledbetter was challenging prior discriminatory acts with later consequences. In that respect, her claim was like those brought in Evans and Ricks. But unlike the plaintiffs in Evans and Ricks, Ledbetter was challenging pay discrimination, an injury that accumulates over time and that might not be readily apparent at first. In these respects, her claim was like those brought in Bazemore and Morgan, as well as the statutory amendment made by Congress to override Lorance concerning seniority systems. If the reasoning in Lorance was presumptively rejected, would the Supreme Court still have followed Ricks and Evans rather than Bazemore and Morgan? Perhaps. But, tellingly, it is language from Lorance--that "facially nondiscriminatory and neutrally applied" systems do not perpetuate a charging period--rather than the other opinions that the majority opinion in Ledbetter relied upon to distinguish Bazemore. (280)

4. Scope of the Proposal

I have illustrated the problem of shadow precedents with a series of examples from employment discrimination, and I have noted that shadow precedents may be more likely to arise in areas of the law, like employment discrimination, that are subject to partisan divides. I have also explained that I chose employment discrimination as a focus in part because it is an area of the law where overrides are common. (281) There are potential risks in suggesting a general interpretive convention applicable to any area of statutory law based on examples from a single area of the law, particularly one which may not be representative. There are two--in some sense opposite--concerns that should be addressed. First is a concern that reliance on shadow precedents may be unusual; if so, why should a general rule be developed? Second is the converse concern that reliance on shadow precedents may be relatively common; if so, are there significant factors that are not necessarily evident in the employment discrimination context that should be considered crafting a rule to address the phenomenon?

Further research is appropriate before definitively answering these questions. Nonetheless, my preliminary conclusion is that these concerns are neither fatal to the proposal nor so significant that the interpretive convention I suggest should only apply in employment discrimination cases. The first concern is readily addressed. If reliance on shadow precedents is truly unusual outside the employment discrimination context, then, notwithstanding the fact that the rule I propose is a general one, it would have little significance in other contexts for the simple reason that courts already abide by it. That is, overrides always pose some interpretive challenges. If, in other contexts, courts conceive the override as effectively superseding the interpretative rationales expressed in the precedent case such that they do not rely on shadow precedents, then the rule I propose would merely bring employment discrimination in line with interpretation of overrides in other areas of the law. (282)

The second concern--that is, whether there may be other contexts where reliance on shadow precedents is more likely warranted--is harder to answer in the abstract. Although this is certainly possible (and thus it may be that the interpretive convention I suggest is less necessary in some other areas of the law), it seems relatively unobjectionable to suggest that it is always at least worth considering carefully whether Congress, in enacting an override, intended to end reliance on the prior judicial interpretation. As discussed more fully above, the rebuttable presumption I propose is not a hard rule that the interpretation in an overridden precedent is clearly rejected. Rather, courts could (and should) consider whether factors suggest that reliance on the prior judicial interpretation is merited. If so, courts could, under the interpretive convention I propose, reaffirm the prior interpretation.

That said, a rebuttable presumption makes it somewhat less likely that the previous interpretation would control, and this could increase, in some respects, unpredictability associated with overrides. This might be a more serious cost in areas of the law, such as statutes regulating commercial transactions, where predictability and efficiency are arguably even more important than in the employment discrimination or comparable contexts. It may also be that in areas of the law that are typically less subject to partisan divide, there is less reason to be concerned that courts could consciously seek to thwart congressional will through their interpretation of overrides. (283) One way to address such concerns would be to advocate that the interpretive conventions I propose only apply in employment cases (or perhaps other antidiscrimination contexts).

However, there would be costs to suggesting an interpretive reform that was limited to the employment discrimination context. As a general matter, metaprinciples of statutory interpretation, such as the rule I propose here, are not subject specific. This has the clear benefit that neither Congress nor the courts needs to be concerned about precisely how a given statute would be categorized when considering how it should be interpreted. (284) Indeed, limiting a proposal to employment discrimination would lead to awkward questions with respect to statutes that bridge several subject matters. For example, the recent amendments to the Americans with Disabilities Act amend not only provisions in the employment title but also definitional provisions that govern the other titles as well; if an employment discrimination specific rule were adopted, would interpretation of the scope of the override depend on which title it was applied to? Suggesting that the rule would apply to a broader category of statutes (for example, any antidiscrimination law) would simply push line-drawing questions a little further out. Additionally, as a practical matter, such a circumscribed reach would make it less likely that any given court would adopt such a rule, since the political valence of its effects would likely be more evident and more uniform if it were limited to a particular area of the law. (285) Since, as noted above, I believe the proposed rule is flexible enough to address competing concerns, my preliminary conclusion is that a general rule of statutory interpretation that discourages reliance on shadow precedents is warranted.

D. Benefits of the Proposed Interpretive Reforms

Reducing reliance on shadow precedents would better accord with principles of legislative supremacy. As noted, overrides are the ultimate expression of congressional disagreement with a court's interpretation. Reliance on a shadow precedent privileges the application of a judicial interpretation that has clearly been repudiated by Congress over the expression, admittedly sometimes obscure, of congressional intent in an override. It has the counterintuitive effect of making a partial or narrow override of a precedent serve as a stronger indication of (supposed) congressional acquiescence in the interpretive rationales expressed in the precedent than congressional silence. Reliance on shadow precedents turns the principles of legislative supremacy on their head. By contrast, a presumption against relying on an overridden precedent refocuses the analysis on the statutory language and congressional signals, rather than on judicial interpretations.

In the long run at least, the proposed reforms would also better accord with the values served by precedent. Reducing reliance on shadow precedents better promotes fairness in that it makes it more likely that cases that are relevantly similar are treated alike. This is true both for substantive similarity (for example, breastfeeding and pregnancy) and "statutory" similarity (for example, mixed-motive decisions in disability discrimination cases and mixed-motive decisions in sex discrimination cases).

The reforms' effects on predictability would be more mixed. As a threshold matter, the proposed reforms would make it clear that the shadow precedent itself was unlikely to be applied, thus increasing predictability. This does not mean, however, that the new interpretation would always be predictable. That would depend on the clarity of the override and whether it is obvious how it would "reinterpret" the preexisting law. In cases, such as Patterson, where the overridden case departed from a preexisting, relatively uniform statutory interpretation, or where an alternative interpretation is clear, the new rule would increase predictability. In cases where the prior law was unsettled or where the override itself is ambiguous, unpredictability would remain. (286)

Likewise, the rule would probably increase efficiency in the long run but with some short-term efficiency costs. Courts generally would no longer need to struggle with whether and how to apply a shadow precedent, nor would they (absent clear congressional directives) need to develop parallel bodies of law addressing common issues in so-called "related statutes." On the other hand, the rule would call for "fresh" statutory analysis rather than simply relying on prior interpretations. This would of course take time, but it would replace the complicated task of trying to reconcile competing signals offered by the override and the overridden precedent.

Finally, the proposed reforms would nicely complement existing rules regarding retroactivity. As discussed in Part II, the Supreme Court has established a general rule that overrides will be effective prospectively only. In reaching this decision, it noted that an authoritative interpretation of a statute by the Supreme Court--even one that disagrees with pre-existing lower court interpretations and is subsequently repudiated by Congress--is presumed to establish what the statute has always meant. Prospectively, however, legislative supremacy dictates that it should be within Congress' power to determine what the statute will mean. Reliance on shadow precedents undermines this basic precept. Establishing a general rule that interpretations that Congress repudiates will no longer hold sway helps ensure that Congress can truly realize its power not only to add language to a statute but also to reinterpret the language that is already there.

CONCLUSION

Under the separation of powers set forth in the Constitution, Congress has the power to make laws and courts are charged with applying them. Although it is widely recognized that through statutory interpretation courts can be said to "make" statutory policy, it is always with the assumption--often made explicit--that Congress can override the courts' interpretations if they are incorrect or not in accordance with congressional will. This Article reveals that in practice, however, courts in employment discrimination cases often show surprising disregard for Congress' disapproval of precedent in an override. Empirical studies suggest that this may be particularly common in areas in which judges tend to be partisan in their rulings. The interpretive reforms proposed in this Article--which would create a presumption against applying shadow precedents--would help ensure that legislative overrides can serve as a true check on judicial lawmaking. But it is equally important that Congress can take steps to articulate more clearly the extent to which it intends an override to end reliance on the prior judicial interpretation and litigants can use the arguments described above to articulate reasoned justifications against relying on a shadow precedent.

There is a new opportunity to test some of these propositions in the employment discrimination context. As this Article was being finalized for publication, Congress enacted the ADA Amendments Act, overriding two seminal Supreme Court cases that interpret the meaning of "disability" under the ADA. Numerous commentators have contended that the Supreme Court's interpretations were far narrower than Congress had intended, (287) and, although it is impossible to ascertain definitively the intent of the Congress that initially enacted the ADA, the current Congress apparently agreed. The findings for the ADA Amendments state explicitly that the Supreme Court's prior interpretations "eliminat[ed] protection for many individuals whom Congress intended to protect" (288) and directs that the (revised) definition be construed "in favor of broad coverage." (289) Moreover, the purposes clauses explicitly "reject" aspects of the Supreme Court's "reasoning" and "standards" announced in the overridden precedents. (290) But undoubtedly questions will arise that are not squarely addressed by the substantive text of the override. Courts then will be forced to struggle with the questions at the heart of this project: whether to follow the precedents that Congress has repudiated, and, if not, how to interpret or re-interpret the language that gave rise to them.

(1) Johnson v. Transp. Agency, 480 U.S. 616, 630 n.7 (1987) (quoting GUIDO CALABRESI, A COMMON LAW FOR THE AGE OF STATUTES 31-32 (1982)).

(2) See, e.g, William N. Eskridge, Jr., Overriding Supreme Court Statutory Interpretation Decisions, 101 YALE L.J. 331 (1991).

(3) See Nancy C. Staudt et al., Judicial Decisions As Legislation: Congressional Oversight of Supreme Court Tax Cases, 1954-2005, 82 N.Y.U. L. REV. 1340, 1354 (2007).

(4) JEB BARNES, OVERRULED? LEGISLATIVE OVERRIDES, PLURALISM, AND CONTEMPORARY COURT-CONGRESS RELATIONS 90 (2004).

(5) James Brudney considers many related questions in an article exploring the extent to which courts should consider legislative history expressing approval or disapproval of court decisions. See James J. Brudney, Congressional Commentary on Judicial Interpretations of Statutes: Idle Chatter or Telling Response?, 93 MICH. L. REV. 1 (1994). Additionally, in the employment discrimination context, there are a handful of articles that explore the specific question of whether an interpretation that has been overridden with respect to one statute should be applied to other statutes that are generally interpreted in tandem. See infra note 163.

(6) See infra Part III.D.

(7) See, e.g., Shearson/Am. Express, Inc. v. McMahon, 482 U.S. 220, 268 (1986) (Stevens, J., concurring in part and dissenting in part) ("[A]fter a statute has been construed, either by this Court or by a consistent course of decision by other federal judges or agencies, it acquires a meaning that should be as clear as if the judicial gloss had been drafted by the Congress itself.").

(8) See, e.g., Gomez-Perez v. Potter, 128 S. Ct. 1931, 1936 (2008) (relying on prior statutory precedents to construe a similar discrimination statute).

(9) See Frederick Schauer, Precedent, 39 STAN. L. REV. 571, 595-601 (1987).

(10) Pub. L. No. 102-166, 105 Stat. 107 (codified as amended in scattered sections of 2, 16, 29, 42 U.S.C.).

(11) See also Eskridge, supra note 2, at 333 n.4 (cataloguing overridden cases).

(12) Pub. L. No. 110-325, 122 Stat. 3553 (2008) (codified in scattered sections of 29, 42 U.S.C.) (effective Jan. 1, 2009).

(13) See Lilly Ledbetter Fair Pay Act of 2007, H.R. 2831, 110th Cong. (as passed by House, July 31, 2007); Fair Pay Restoration Act, S. 1843, 110th Cong. (2007); Carrie Sheffield, Filibuster Blocks Wage-Bias Lawsuit Bill, WASH. TIMES, Apr. 24, 2008, at A4 (discussing the filibuster).

(14) An empirical study suggests that areas of the law that are typically subject to partisan divides may yield greater judicial dissensus after overrides than other areas. See BARNES, supra note 4, at 169-70. In this respect, the prevalence of shadow precedents in employment discrimination may be somewhat unusual. Nonetheless, it is particularly in such areas of the law that there is greatest reason for concern that reliance on shadow precedents may undermine Congress' authority to shape statutory law.

(15) 490 U.S. 900 (1989).

(16) 490 U.S. 228 (1989).

(17) 429 U.S. 125 (1976).

(18) U.S. CONST. art. I, [section] 1. This Article primarily discusses federal legislation and the federal courts' interpretation of it. Similar principles would apply, however, under most state constitutional structures.

(19) See, e.g., Plaut v. Spendthrift Farm, Inc., 514 U.S. 211, 222 (1995) ("The power of '[t]he interpretation of the laws' [is] 'the proper and peculiar province of the courts.'" (first alteration in original) (quoting THE FEDERALIST No. 78, at 525 (Alexander Hamilton) (J. Cooke ed., 1961))).

(20) See, e.g., Edward O. Correia, A Legislative Conception of Legislative Supremacy, 42 CASE W. RES. L. REV. 1129, 1141-43 (1992) (explaining the debate over the proper meaning and significance of legislative supremacy and how that debate shapes alternative approaches to statutory interpretation); Daniel A. Farber, Statutory Interpretation and Legislative Supremacy, 78 GEO. L.J. 281, 283-94 (1989) (describing "strong" and "weak" conceptions of supremacy).

(21) See Chevron U.S.A. Inc. v. Nat'l Res. Def. Council, Inc., 467 U.S. 837, 844 (1984). In practice, however, the level of judicial deference to agency interpretations varies dramatically. See William N. Eskridge, Jr. & Lauren E. Baer, The Continuum of Deference: Supreme Court Treatment of Agency Statutory Interpretations from Chevron to Hamdan, 96 GEO. L.J. 1083 (2008).

(22) Jerome Frank, Words and Music: Some Remarks on Statutory Interpretation, 47 COLUM. L. REV. 1259, 1269 (1947).

(23) See generally REED DICKERSON, THE INTERPRETATION AND APPLICATION OF STATUTES 43-53 (1975) (discussing the inherent limitations of language and typical causes of ambiguity in statutes). Sometimes legislators leave questions unresolved in order to build a consensus to enact a bill. See, e.g., Victoria F. Nourse & Jane S. Schacter, The Polities of Legislative Drafting: A Congressional Case Study, 77 N.Y.U. L. REV. 575, 594-96 (2002) (documenting "deliberate ambiguity" in statutes).

(24) See, e.g., Green v. Bock Laundry Mach. Co., 490 U.S. 504, 527 (1989) (Scalia, J., concurring) (declining to apply plain language of a statute which, "if interpreted literally, produces an absurd, and perhaps unconstitutional, result"); see also Jonathan R. Siegel, What Statutory Drafting Errors Tech Us About Statutory Interpretation, 69 GEO. WASH. L. REV. 309, 326-32 (2001) (exploring the textualist "absurd results exception" as it relates to the federal venue statute).

(25) Jane S. Schacter, Metademocracy: The Changing Structure of Legitimacy in Statutory Interpretation, 108 HARV. L. REV. 593, 593 (1995).

(26) The term was coined by Alexander Bickel. See ALEXANDER M. BICKEL, THE LEAST DANGEROUS BRANCH 16 (2d ed. Yale University Press 1986).

(27) See, e.g., Lawrence C. Marshall, "Let Congress Do It": The Case for an Absolute Rule of Statutory Stare Decisis, 88 MICH. L. REV. 177, 204 (1989) ("There is no doubt that the potential for congressional override significantly mitigates the countermajoritarian difficulty of statutory interpretation....").

(28) Cf BARNES, supra note 4, at 31-34 (discussing the relative disadvantages of alternative control mechanisms such as impeachment, review of judicial appointments, control of the budget, and the ability to propose constitutional amendments).

(29) The term "super-strong" is from William N. Eskridge, Jr., Overruling Statutory Precedents, 76 GEO. L.J. 1361, 1366 (1987).

(30) See, e.g., Flood v. Kuhn, 407 U.S. 258, 284 (1972) ("If there is any inconsistency or illogic in all this, it is an inconsistency and illogic of long standing to be remedied by the Congress and not by this Court.").

(31) See, e.g., Hohn v. United States, 524 U.S. 236, 251-53 (1998).

(32) See, e.g., Griffin v. Oceanic Contractors, Inc., 458 U.S. 564, 575-76 (1982) (applying statutory language notwithstanding odd or arguably absurd policy results on the grounds that any changes should be made by Congress); Tenn. Valley Auth. v. Hill, 437 U.S. 153, 194-95 (1978) (same).

(33) See EDWARD H. LEVI, AN INTRODUCTION TO LEGAL REASONING 32 (1949); Marshall, supra note 27, at 200-15.

(34) See, e.g., WILLIAM N. ESKRIDGE, JR., DYNAMIC STATUTORY INTERPRETATION 52-55 (1994); T. Alexander Aleinikoff, Updating Statutory Interpretation, 87 MICH. L. REV. 20, 46-66 (1988); Einer Elhauge, Preference-Estimating Statutory Default Rules, 102 COLUM. L. REV. 2027, 2112-26 (2002). Amanda Tyler groups these theorists together as "dynamic" theorists. Amanda L. Tyler, Continuity, Coherence, and the Canons, 99 NW. U. L. REV. 1389, 1391 (2005).

(35) See GUIDO CALABRESI, A COMMON LAW FOR THE AGE OF STATUTES 105 (1982).

(36) See, e.g., id. at 169 (observing that the potential for override creates a check against "willful" judicial abuse, and arguing that the check might be insufficient); ESKRIDGE, supra note 34, at 151 ("[D]ynamic statutory interpretation, even against legislative expectations, is subject to override by the legislature and in fact may even be a stimulus to legislative deliberation.").

(37) Professors Elhauge and Eskridge each also suggest that courts should sometimes intentionally trigger a congressional response. See ESKRIDGE, supra note 34, at 153; Elhauge, supra note 34, at 2165 ("[O]ften the best choice is instead a ... rule that is more likely to provoke a legislative reaction.").

(38) This term is from Richard A. Paschal, The Continuing Colloquy: Congress and the Finality of the Supreme Court, 8 J.L. & Pot.. 143, 143 (1991). Even if the Court is acting in good faith, it may be perceived as seeking to impose its own policy preferences and thus overrides may come at an institutional cost to the Court. See Abner J. Mikva & Jeff Bleich, When Congress Overrules the Court, 79 CAL. L. REV. 729, 747 (1991).

(39) For a good overview of the theory and application to a few examples, see Rafael Gely & Pablo T. Spiller, A Rational Choice Theory of Supreme Court Statutory Decisions with Applications to the State Farm and Grove City Cases, 6 J.L. ECON. & ORG. 263 (1990); cf Pablo T. Spiller & Emerson H. Tiller, Invitations to Override: Congressional Reversals of Supreme Court Decisions, 16 INT'L REV. L. & ECON. 503 (1996) (theorizing that the Court may specifically "invite" an override when it prefers to decide a case in a given manner, e.g., relying on textualism, and is confident that Congress will override the resulting policy choice that the Court also does not prefer). Other models focus on the extent to which the Court responds to the sitting Congress and President and their wishes rather than to the enacting Congress and its wishes. See, e.g., William N. Eskridge, Jr., Reneging on History ? Playing the Court/Congress/President Civil Rights Game, 79 CAL. L. REV. 613, 641-66 (1991) (using game theory modeling to demonstrate that statutory interpretation decisions by the Supreme Court are influenced by the policy preferences of the sitting Congress and President); John Ferejohn & Barry Weingast, Limitations of Statutes: Strategic Statutory Interpretation, 80 GEO. L.J. 565, 574-76 (1992) (using Euclidean modeling to demonstrate that judicial review, if done strategically, either leaves outcomes undisturbed or moves outcomes toward the sitting legislature's preferences).

(40) See, e.g., Pablo T. Spiller & Rafael Gely, Congressional Control or Judicial Independence: The Determinants of U.S. Supreme Court Labor-Relations Decisions, 1949-1988, 23 RAND J. ECON. 463, 489 (1992) (concluding that court decisions are often constrained by preferences of elected officials). But see, e.g., Jeffrey A. Segal, Separation-of-Powers Games in the Positive Theory of Congress and Courts, 91 AM. POL. SCI. REV. 28, 42 (1997) (concluding that Justices typically rule according to their own sincere policy preferences without being constrained by potential congressional response).

(41) See BARNES, supra note 4, at 118.

(42) See generally, e.g., WILLIAM N. ESKRIDGE, JR. ET AL., CASES AND MATERIALS ON LEGISLATION 54-60 (4th ed. 2007) (summarizing central theorists).

(43) See William N. Eskridge, Jr., Politics Without Romance: Implications of Public Choice Theory for Statutory Interpretation, 74 VA. L. REV. 275, 283 (1988) (arguing that majoritarian ideals of statutory enactment are flawed, in part, because of rent-seeking).

(44) See, e.g., Victor M. Sher & Carol Sue Hunting, Eroding the Landscape, Eroding the Laws: Congressional Exemptions from Judicial Review of Environmental Laws, 15 HARV. ENVTL. L. L. REV. 435, 487-89 (1991) (describing how industry representatives have lobbied successfully for overrides that provide project-specific exemptions from environmental laws).

(45) Eskridge, supra note 2, at 332 n.1 (emphasis added).

(46) See, e.g., BARNES, supra note 4, at 76-78 (using Eskridge's data but cross-referencing it with other studies and an independent Lexis search); Lori Hausegger & Lawrence Baum, Behind the Scenes: The Supreme Court and Congress in Statutory Interpretation, in GREAT THEATRE: THE AMERICAN CONGRESS IN THE 1990S, at 224, 225 n.1, 227 (Herbert F. Weisberg & Samuel C. Patterson eds., 1998) (using Eskridge's data but extending analysis through 1996).

(47) Studies before 1975 typically found very few overrides by Congress. See Eskridge, supra note 2, at 338; see also, e.g., Beth Henschen, Statutory Interpretations of the Supreme Court: Congressional Response, 11 AM. POL. Q. 441, 445 (1983) (finding very few overrides of antitrust and labor legislation between 1950 and 1972).

(48) This term is derived from "statutorification," coined by Guido Calabresi. CALABRESI, supra note 35, at 1.

(49) See Eskridge, supra note 2, at 339-40 (discussing the significant correlation between the increase in overrides and the size of congressional staffs). He also suggests that increases in the number of lower federal court decisions, omnibus bills, and organized interest group activity contributed to the rapid growth in overrides. Id. at 338.

(50) Hausegger & Baum, supra note 46, at 224.

(51) See Eskridge, supra note 2, at 342 tbl.3 (finding that the judiciary committee held hearings on 38% of Supreme Court statutory decisions that fell within its jurisdiction between 1977 and 1983); Staudt et al., supra note 3, at 1352-53 (finding 54% of Supreme Court tax decisions were discussed in some fashion).

(52) Eskridge, supra note 2, at 338; cf BARNES, supra note 4, at 197-209 app. (listing overrides, including many that are overrides of circuit court decisions); Daniel J. Bussel, Textualism's Failures: A Study of Overruled Bankruptcy Decisions, 53 VAND. L. REV. 887, app. at 930-38 (2000) (listing overrides of bankruptcy decisions, including significant numbers of lower court decisions).

(53) See generally Brudney, supra note 5, at 5 (discussing this phenomenon in detail).

(54) Eskridge, supra note 2, at 335-36, 338 tbl.1.

(55) See, e.g., Hausegger & Baum, supra note 46, at 238 (finding that Congress has overridden at least 5.6% of Supreme Court decisions in the 1978-1989 Terms).

(56) See Brudney, supra note 5, at 21.

(57) See ESKRIDGE ET AL, supra note 42, at 24-38.

(58) Factors that typically are found to increase the likelihood of a congressional response include the extent to which organized interest groups care about the issue, whether the U.S. government is a losing party, and whether the decision was nonunanimous. See, e.g., Hausegger & Baum, supra note 46, at 236-37; Beth M. Henschen & Edward I. Sidlow, The Supreme Court and the Congressional Agenda-Setting Process, 5 J.L. & POL. 685, 690-91 (1989); Virginia A. Hettinger & Christopher Zorn, Explaining the Incidence and Timing of Congressional Responses to the U.S. Supreme Court, 30 LEGIS. STUD. Q. 5, 10-14 (2005); Joseph Ignagni & James Meernik, Explaining Congressional Attempts to Reverse Supreme Court Decisions, 47 POL. RES. Q. 353, 358-59, 361 (1994); Harry P. Stumpf, Congressional Response to Supreme Court Rulings: The Interaction of Law and Politics, 14 J. PUB. L. 377, 391-92 (1965).

(59) See BARNES, supra note 4, at 16.

(60) Johnson v. Transp. Agency, 480 U.S. 616, 672 (1987) (Scalia, J., dissenting).

(61) See id. at 671-72; see also, e.g., William N. Eskridge, Jr., Interpreting Legislative Inaction, 87 MICH. L. REV. 67, 90-108 (1988) (discussing the various problems that result when inferring intent from legislative inaction); Marshall, supra note 27, at 186-96 (attributing acquiescence to four factors: ignorance, inertia, interpretational ambiguity, and irrelevance). But see, e.g., Johnson, 480 U.S. at 629 n.7 (considering Congress' failure to enact an override significant where the prior decision was widely publicized and there were no override bills proposed).

(62) See BARNES, supra note 4, at 76-77, 84.

(63) Intracircuit disagreement is defined as cases in which there are either reversals of a lower court decision or a dissent on appeal; "significant" intracircuit dissensus is defined as the presence of multiple dissents and reversals, not merely an isolated dissent or reversal. Id. at 84-85.

(64) Id. at 146.

(65) Id. at 90 fig.4.1 (finding forty-nine cases of dissensus and fifty-one cases of consensus).

(66) Id. at 91 fig.4.2 (finding twenty-one cases of dissensus and no cases of consensus).

(67) Id. (finding twenty-three cases of dissensus and forty-five cases of consensus).

(68) Id. at 169, 171.

(69) Id. at 169.

(70) Influential discussions of precedent and the rationales that underlie the rule of stare decisis in both American and British law include: LEVI, supra note 33, at 2; PRECEDENT IN LAW (Laurence Goldstein ed., 1987); Earl Maltz, The Nature of Precedent, 66 N.C.L. REV. 367 (1988); Schauer, supra note 9; A.W.B. Simpson, The Ratio Decidendi of a Case and the Doctrine of Binding Precedent, in OXFORD ESSAYS IN JURISPRUDENCE 148 (A.G. Guest ed., 1961).

(71) The precise meaning of holding, ratio decidendi, and dictum has long been a subject of much dispute. See Maltz, supra note 70, at 372 & n.21.

(72) See generally Peter M. Tiersma, The Textualization of Precedent, 82 NOTRE DAME L. REV. 1187, 1190-1247 (2007) (discussing the evolution of this process). Similarly, Earl Maltz suggests that precedent may be reframed as including the general doctrine of the decision (that is, judicial conventions that are at play but that are not explicitly linked to the substantive scope of the decision), the specific doctrine of the decision (the rule of law that the court describes as controlling the particular matter), and the rationale (the reasons given by the court as supporting the specific doctrine announced in the case). See Maltz, supra note 70, at 373-83.

(73) See Richard W. Murphy, Separation of Powers and the Horizontal Force of Precedent, 78 NOTRE DAME L. REV. 1075, 1085-86 (2003).

(74) See, e.g., Stefanie A. Lindquist & Frank B. Cross, Empirically Testing Dworkin's Chain Novel Theory: Studying the Path of Precedent, 80 N.Y.U. L. REV. 1156, 1161 n.21 (2005).

(75) Marshall, supra note 27, at 218.

(76) See Murphy, supra note 73, at 1085-86.

(77) Patterson v. McLean Credit Union, 491 U.S. 164, 173 (1989).

(78) See, e.g., Lindquist & Cross, supra note 74, at 1159-61 (summarizing several scholars on precedents); Schauer, supra note 9, at 595-99 (same).

(79) See Lindquist & Cross, supra note 74, at 1164-65 (summarizing several scholars on precedent).

(80) Boumediene v. Bush, 128 S. Ct. 2229, 2265 (2008) ("When interpreting a statute, we examine related provisions in other parts of the U.S. Code.").

(81) See, e.g., Smith v. City of Jackson, 544 U.S. 228, 233 (2005) (plurality opinion) (" [W] hen Congress uses the same language in two statutes having similar purposes, particularly when one is enacted shortly after the other, it is appropriate to presume that Congress intended that text to have the same meaning in both statutes.").

(82) See, e.g., id. at 240 (plurality opinion) (applying a 1971 interpretation of a statute enacted in 1964 to a related statute enacted in 1967). But see id. at 260 (O'Connor, J., concurring) (criticizing the plurality's approach on this point).

(83) For example, Price Waterhouse v. Hopkins, 490 U.S. 228 (1989), addressed several important issues in employment discrimination law. One was the liability standard for "mixed-motive" claims. Id. at 252. The Supreme Court's resolution of the issue was overridden by Congress in the 1991 Civil Rights Act. See infra Part III.C. Another was that decisions based on gender-based stereotypes could be actionable under Title VII. Price Waterhouse, 490 U.S. at 250-52. The 1991 Civil Rights Act did not in any way disturb this holding and it unquestionably remains persuasive precedent today. See, e.g., Back v. Hastings on Hudson Union Free Sch. Dist., 365 F.3d 107, 119-20 (2d Cir. 2004).

(84) See J.M. BARRIE, PETER PAN AND WENDY 17-33 (1911).

(85) I use "rationale" here and below to mean reasoning that analyzes the specific statutory language in question, not general canons of statutory interpretation. Cf. Maltz, supra note 70, at 373-83 (distinguishing between what he classifies as "general doctrine," such as canons of construction, that is not linked to the substantive statutory language and "rationales").

(86) Of course, there is significant disagreement among judges and commentators regarding the extent to which legislative history should be consulted in statutory interpretation. For two influential discussions, see ANTONIN SCALIA, A MATTER OF INTERPRETATION 29-37 (1997) (arguing against use); Stephen Breyer, On the Uses of Legislative History in Interpreting Statutes, 65 S. CAL. L. REV. 845, 847 (1992) (arguing for careful use) ; and also see ESKRIDGE ET AL., supra note 42, at 990 n.j (collecting scholarly commentary on both sides of the debate). To the extent that judges are willing to consult legislative history, committee reports are generally considered particularly authoritative indications of congressional intent, in part because legislators often rely upon the summary of legislation they provide. See id. at 981-83; see also Nourse & Schacter, supra note 23, at 607 (providing an empirical study documenting reliance on committee reports by members of Congress).

(87) See infra text accompanying notes 196-212.

(88) BLACK'S LAW DICTIONARY 620 (8th ed. 2004).

(89) See infra notes 197-202 and accompanying text.

(90) 511 U.S. 244 (1994).

(91) 511 U.S. 298 (1994).

(92) Landgraf 511 U.S. at 258-59.

(93) Id. at 259.

(94) Id. at 265.

(95) Id. at 304.

(96) Rivers, 511 U.S. at 309-10.

(97) Id. at 310.

(98) The Court specifically cited specific modifications regarding the retroactivity provisions and the purposes language made from a version of the bill that was vetoed in 1990. See id. at 307-08; see also infra notes 120-21 and accompanying text.

(99) Rivers, 511 U.S. at 312-13 (emphasis added).

(100) Id. at 313 n.12.

(101) Id. at 304 (internal quotation marks omitted).

(102) See supra note 68 and accompanying text.

(103) See BARNES, supra note 4, at 171; see also U.S. CONST. amend. XIV, [section] 1 ("No State shall ... deny to any person within its jurisdiction the equal protection of the laws.").

(104) 42 U.S.C.A. [section][section] 2000e to 2000e-17 (West 2003 & Supp. 2008).

(105) Id. [section] 2000e-2.

(106) Id. [section] 2000e-3(a).

(107) Pub. L. No. 90-202, 81 Stat. 602 (codified as amended at 29 U.S.C.A. [section][section] 621-624 (West 2000 & Supp. 2008)).

(108) 29 U.S.C.A. [section] 623.

(109) Pub. L. No. 101-336, 104 Stat. 327 (codified as amended at 42 U.S.C. [section][section] 12101-12213 (2000)).

(110) [section][section] 101-108, 104 Stat. at 330-37 (Title I).

(111) 42 U.S.C. [section] 12112 (2000). Title VII also requires employers to reasonably accommodate employees' religious observance. See id. [section] 2000e(j).

(112) Id. [section] 1981.

(113) See, e.g., Patterson v. McLean Credit Union, 491 U.S. 164, 171 (1989) (citing Runyon v. McCrary, 427 U.S. 160, 168 (1976)).

(114) See ADMIN. OFFICE OF THE U.S. COURTS, JUDICIAL BUSINESS OF THE UNITED STATES COURTS, 145-47 tbl.C-2 (2007), available at http://www.uscourts.gov/judbus 2007/appendices/C02Sep07.pdf (showing that employment discrimination cases were the most common nonprisoner federal question filing in U.S. district courts).

(115) See Staudt et al., supra note 3, at 1351 (showing Title VII, ADA, and ADEA among the most frequently litigated statutes, although they rank well below the Internal Revenue Code, the Federal Rules of Civil Procedure, and the Bankruptcy Code).

(116) The summary that follows was drawn from two extremely detailed accounts of the negotiations that gave rise to the Act. See Reginald C. Govan, Honorable Compromises and the Moral High Ground: The Conflict Between the Rhetoric and the Content of the Civil Rights Act of 1991, 46 RUTGERS L. REV. 1 (1993) (written by a former House Democratic staffer); Peter M. Leibold et al., Civil Rights Act of 1991: Race to the Finish--Civil Rights, Quotas, and Disparate Impact in 1991, 45 RUTGERS L. REV. 1043 (1993) (written by former Senate Republican staffers).

(117) Govan, supra note 116, at 23 (quoting Julie Johnson, High Court Called Threat to Blacks, N.Y. TIMES, July 10, 1989, at A14).

(118) Id. at 23-24 (quoting Metzenbaum Seeks Reversal of Wards Cove, 131 Lab. Rel. Rep. (BNA) 309, 309 (July 3, 1989)).

(119) See Landgraf v. USI Film Prods., 511 U.S. 244, 255 & n.8 (1994).

(120) Civil Rights Act of 1991, H.R. 1,102d Cong. [section] 2 (1991) (emphasis added); Civil Rights Act of 1990, S. 2104, 101st Cong. [section] 2 (1990).

(121) Civil Rights Act of 1991, Pub. L. No. 102-166, [section] 3, 105 Star. 1071, 1071 (emphasis added).

(122) See Rivers v. Roadway Express, Inc., 511 U.S. 298, 307-08 (1994).

(123) See infra text accompanying notes 152-55.

(124) See H.R. REP. No. 102-40, pts. 1-2 (1991), reprinted in 1991 U.S.C.C.A.N. 549.

(125) See 137 CONG. REC. 29,045-47 (1991) (memorandum introduced by Sen. Danforth).

(126) See H.R. PEP. No. 101-644, pts. 1-2 (1990); S. REP. No. 101-315 (1990); H.R. PEP. No. 101-856 (1990) (Conf. Rep.); H.R. REP. No. 101-755 (1990) (Conf. Rep.).

(127) Of course, some prominent judges and commentators contend that legislative history should not be consulted in statutory interpretation. See supra note 86. Although a full discussion of this thorny issue is beyond the scope of this Article, I generally believe that judicious use of legislative history is appropriate in statutory interpretation.

(128) 127 S. Ct. 2162 (2007).

(129) 42 U.S.C. [section] 2000e-5(e) (2000).

(130) 490 U.S. 900, 902 (1989).

(131) Id.

(132) Id. at 906-08.

(133) 449 U.S. 250 (1980).

(134) Id. at 258.

(135) 431 U.S. 553 (1977).

(136) Id. at 560.

(137) 478 U.S. 385 (1986).

(138) See id. at 395-96 (Brennan, J., concurring). The per curiam decision of the Court adopted Justice Brennan's reasoning. Id. at 386-87 (per curiam).

(139) Lorance, 490 U.S. at 912 n.5. This distinction is somewhat unsatisfying because, although the pay structure in Bazemore was based on a facially discriminatory division of jobs that predated the enactment of Title VII, the pay structure was not "facially" discriminatory when the plaintiffs brought suit.

(140) Lorance, 490 U.S. at 917 (Marshall, J., dissenting) (quoting Del. State Coll. v. Ricks, 449 U.S. 250, 257-58 (1980)).

(141) Id. at 916-17.

(142) Id. at 915.

(143) Id. at 914.

(144) Civil Rights Act of 1991, Pub. L. No. 102-166, [section] 112, 105 Stat. 1071, 1079 (codified at 42 U.S.C. [section] 2000e-5 (2000)).

(145) 137 CONG. REC. 29,047 (1991) (memorandum introduced by Sen. Danforth).

(146) Id.

(147) H.R. REP. No. 102-40, pt. 2, at 22-24, 40-41 (1991), reprinted in 1991 U.S.C.C.A.N. 694, 715-18, 733-35. The House bill would have modified the time limit provision to include a general rule that claims must be brought within the specified time after they have "occurred" or have "been applied to affect adversely the person aggrieved, whichever is later." H.R. 1, 102d Cong. [section] 7 (1991). This language largely tracks that which passed both houses in 1990. See S. 2104, 101st Cong. [section] 7 (1990). It obviously would have made it clearer that Lorance's rationale was not to be extended to other contexts, but it arguably would have also dramatically expanded potential liability for all kinds of other actions. Thus, particularly given the Senate sponsors' interpretative memorandum, see supra note 125, the ultimate failure to enact this broader language should not be understood as clear congressional intent to cabin the override specifically to seniority systems and leave unchallenged the general rationales of Lorance.

(148) Ledbetter v. Goodyear Tire & Rubber Co., 421 F.3d 1169, 1175-76 (11th Cir. 2005).

(149) Id. at 1180, 1184.

(150) Id. at 1169.

(151) Ledbetter v. Goodyear Tire & Rubber Co., 127 S. Ct. 2162, 2165 (2007).

(152) Id. at 2169 n.2.

(153) Id.

(154) Id.

(155) Id. (quoting id. at 2183 (Ginsburg, J., dissenting)).

(156) Id. at 2173.

(157) Id. at 2183 (Ginsburg, J., dissenting).

(158) Id. (alteration in original) (quoting Lorance v. AT&T Techs., 490 U.S. 900, 914 (1989) (Marshall, J., dissenting)).

(159) Id. at 2183-84 (quoting, inter alia, the Senate memorandum discussed in the text accompanying note 125). Justice Ginsburg's support for her contention that Congress intended to generalize Bazemore is rather puzzling. She relies upon a Senate committee report that accompanied the 1990 bill. Id. (quoting S. Rep. 101-315, at 54 (1990)). However, although Ginsburg characterizes the 1991 Act as "in all material respects identical to the proposed 1990 Act," id., in fact the version of the legislation passed by the Senate in 1990 was more explicit about generalizing the override (it was identical to the broader language that passed in the House in 1991). See supra note 147. The language Ginsburg quotes regarding Bazemore was identified as explaining this more general provision, not the specific reference to seniority systems ultimately enacted in 1991.

(160) See, e.g., Cox v. City of Memphis, 230 F.3d 199, 203 (6th Cir. 2000) ("Lorance stands for the proposition that 'the distinction does not turn on the type of discrimination, but on whether the practice at issue is part of, or a repetition of, a past discriminatory act....'" (quoting Anderson v. City of Bristol, 6 F.3d 1168, 1175 (6th Cir. 1993))); Lettis v. U.S. Postal Serv., 39 F. Supp. 2d 181, 195 (E.D.N.Y. 1998) (quoting Lorance, 490 U.S. at 407, without acknowledging the 1991 amendments, for the proposition that in determining when a statute of limitations begins to run, "the proper focus is upon the time of the discriminatory acts, not upon the time at which the consequences of the acts became most painful" (internal quotation marks omitted)).

(161) 490 U.S. 642 (1989).

(162) The most prominent example of Wards Cove as a shadow precedent in a related statute is the Supreme Court's use of it to interpret the ADEA in Smith v. City of Jackson, 544 U.S. 228, 240 (2005) ("While the relevant 1991 amendments expanded the coverage of Title VII, they did not amend the ADEA or speak to the subject of age discrimination. Hence, Wards Cove's pre-1991 interpretation of Title VII's identical language remains applicable to the ADEA."). The Supreme Court subsequently characterized this statement as meaning only that disparate impact claims are cognizable under the ADEA and that the plaintiff-employee has the burden of identifying particular practices that cause the disparate impact. See Meacham v. Knolls Atomic Power Lab., 128 S. Ct. 2395, 2404-07 (2008). For the application of Lorance to other statutes, see, for example, Huels v. Exxon Coal USA, Inc., 121 F.3d 1047, 1050 n.1 (7th Cir. 1997) (which applies Lorance to an ADA claim), and Casteel v. Executive Bd. of Local 703 of the International Brotherhood of Teamsters, 272 F.3d 463, 467 (7th Cir. 2001), which applies Lorance to an ADEA claim.

(163) Most commentators argue against applying the overridden precedents to other statutes. See, e.g., Robert Belton, Mixed-Motive Cases in Employment Discrimination Law Revisited: A Brief Updated View of the Swamp, 51 MERCER L. REV. 651, 664 (2000) (arguing that Price Waterhouse should not apply to the ADA) ; Martin J. Katz, Unifying Disparate Treatment (Really), 59 HASTINGS L.J. 643, 664-81 (2008) (arguing that courts should stop applying Price Waterhouse to other statutes) ; Seam Park, Comment, Curing Causation: Justifying a "Motivating-Factor" Standard Under the ADA, 32 FLA. ST. U. L. REV. 257, 273-75 (2004) (arguing that Price Waterhouse should not be applied to the ADA); Jamie Darin Prenkert, Bizarro Statutory Stare Decisis, 28 BERKELEY J. EMP. & LAB. L. 217, 247 (2007) (arguing that the Supreme Court erred in suggesting Wards Cove applied to the ADEA). But see John L. Flynn, Note, Mixed-Motive Causation Under the ADA: Linked Statutes, Fuzzy Thinking, and Clear Statements, 83 GEO. L.J. 2009, 2035-37 (1995) (arguing that Price Waterhouse should apply to the ADA).

(164) 490 U.S. 228, 253-55 (1989).

(165) 42 U.S.C. [section] 2000e-2(a) (2000) (emphasis added).

(166) Price Waterhouse, 490 U.S. at 244-45.

(167) See, e.g., Fakete v. Aetna, Inc., 308 F.3d 335, 337 n.2 (3d Cir. 2002).

(168) Price Waterhouse, 490 U.S. at 276 (O'Connor, J, concurring). There was also a concurrence by Justice White that largely tracks the standards applied by Justice O'Connor. Id. at 258-60 (White, J., concurring). A dissent by Justice Kennedy, joined by Justices Scalia and Rehnquist, would have required a showing that the illegitimate factor made a "difference" in the decision ("but for" causation) and then followed the burden-shifting laid out in Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 252-56 (1981). See Price Waterhouse, 490 U.S. at 281, 286 (Kennedy, J., dissenting).

(169) See Price Waterhouse, 490 U.S. at 244-45; id. at 276-77 (O'Connor, J., concurring).

(170) 42 U.S.C. [section] 2000e-2(m).

(171) Id. [section] 2000e-5(g)(2)(B) (permitting declaratory and certain injunctive relief, as well as attorney's fees, but not damages or orders of reinstatement).

(172) H.R. REP. No. 102-40, pt. 1, at 47 (1991), reprinted in 1991 U.S.C.C.A.N. 549, 585.

(173) H.R. REP. No. 102-40, pt. 2, at 4 (1991), repented in 1991 U.S.C.C.A.N. 694, 696-97.

(174) Martin Katz compiles the mixed case law on the issue under the ADA, ADEA, and Title VII retaliation claims, as well as [section] 1981 and the Family and Medical Leave Act. See Katz, supra note 163, at 647 n.22, 650 n.31.

(175) For ADEA claims, see, for example, Gross v. FBL Financial Services, Inc., 526 F.3d 356, 361-62 (8th Cir. 2008); Glanzman v. Metropolitan Management Corp., 391 F.3d 506, 512 n.3 (3d Cir. 2004); Equal Employment Opportunity Commission v. Warfield-Rohr Casket Co., 364 F.3d 160, 164 n.2 (4th Cir. 2004); Lewis v. Young Men's Christian Ass'n, 208 F.3d 1303, 1305 & n.2 (11th Cir. 2000) (per curiam), each of which apply Price Waterhouse to ADEA mixed-motive age claims or mixed-motive retaliation claims. But see Fast v. S. Union Co., 149 F.3d 885, 889 (8th Cir. 1998) (applying the "motivating factor" standard enacted in the 1991 CRA). For Title VII retaliation claims, see, for example, Matima v. Celli, 228 F.3d 68, 80-81 (2d Cir. 2000), Kubicko v. Ogden Logistics Seroices, 181 F.3d 544, 553 n.8 (4th Cir. 1999), Medlock v. Ortho Biotech Inc., 164 F.3d 545, 549-51 (10th Cir. 1999).

(176) See, e.g., Parker v. Columbia Pictures Indus., 204 F.3d 326, 337 (2d Cir. 2000) (requiring a showing that disability was a motivating factor in a decision); Foster v. Arthur Andersen, LLP, 168 F.3d 1029, 1033-34 (7th Cir. 1999) (same); Baird v. Rose, 192 F.3d 462, 468-70 (4th Cir. 1999) (same). But see Garcia v. S.U.N.Y. Health Scis. Ctr., 280 F.3d 98, 112 (2d Cir. 2001) (suggesting, without clearly deciding, that the analysis might be governed by Price Waterhouse); McNely v. Ocala Star-Banner Corp., 99 F.3d 1068, 1076 (11th Cir. 1996). There is also some variance regarding whether mixed-motive claims may be brought under the ADA at all. See Macy v. Hopkins County Sch. Bd., 484 F.3d 357, 363 n.2 (6th Cir. 2007) (discussing this issue and noting that all circuits but the 6th permit mixed-motive claims).

(177) Parker, 204 F.3d at 337.

(178) See Sista v. CDC Ixis N. Am., Inc., 445 F.3d 161, 174 (2d Cir. 2006).

(179) See, e.g., Woodson v. Scott Paper Co., 109 F.3d 913, 934 (3d Cir. 1997).

(180) Id.

(181) Id.

(182) Id. at 934 n.25 (quoting H.R. REP. No. 104-20, pt. 1, at 47 (1991), reprinted in 1991 U.S.C.C.A.N. 549, 585) (finding the House Report unpersuasive because it referred to other "statutes" rather than other parts of Title VII).

(183) Id. at 934.

(184) 429 U.S. 125 (1976).

(185) 417 U.S. 484 (1974).

(186) 429 U.S. at 135 (quoting Geduldig, 417 U.S. at 496-97).

(187) Id. at 139.

(188) Id. at 161-62 (Stevens, J., dissenting).

(189) Id. at 148-49 (Brennan, J., dissenting) (quoting Geduldig, 417 U.S. at 496 n.20) (citations omitted).

(190) Id. at 146-47.

(191) Pub. L. No. 95-555, 92 Star. 2076 (codified at 42 U.S.C. [section] 2000e(k) (2000)).

(192) [section] 1, 92 Stat. at 2076.

(193) H.R. REP. No. 95-948, at 2 (1978), reprinted in 1978 U.S.C.C.A.N. 4749, 4750.

(194) Id.

(195) S. REP. No. 95-331, at 2-3 (1977).

(196) Newport News Shipbuilding & Dry Dock Co. v. EEOC, 462 U.S. 669, 676 (1983) (citation omitted).

(197) Id. at 678-79 & n.17 (cataloguing references in the PDA's legislative history to the Gilbert dissents' reasoning and to a desire to "restore" the law).

(198) See supra notes 182-98 and accompanying text.

(199) Martinez v. N.B.C. Inc., 49 F. Supp. 2d 305, 309 (S.D.N.Y. 1999).

(200) Wallace v. Pyro Mining Co., 789 F. Supp. 867, 869 (W.D. Ky. 1990), aff'd, 951 F.2d 351 (6th Cir. 1991) (unpublished table decision). The Sixth Circuit decision neither approves nor disapproves of the district court's analysis regarding Gilbert.

(201) Wallace, 789 F. Supp. at 869.

(202) Id. at 870; see also, e.g., Fejes v. Gilpin Ventures, Inc., 960 F. Supp. 1487, 1492 (D. Colo. 1997) (finding that breastfeeding is not a "medical condition" under the PDA).

(203) 479 F.3d 936 (8th Cir. 2007). Decisions by district courts considering the issue are split. See id. at 940 n.1 (providing an overview of recent district court decisions regarding whether Title VII "requires companies to provide coverage of contraception").

(204) Id. at 942.

(205) See id. at 943-45.

(206) Id. at 944-45 & n.5.

(207) See, e.g., Cummins v. Illinois, No. 2002-cv-4201-JPG, 2005 U.S. Dist. LEXIS 42634, at *12-14 (S.D. Ill. Aug. 30, 2005).

(208) Union Pacific, 479 F.3d at 945-49 (Bye, J., dissenting).

(209) 141 F. Supp. 2d 1266, 1270 (W.D. Wash. 2001).

(210) Id.

(211) Id. at 1271.

(212) See U.S. Equal Employment Opportunity Comm'n, Decision on Coverage of Contraception (Dec. 14, 2000), http://www.eeoc.gov/policy/docs/decision-contraception.html.

(213) 128 S. Ct. 1951 (2008).

(214) 491 u.s. 164 (1989).

(215) 42 U.S.C. [section] 1981(a) (2000).

(216) 491 U.S. at 180-181. The Court also sua sponte requested that the parties brief whether a prior decision, holding that [section] 1981 was applicable to private actors, should be reconsidered but ended up upholding the prior decision on stare decisis grounds. See id. at 172-75.

(217) Id. at 180.

(218) Id. at 189 (Brennan, J., dissenting).

(219) Id. at 206.

(220) Id. at 206, 211.

(221) Civil Rights Act of 1991 [section] 101, 42 U.S.C. [section] 1981(b) (2000). This language was found not only in S. 1745, 102d Cong. (1991) (enacted), the bill that was ultimately enacted, but also in H.R. 1, 102d Cong. (1991), the Democratic-sponsored bill that passed the House.

(222) 137 CONG. REC. 29,045-47 (1991) (memorandum introduced by Sen. Danforth).

(223) H.R. REP. No. 102-40, pt. 1, at 92 (1991), reprinted in 1991 U.S.C.C.A.N. 549, 630.

(224) CBOCS West, Inc. v. Humphries, 128 S. Ct. 1951, 1954 (2008).

(225) See 29 U.S.C. [section] 623(d) (2006) (providing a separate antiretaliation provision in the ADEA); 42 U.S.C. [section] 2000e-3(a) (2000) (providing a separate antiretaliation provision within Title VII); id. [section] 12203(a)-(b) (providing a separate antiretaliation provision in the ADA); Burlington N. & Santa Fe Ry. Co. v. White, 548 U.S. 53, 63 (2006) ("There is strong reason to believe that Congress intended the differences that [separate retaliation and discrimination statutes] suggests....").

(226) CBOCS, 128 S. Ct. at 1961.

(227) Id. at 1958.

(228) Id. at 1955.

(229) Id. at 1965 (Thomas, J., dissenting).

(230) 396 U.S. 229 (1969). The Court also referenced its more recent decision in Jackson v. Birmingham Board of Education, 544 U.S. 167, 176-77 (2005), which relied upon Sullivan to find that Title IX also permits retaliation claims not withstanding its lack of explicit language regarding retaliation. See CBOCS, 128 S. Ct. at 1955, 1957-59.

(231) CBOCS, 128 S. Ct. at 1955-56.

(232) See id. at 1956 (citing Choudhury v. Polytechnic Inst. of N.Y., 735 F.2d 38, 42-43 (2d Cir. 1984)); Goff v. Cont'l Oil Co., 678 F.2d 593, 598-99 (5th Cir. 1982), overruled by Carter v. S. Cent. Bell, 912 F.2d 832 (5th Cir. 1990); Winston v. Lear-Siegler, Inc., 558 F.2d 1266, 1270 (6th Cir. 1977)).

(233) CBOCS, 128 S. Ct. at 1955-56 (citing Walker v. S. Cent. Bell Tel. Co., 904 F.2d 275, 276 (5th Cir. 1990) (per curiam); Sherman v. Burke Contracting, Inc., 891 F.2d 1527, 1534-35 (11th Cir. 1990) (per curiam); Overby v. Chevron USA, Inc., 884 F.2d 470, 473 (9th Cir. 1989).

(234) Id. at 1959.

(235) Id. at 1957; see also supra note 223 and accompanying text (referencing a House Report which stated the 1991 amendment was intended to reinstate retaliation claims).

(236) See supra notes 152-56 and accompanying text.

(237) See supra notes 98-100 and accompanying text; see also Jones v. R.R. Donnelley & Sons Co., 541 U.S. 369, 377-83 (2004) (holding for purposes of a federal statute of limitations that a claim under [section] 1981 concerning postformation conduct is brought pursuant to legislation "enacted" after 1990).

(238) 128 S. Ct. at 1954-58.

(239) Id. at 1963 (Thomas, J., dissenting).

(240) Humphries v. CBOCS West, Inc., 474 F.3d 387, 409 (7th Cir. 2007) (Easterbrook, C.J., dissenting).

(241) Id.

(242) See BARNES, supra note 4, at 169, 171.

(243) See supra note 78 and accompanying text.

(244) See, e.g., SCALIA, supra note 86.

(245) See, e.g., Marshall, supra note 27, at 211-15.

(246) There might be some constitutional question regarding whether Congress could prospectively dictate future judicial interpretation, see infra note 258, but such statements would certainly help signal congressional intent.

(247) See, e.g., Ledbetter v. Goodyear Tire & Rubber Co., 127 S. Ct. 2162, 2169 n.2 (2007).

(248) See generally James J. Brudney & Corey Ditslear, The Decline and Fall of Legislative History? Patterns of Supreme Court Reliance in the Burger and Rehnquist Eras, 89 JUDICATURE 220 (2006) (documenting declining reliance on legislative history in Supreme Court jurisprudence); see also supra note 86 (describing the debate regarding the extent to which legislative history should direct statutory interpretation).

(249) To the extent that courts thus thwart congressional will, such decisions may be ripe for override, a suggestion in accordance with empirical studies finding that opinions that rely exclusively on "plain meaning" of the text are more likely to be overridden. See, e.g., Bussel, supra note 52, at 910 (focusing on bankruptcy decisions exclusively); Eskridge, supra note 2, at 348.

(250) See, e.g., Brudney, supra note 5, at 11-16 (discussing the override of a narrow interpretation of an override in the ADEA). In 2008, Congress considered an override of the Ledbetter decision. The bill easily passed the House but fell subject to a filibuster in the Senate. See supra note 13. As Kathryn Eidmann points out, however, the proposed override was written to address only pay discrimination and thus might have failed to end reliance on Ledbetter as a precedent in other contexts. See Kathryn A. Eidmann, Comment, Ledbetter in Congress: The Limits of a Narrow Legislative Override, 117 YALE L.J. 971, 973 (2008). In other words, even if a comparable bill were to pass, it likely would leave Ledbetter standing as a shadow precedent.

(251) Prenkert, supra note 163, at 255. In a different article, Prenkert advocates that Congress amend all "disparate treatment" statutes to make clear that the 1991 CRA's mixed-motive standard applies. See Jamie Darin Prenkert, The Role of Second-Order Uniformity in Disparate Treatment Law: McDonnell Douglas's Longevity and the Mixed-Motive Mess, 45 AM. Bus. L.J. 511, 559-61 (2008).

(252) See, e.g., Huntington Branch, NAACP v. Town of Huntington, 844 F.2d 926, 935-36 (2d Cir. 1988) (recognizing that Title VII and Title VIII, the Fair Housing Act, are typically interpreted similarly).

(253) ,See, e.g., Hayut v. State Univ. of N.Y., 352 F.3d 733, 744 (2d Cir. 2003) (stating that hostile environment claims under Title IX "are governed by traditional Title VII 'hostile environment' jurisprudence").

(254) Cf. Brudney, supra note 5, at 20-40 (discussing such opportunity costs).

(255) See supra notes 184-97 and accompanying text.

(256) See supra notes 196-97 and accompanying text.

(257) 42 U.S.C. [section] 2000e(k) (2000).

(258) See generally Nicholas Quinn Rosenkranz, Federal Rules of Statutory Interpretation, 115 HARV. L. REV. 2085 (2002) (advocating that Congress adopt principles of statutory interpretation and addressing potential constitutional and jurisprudential objections to the suggestion). But see, e.g., Larry Alexander & Saikrishna Prakash, Mother May I? Imposing Mandatory Prospective Rules of Statutory Interpretation, 20 CONST. COMMENT. 97 (2003) (questioning whether Congress has the power to dictate courts' prospective interpretations). There are similar debates regarding whether Congress has the power to abrogate stare decisis with respect to constitutional decisions. Compare, e.g., Michael Stokes Paulsen, Abrogating Stare Decisis by Statute: May Congress Remove the Precedential Effect of Roe and Casey?, 109 YALE L.J. 1535 (2000) (arguing that it may), with Gary Lawson, Controlling Precedent: Congressional Regulation of Judicial Decision-Making, 18 CONST. COMMENT. 191 (2001) (arguing that it may not).

(259) See supra Parts III.B & III.C.

(260) See supra note 61 and accompanying text; cf. Mid-Con Freight Sys., Inc. v. Mich.

Pub. Serv. Comm'n, 545 U.S. 440, 468 (2005) (Kennedy, J., dissenting) ("Instead of heeding what Congress actually said, the Court relies on ... dubious inferences from legislative silence to impose the Court's view....").

(261) See supra note 61 and accompanying text.

(262) See supra notes 179-83.

(263) This is similar to a proposal made by Jamie Prenkert specifically with respect to ending reliance on Wards Cove in other statutes. See Prenkert, supra note 163, at 263.

(264) To some extent, this proposal could be said to rest on the interpretive fiction that the sitting Congress can "know" the intentions of the enacting Congress. Cf. William W. Buzbee, The One-Congress Fiction in Statutory Interpretation, 149 U. PA. L. REV. 171, 171 (2000) (discussing judges' tendency to imagine that Congress expresses itself in an "authoritative way" when engaging in statutory interpretation, even "where statutes were enacted by different sessions of Congress, arose out of different contexts, and concerned different subjects"). Alternatively, the proposal could rest on an understanding that the sitting Congress, by enacting an override, can reinterpret the preexisting language. Significantly, however, since the enacting Congress no longer exists, the question is whether the interpretation of the preexisting language is presumptively governed by a judicial interpretation (that the sitting Congress has at least partially repudiated) or by the sitting Congress. On balance, given that choice and the general principles of legislative supremacy, it seems appropriate to follow legislative signals rather than judicial signals.

(265) See, e.g., McNally v. United States, 483 U.S. 350, 359-60 (1987) (" [W] hen there are two rational readings of a criminal statute, one harsher than the other, we are to choose the harsher only when Congress has spoken in a clear and definite language.").

(266) See, e.g., 42 U.S.C. [section] 2000e-2(k) (2000) (partially codifying and partially overriding Wards Cove).

(267) 15 U.S.C. [section][section] 1-7 (2006).

(268) Analysis of such legislative history should be sensitive to the dangers implicit in consideration of post-enactment legislative history. See generally Brudney, supra note 5.

(269) This is similar to concerns frequently leveled regarding reliance on legislative history. See, e.g., SCALIA, supra note 86, at 34-35 (asserting that committee reports cannot be assumed to represent the will of both houses). But cf. Nourse & Schacter, supra note 23, at 605-10 (presenting an empirical study which suggests checks built into congressional processes mitigate this concern).

(270) of course, there is some reason to believe that the Supreme Court would never adopt the more moderate "rebuttable" presumption rule either. However, given its espoused commitment to legislative supremacy, and that policy ramifications of the general principle would shift over time according to the make-up of the courts and Congress, this seems at least plausible. See supra notes 20-21 and accompanying text.

(271) This would be subject to the same limitations regarding constitutional interpretations and aspects of the precedent unrelated to the override.

(272) Patterson v. McLean Credit Union, 491 U.S. 164, 173 (1989).

(273) See Chevron U.S.A. Inc. v. Natural Res. Def. Council, 467 U.S. 837, 842, 844 (1984). Regulations issued after notice-and-comment rulemaking typically receive full Chevron deference; other forms of agency guidance, that are not intended to speak with "force of law," do not receive Chevron deference but may nonetheless be deemed persuasive. See United States v. Mead Corp., 533 U.S. 218 (2001).

(274) This is possible even though the statutory language of the override technically changes what an "unlawful practice" is since this implicitly relates back to conduct "because of" a protected class. Cf. Katz, supra note 163, at 664; Prenkert, supra note 163, at 263-67 (making similar arguments).

(275) See supra notes 164-69 and accompanying text.

(276) I am not contending that this understanding of Title VII would necessarily require employers to make accommodations, such as additional break time, for an employee who sought to breastfeed or express breast milk, but solely that it would prohibit discrimination (that is, punitive conduct) against an employee for choosing to breastfeed.

(277) The majority opinion in Ledbetter, seeking to minimize the significance of Congress' override, states, "Evans and Ricks, on which Lorance relied, and which employed identical reasoning, were left in place, and these decisions are more than sufficient to support our holding today." Ledbetter v. Goodyear Tire & Rubber Co., 127 S. Ct. 2162, 2169 n.2 (2007) (citation omitted). The question remains, however, are they really?

(278) 536 U.S. 101 (2002).

(279) Bazemore v. Friday, 478 U.S. 385, 395-96 (1986).

(280) Ledbetter, 127 S. Ct. at 2173.

(281) See supra text accompanying notes 10-17. It is beyond the scope of this project to document the existence of shadow precedents in other areas of the law. However, informal conversations with colleagues with expertise in other areas of the law suggest that they routinely arise in other contexts as well.

(282) If reliance on shadow precedents were unusual in other areas of the law (or in other, less partisan, areas of the law), it could suggest that the practice is the result of judicial activism rather than a good faith misreading of congressional signals. On the other hand, it could also suggest that Congress passes clearer overrides in other contexts. If this were shown to be the case, then it would be more appropriate to place a greater onus on Congress to be similarly clear--again in statutory text--in overrides in the employment discrimination context.

(283) Cf. text accompanying supra note 69 (discussing the empirical finding that judicial dissensus regarding the scope of the meaning of overrides is correlated with the extent to which judges' rulings in the area tended to be partisan).

(284) Cf. Nourse & Schacter, supra note 23, at 604 (finding that congressional staff members involved in legislative drafting have a general awareness of the courts' statutory interpretation doctrines even though political considerations often play a larger role in drafting decisions).

(285) To the extent that Congress could prescribe prospective rules to govern the interpretation of a particular override or a class of overrides, this concern would obviously be less relevant. See supra notes 246, 258 (discussing the constitutional debate regarding Congress' power to dictate judicial interpretation).

(286) For example, in enacting the 1991 CRA, Congress was clear that it disagreed with Wards Cove but not whether the precedent Wards Cove itself relied upon was overridden. See Civil Rights Act of 1991 [section][section] 2-3, 42 U.S.C. [section] 1981 (2000). Courts have therefore struggled with how to apply the new standard. See Lanning v. Se. Pa. Transp. Auth., 181 F.3d 478, 488-90 (3d Cir. 1999) (reviewing pre-Wards Cove decisions and the ambiguity of the 1991 CRA regarding how to interpret these decisions); see also Earl M. Maltz, The Legacy of Griggs v. Duke Power Co. : A Case Study in the Impact of a Modernist Statutory Precedent, 1994 UTAH L. REV. 1353, 1370-71 (discussing confusion among lower courts regarding the status of pre-Wards Cove case law after the enactment of the 1991 CRA).

(287) See, e.g., Samuel R. Bagenstos, The Americans with Disabilities Act As Welfare Reform, 44 WM. & MARY L. REV. 921, 933-36 (2003) (collecting such commentary). Bagenstos himself, however, argues that the judicial interpretations were not inconsistent with the enacting Congress' focus on the ADA as an aspect of welfare reform. See id. at 953-85.

(288) ADA Amendments Act of 2008 [section][section] 2(a) (4)-(5), 42 U.S.C.A. [section] 12101 note (West 2005 & Supp. 2008).

(289) Id. [section] 4(a), 42 U.S.C.A. [section] 12102 (West 2005 & Supp. 2008).

(290) Id. [section] 2(b), 42 U.S.C.A. note (West 2005 & Supp. 2008).

[c] 2009 Deborah A. Widiss. Individuals and nonprofit institutions may reproduce and distribute copies of this Article in any format, at or below cost, for educational purposes, so long as each copy identifies the author, provides a citation to the Notre Dame Law Review, and includes this provision and copyright notice.

Deborah A. Widiss, Visiting Assistant Professor, Brooklyn Law School. J.D., Yale Law School; B.A., Yale University. I am grateful to Stephen Befort, Martha Chamallas, Drew Days, Elizabeth Emens, Bill Eskridge, Matthew Finkin, Timothy Glynn, Tristin Green, Jeffrey Hirsch, Ted Janger, Minna Kotkin, Anita Krishnakumar, Jerry Mashaw, Leticia Saucedo, Elizabeth Schneider, Reva Siegel, Larry Solan, Lior Strahilevitz, Charles Sullivan, Nelson Tebbe, Michelle Travis, Michael Waterstone, and Michael Zimmer, for their thoughtful comments and encouragement on earlier drafts. I am also grateful to the organizers and participants in the Scholarly Papers panel at the 2009 Association of American Law Schools' Annual Conference; 2008 Seton Hall Labor and Employment Law Scholars' Forum; and the Brooklyn Law School Junior Faculty Workshop for the opportunity to present and discuss this project. This project was supported by the Brooklyn Law School Dean's Summer Research Stipend Program, with particular thanks to Dean Joan Wexler. Excellent research assistance was provided by Amanda Lutrell. My thanks, as well, to the editors at the Notre Dame Law Review for their extremely conscientious work.
COPYRIGHT 2009 University of Notre Dame Law School
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Widiss, Deborah A.
Publication:Notre Dame Law Review
Date:Jan 1, 2009
Words:31418
Next Article:Congress' new infrastructural model of medical privacy.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters