The forum decides the case

Most people who believe they were fired, demoted, or harassed for an unlawful reason ask what the law forbids. The question that decides the dispute is usually different: where the dispute will be heard, what proof standard will apply, and whether the forum permits the claim at all. In modern American employment law the substantive standard matters less than the forum, because arbitration agreements, class certification requirements, and immunity doctrines determine whether the substantive standard is ever applied. A reader who tracks only the discrimination doctrine is watching the wrong variable.

This article owns the cross-statute procedural questions and leaves the substantive holdings to the statute articles. The substantive holdings behind these procedural rules are treated in the statute articles themselves; what follows here is organized around four lines of decisions. The first is the causation patchwork: status discrimination under the civil rights title may be shown on a motivating-factor standard because Congress wrote that standard into the statute in 1991, retaliation under the same title requires but-for causation after a 2013 decision because the retaliation provision was never amended the same way, age claims require but-for causation after a 2009 decision for the same reason, and disability claims sit in a genuinely contested position. Four statutes, three standards, one explanation. The second is the arbitration line: the arbitration act has been held to reach most employment contracts, and agreements requiring individual arbitration were, at this article’s 2015 reference date, the subject of an unresolved fight over class and collective waivers, which is why a large share of employment claims never reaches a court at all. The third is the class action line: a 2011 decision on commonality decertified the largest employment class ever certified and reshaped what plaintiffs must show to aggregate claims, compounding the arbitration line’s effect. The fourth is the sovereign immunity patchwork: state employees may sue their employer for damages under some federal employment statutes and not others, depending on whether the Court has found the statute a valid exercise of the enforcement power, producing a map in which the answer differs by statute and, under the leave statute, by which kind of leave was taken.

The four lines interact rather than merely coexisting. The arbitration line and the class action line compound each other: arbitration agreements remove claims from court, and the certification standard raises the bar for the claims that remain, so the two doctrines together narrow collective redress more than either would alone. The causation line and the immunity line interact differently: a state employee’s ADEA claim faces both but-for causation and no damages remedy against the state, a double procedural discount that makes the claim’s expected value vanishingly small regardless of its substantive strength. Reading the lines together, rather than as isolated doctrines, is what turns the table from a reference into an analysis.

The reference date for every statement of law in this article is April 15, 2015. Decisions issued after that date are identified by their explicit dates and presented as later developments, not as the law described in the main text.

Procedure prices the right. A statutory protection that cannot be enforced in an accessible forum, within the limitations period, under a provable causation standard, and against a defendant subject to suit, has a practical value near zero no matter how strongly the statute phrases the prohibition. That is the sense in which litigation is the stage where a statutory right is priced: the substantive sections of the employment statutes announce the rights, and the procedural architecture built by the decisions in this article determines what those rights are worth in practice. The four-statute table near the end of the article collects the pricing variables in one place, so a reader can compare any two statutes across causation, exhaustion, timing, damages, forum, and immunity without reconstructing the analysis from the case law.

The economics explain why the forum variable dominates. Litigation is expensive, slow, and uncertain, and the expected value of a claim equals its probability of success multiplied by its likely recovery minus its cost. Arbitration agreements change the cost and recovery terms; certification standards change whether costs can be shared across claimants; immunity doctrines change whether recovery is possible at all; and causation standards change the probability of success. Each of the four lines of decisions in this article moves one of those terms, which is why the procedural questions deserve an article of their own rather than a footnote in each statute’s article.

A note on scope is necessary before the analysis begins. This article describes the procedural systems Congress and the courts built; it does not evaluate any reader’s situation, predict any outcome, or recommend any course of action for a specific dispute. The standards vary by circuit on contested questions, deadlines run from events that readers may not recognize as triggering them, and the table’s cells change as later decisions issue. The map that follows is a guide to the system, not counsel for navigating an individual case. State-law protections, which often exceed the federal floor, lie outside the article’s scope; the map describes the federal architecture only. And the article takes no position on the policy disputes the cases provoked, presenting each majority and dissent on its own terms and the empirical record with its caveats, so readers can draw their own conclusions about the system the decisions built.

This article sits in the series’ litigation cluster, alongside the statute articles that own the substantive holdings and the enforcement articles that own the agency mechanics. The cluster structure reflects a deliberate anti-cannibalization rule: each article owns its questions, and this one owns the questions that cut across statutes. A reader who wants the full history of the Age Discrimination in Employment Act will find it in that statute’s article; a reader who wants to know why age claims use a different causation standard than race claims, or why state employees can sue for race discrimination but not age discrimination, needs the cross-statute comparison only this article provides. Its distinctive contribution is the cross-statute comparison: no single statute’s article can show why retaliation uses a different causation standard than status discrimination, or why the immunity answer changes between the ADEA and Title VII, because those answers emerge only from reading the statutes against each other.

The causation patchwork

Employment discrimination law does not have one causation standard. It has several, and the differences are not the product of judicial whim. They are the product of what Congress wrote into each statute and what Congress left out. The most useful thing a reader can take from this article is the ability to name the standard that applies under each major statute and to explain why the standards differ, because the explanation is the same in every case: Congress added an explicit motivating-factor standard to one provision of one statute in 1991 and did not add it anywhere else, and the Supreme Court has enforced that textual boundary with precision.

The 1991 amendment and the motivating-factor standard

The baseline is the civil rights title’s status-discrimination provision. Title VII of the Civil Rights Act of 1964 makes it unlawful for an employer to discriminate against an individual with respect to compensation, terms, conditions, or privileges of employment because of race, color, religion, sex, or national origin, in the language of 42 U.S.C. 2000e-2(a). For the first twenty-five years of the statute’s life, the courts argued over what a plaintiff had to show when an employer’s decision was driven by both a forbidden reason and a lawful one, the mixed-motive case. The Supreme Court’s 1989 decision in Price Waterhouse v. Hopkins answered the question but in a fractured set of opinions that left the lower courts uncertain about the governing framework. The case concerned Ann Hopkins, a senior manager denied partnership at an accounting firm after partners described her as needing charm school and advised her to walk, talk, and dress more femininely. A plurality held that once a plaintiff shows the forbidden characteristic played a motivating part in the decision, the burden of persuasion shifts to the employer to prove it would have made the same decision regardless. A concurrence would have required a stricter showing before shifting the burden. The dissent would have kept the burden on the plaintiff throughout. With no majority rationale, the lower courts spent two years improvising, and Congress’s 1991 intervention replaced the improvisation with a statute.

The 1991 Act itself was a compromise shaped by failure. A broader civil rights bill passed Congress in 1990 and was vetoed; the 1991 version that became law narrowed several provisions to secure the votes for enactment and for overriding a second veto. The motivating-factor section survived that negotiation. Section 107(a) of the Act added a new subsection, codified at 42 U.S.C. 2000e-2(m), to the status-discrimination section of Title VII: an unlawful employment practice is established when the complaining party demonstrates that race, color, religion, sex, or national origin was a motivating factor for any employment practice, even though other factors also motivated the practice. The same section amended the remedies provision, 42 U.S.C. 2000e-5(g), to add a limited affirmative defense: when the employer demonstrates that it would have taken the same action in the absence of the impermissible motivating factor, the plaintiff still establishes liability but may receive only declaratory relief, injunctive relief, and attorney’s fees, with no damages and no reinstatement. The 1991 amendments are examined in detail in the companion guide to the Civil Rights Act of 1991 amendments.

Congress deliberately placed this provision in the status-discrimination section of the title. What Congress did not do, in the same Act, was add a motivating-factor provision anywhere else: not to the retaliation section of Title VII, not to the age statute, and not to the disability statute. That omission became the foundation of two 5-4 decisions that split the causation map.

The standard matters because most employment decisions have more than one input. A manager may dislike a worker’s protected trait and also dislike the worker’s attendance record. Under the motivating-factor test, the plaintiff does not need to isolate the trait as the decisive input; showing it was one motivating input suffices for liability. The employer’s same-decision showing then caps the remedy rather than defeating the claim. A worked example shows how the defense operates. Suppose a jury finds that a manager’s bias against a worker’s religion was a motivating factor in a termination, but the employer proves through attendance records and documented warnings that the worker would have been fired for absenteeism alone. Liability is established, the court may declare the violation and enjoin future discrimination, and the worker’s lawyer recovers fees, but the worker receives no damages and no reinstatement. The defense thus functions as a damages cutoff rather than a liability shield, which is why employers invest heavily in documenting legitimate reasons contemporaneously: the paper trail built before litigation becomes the same-decision proof during it.

One more decision governs how the motivating-factor standard reaches the jury. In Desert Palace, Inc. v. Costa, decided in 2003, the Court held unanimously that a plaintiff need not present direct evidence of discrimination, such as an explicit statement of bias, to obtain a mixed-motive jury instruction under 42 U.S.C. 2000e-2(m); circumstantial evidence suffices. The decision matters because most discrimination cases are built on circumstantial evidence, and a direct-evidence requirement would have confined the 1991 standard to the rare case with a smoking gun.

What is the motivating-factor standard?

Under the motivating-factor standard, a plaintiff prevails by proving that race, color, religion, sex, or national origin was a motivating factor in the employment decision, even if other lawful factors also motivated it. If the employer proves it would have made the same decision anyway, the plaintiff keeps liability but receives only declaratory relief, an injunction, and attorney’s fees.

The McDonnell Douglas framework

McDonnell Douglas Corp. v. Green (1973) supplies the background framework for most disparate-treatment cases that lack direct evidence of bias. The framework does not change the causation standard; it structures how parties prove their positions under whichever standard applies. Under the motivating-factor regime, a plaintiff can also proceed with direct evidence that the trait was a motivating factor, bypassing the framework’s steps. Under the but-for regime of Gross and Nassar, the pretext stage carries the full weight: the plaintiff must persuade the factfinder that the employer’s stated reason is false and that the forbidden consideration was the reason the action would not otherwise have occurred. The interaction explains why the causation standard matters most at summary judgment, where judges ask whether a reasonable jury could find the required causal link on the paper record.

The prima facie case itself is deliberately undemanding: the plaintiff shows membership in a protected group, qualification for the position, an adverse action, and circumstances suggesting discrimination, such as the position remaining open or a similarly situated worker outside the group receiving better treatment. The employer’s burden at the second step is equally light, requiring only articulation of a legitimate reason, not proof of it. The weight falls on the third step, where the plaintiff must produce evidence that the reason is pretextual: shifting explanations, comparator evidence, procedural irregularities, or remarks suggesting bias. Because most cases lack direct evidence, the pretext battle on the paper record is where employment litigation is typically won or lost.

How does the McDonnell Douglas framework allocate proof?

The framework assigns three sequential burdens. The plaintiff first shows a basic case of discrimination, such as membership in a protected group and an adverse action under suspicious circumstances. The employer must then articulate a legitimate nondiscriminatory reason. The plaintiff finally must prove that reason is a pretext for discrimination.

Nassar and retaliation: the textual gap

University of Texas Southwestern Medical Center v. Nassar, 570 U.S. 338 (2013), asked what causation standard governs Title VII retaliation claims. The plaintiff, a physician of Middle Eastern descent, alleged that the medical center withdrew an employment offer after he complained of harassment. The Court, 5-4, held that retaliation claims under 42 U.S.C. 2000e-3(a) require proof of traditional but-for causation: the plaintiff must show that the adverse action would not have occurred in the absence of the employer’s desire to retaliate.

Justice Kennedy’s majority opinion rested on the text. The 1991 motivating-factor amendment lists race, color, religion, sex, and national origin; it does not list retaliation. The retaliation provision forbids discrimination because an employee opposed an unlawful practice or participated in proceedings, and the majority read that word because as incorporating common-law but-for causation, mirroring its earlier reading of because of age in the age-discrimination context. Congress, the majority reasoned, deliberately omitted retaliation from the 1991 amendment, and courts should respect the omission rather than repair it. The majority treated the retaliation provision’s because language as carrying the common law’s default meaning of but-for causation unless context indicates otherwise, and read the 1991 amendment’s list of covered traits as meaningful rather than inadvertent. Pre-Nassar, several circuits had already concluded that the 1991 amendment did not reach retaliation claims, so the decision confirmed the direction lower courts had taken. Justice Ginsburg’s dissent, joined by three colleagues, argued that retaliation is functionally inseparable from status discrimination: workers who complain about discrimination often face reprisal precisely because of their protected status, and Congress legislated against a background in which retaliation and discrimination claims were litigated together. The dissent disputed the majority’s textual inference, arguing that the word because does not compel but-for causation as a matter of ordinary usage and that the majority’s reading weakened the enforcement scheme Congress designed. The dissent’s position did not prevail.

The decision’s consequences play out in jury instructions and summary judgment. Trial judges instruct juries that retaliation plaintiffs must prove the protected complaint was the but-for cause of the adverse action, a harder instruction to satisfy than the motivating-factor charge given in status cases. At summary judgment, employers attack the causal chain between complaint and action, emphasizing intervening legitimate reasons, while plaintiffs emphasize temporal proximity and shifting explanations. Because the same set of facts can support both a status claim and a retaliation claim, lawyers plead and prove them on separate causal tracks. A worker who complains about race discrimination and is then fired will typically allege both that the firing was motivated by race and that it was retaliation for the complaint. The status claim goes to the jury on the motivating-factor instruction; the retaliation claim goes on the but-for instruction. Identical testimony about the manager’s motives can thus yield a split verdict, with the jury finding race was a motivating factor but not the but-for cause. The two-track structure is a permanent feature of federal employment litigation.

The asymmetry is easy to misread as a judgment about the seriousness of retaliation. The opinions themselves frame it as a question of statutory architecture: Congress wrote one standard into one section and not into another, and the Court declined to transplant it. The same set of facts can support both a status claim and a retaliation claim, and a plaintiff can win the status claim while losing the retaliation claim on identical evidence, because the status claim goes to the jury on the motivating-factor instruction while the retaliation claim goes on the but-for instruction.

Why does retaliation use a different standard than status discrimination?

Retaliation requires but-for causation because the 1991 amendments added the motivating-factor provision to the status discrimination section of Title VII while leaving the retaliation provision untouched. The Court read that choice in 2013 as deliberate, so a retaliation plaintiff must prove the adverse action would not have occurred without the protected complaint.

Gross and the ADEA: but-for causation for age claims

Gross v. FBL Financial Services, Inc., 557 U.S. 167 (2009), presented the same textual question under the Age Discrimination in Employment Act of 1967. The plaintiff alleged that his reassignment was motivated in part by age. The Court, 5-4, held that an ADEA disparate-treatment plaintiff must prove by a preponderance of the evidence that age was the but-for cause of the challenged action. The burden of persuasion never shifts to the employer, and the mixed-motive burden-shifting framework associated with Price Waterhouse v. Hopkins does not apply to age claims. Justice Thomas’s majority opinion tracked the Nassar logic four years before Nassar: when Congress added the motivating-factor provision to Title VII in 1991, it did not add one to the ADEA, and courts must take that choice seriously. The dissenters, led by Justice Stevens, argued that the Price Waterhouse framework had governed mixed-motive analysis for two decades, that Congress’s silence in 1991 did not repudiate it, and that the majority’s reading made age claims materially harder to prove than race or sex claims without any indication Congress wanted that result. The debate turned on what congressional silence means. The majority treated the 1991 amendment as a deliberate choice to give Title VII plaintiffs a tool denied to ADEA plaintiffs, invoking the canon that Congress acts intentionally when it includes language in one section and omits it from another. The dissent treated the silence as inadvertent, arguing that the Price Waterhouse framework had been the shared background understanding and that Congress’s failure to disturb it for age claims left it in place. Neither side could produce legislative history settling the question, because the 1991 debates barely addressed the ADEA at all. The age statute’s full history is covered in the companion guide to the Age Discrimination in Employment Act of 1967.

The practical consequence is a proof structure with no safety net. Under the motivating-factor regime, a plaintiff who shows the forbidden trait was one input among several still establishes liability and forces the employer to prove the same-decision defense. Under Gross, an age plaintiff who shows age was one input among several, but cannot show the decision would have differed without it, loses outright. The employer need not prove anything about what it would have done; the plaintiff carries the causal question from filing through verdict.

What does but-for causation require under the ADEA?

An ADEA plaintiff must prove by a preponderance of the evidence that age was the but-for cause of the adverse employment action, meaning the action would not have happened without the worker’s age. The burden of persuasion never shifts to the employer, and the mixed-motive framework that Title VII plaintiffs may use does not apply to age claims.

The disability statute: a contested position

The Americans with Disabilities Act presents the hardest causation question because Congress has never answered it. The ADA Amendments Act of 2008 broadened the definition of disability substantially, but it did not add a motivating-factor provision to the disability statute, and the Supreme Court has never ruled on the ADA causation standard directly.

What the 2008 amendments did is worth specifying, because their silence on causation is conspicuous. Congress overturned the Court’s narrowing constructions of disability, rejecting the requirement that mitigating measures be considered and the demanding standard for what counts as substantially limiting, and it expanded the definition of major life activities. The amendments were a direct rebuke to a line of decisions that had shrunk the protected class. Yet the same Congress left the causation provision untouched, even though the amendments were enacted just a year before Gross gave the but-for reading its fullest expression. Whether the omission reflected a deliberate choice to leave causation to the courts or simply the fact that causation was not on the legislative agenda is unknowable from the record, but the consequence is the contested map described here.

The Seventh Circuit moved first. In Serwatka v. Rockwell Automation, Inc. (2010), the court overruled its own prior precedent permitting mixed-motive claims under the ADA, reasoning that Gross’s logic extended to the disability statute: Congress wrote no motivating-factor provision into the ADA, and the ADA’s because of language pointed toward but-for causation. The Sixth Circuit followed in Lewis v. Humboldt Acquisition Corp. (2012), rejecting its older sole-cause formulation and adopting but-for causation under Gross’s reasoning. A 2015 survey of circuit law described courts as having yet to commit definitively to one camp, with post-Nassar decisions in acknowledged disagreement about whether mixed-motive analysis survived under the ADA at all. The result, at the article’s 2015 reference date, was a genuine split in approach rather than a settled rule.

For litigants, the contest has concrete consequences. A disability plaintiff in the Seventh or Sixth Circuit faces a but-for instruction, while a plaintiff elsewhere may persuade a judge to give a motivating-factor instruction, making the circuit of filing a substantive variable. And because the Supreme Court has never resolved the question, every ADA causation ruling from the courts of appeals remains provisional, subject to revision by the one decision that would settle the map.

Why is ADA causation contested?

At the article’s 2015 reference date, no settled answer existed. The ADA Amendments Act of 2008 did not add motivating-factor language to the disability statute, and the Supreme Court had never ruled on ADA causation directly. The Seventh Circuit in 2010 and the Sixth Circuit in 2012 applied but-for causation, while other courts left the question open.

The 1991 amendment and the text it left alone

The Civil Rights Act of 1991 is best understood as Congress rewriting one corner of employment law while deliberately leaving the surrounding corners as it found them. The statute was, among other things, a response to a cluster of Supreme Court decisions from the late 1980s that civil rights constituencies regarded as narrowing Title VII, with Price Waterhouse v. Hopkins as the centerpiece. The mixed-motive framework the new section 107(a) created was a legislative settlement: the plaintiff’s burden would be the motivating-factor showing, and the employer’s same-decision proof would limit remedies rather than defeat liability. What matters for this article is the negative space, the provisions the amendment did not touch.

Congress knew how to write a motivating-factor standard. It had just written one. The retaliation provision, sitting one section away in the same title, uses because language that predates the 1991 Act, and the amendment’s drafters did not add the new standard to it. The age statute uses because of language that predates the 1991 Act, and the amendment’s drafters did not add the new standard to it either. The Nassar and Gross majorities treated those omissions as deliberate legislative choices rather than oversights, and the dissents treated them as gaps to be filled by purposive reading. The reader need not take a side to see the consequence: under the law described here, a worker who claims she was fired partly because of her race and partly because of her lawful opposition to discrimination faces two different causation standards in the same trial, one for each count. The dissents in both cases warned that this result would confuse juries and complicate litigation, and the majorities answered that the confusion, if any, was Congress’s to resolve. That exchange is the clearest illustration in modern employment law of the proposition that the text Congress enacted matters more than the purpose a reader might impute to it.

The same-decision defense deserves attention because it is the compromise embedded in the 1991 settlement. Under 42 U.S.C. 2000e-5(g), an employer that proves it would have taken the same action absent the forbidden motive does not escape liability; it escapes damages, reinstatement, and back pay in that measure, retaining exposure to declaratory and injunctive relief and to the plaintiff’s attorney’s fees. The structure reflects a legislative judgment that establishing a discriminatory motive has independent legal significance even when the employment decision would have occurred anyway. Readers sometimes assume that proving the employer would have acted the same way defeats the claim entirely. Under the 1991 framework it does not. It converts a damages case into a liability-only case.

The method behind the patchwork: reading what Congress left out

The causation patchwork invites a question about judicial method. How does a court justify giving the same word, because, different legal content in provisions of the same title, or giving a 1967 statute a causation rule that Congress wrote for a different statute in 1991 and pointedly did not write for it? The answer the majorities gave, in Gross and then in Nassar, is that statutory interpretation begins and often ends with what the legislature enacted, and that what the legislature declined to enact is evidence too. When Congress adds a provision to one section and leaves neighboring sections untouched, courts applying a textualist method treat the omission as a deliberate choice. The canon has a Latin name, expressio unius est exclusio alterius, the expression of one thing is the exclusion of another, but the idea is common sense: a drafter who knows how to write a motivating-factor standard, because she has just written one, and who does not write it into the next section, has made a decision the court should respect.

The dissents answered with a different theory of what legislatures do. Justice Stevens in Gross and Justice Ginsburg in Nassar each argued, in substance, that Congress legislates against the background of existing judicial frameworks and does not always reenact every provision it means to preserve. On this view, the 1991 Act’s motivating-factor amendment codified and refined a burden-shifting approach the courts had been developing since Price Waterhouse, and there was no reason to think Congress meant to strip that approach from retaliation claims or age claims by failing to mention them. Legislative silence, the dissents argued, is ambiguous: it can mean approval of the status quo, inadvertence, or a compromise that left the question for another day, and a court should not convert ambiguity into a rule that makes litigation harder to win. The two theories cannot be reconciled by closer reading of the statutes, because they are theories about how to read statutes in general, and each side in these cases accused the other of displacing Congress’s judgment with its own.

Employment law became the showcase for this methodological fight for a structural reason. The field’s major statutes were enacted decades apart, in 1963, 1964, 1967, 1990, and 1993, and amended sporadically, so the code reads like sedimentary rock, with each layer reflecting the interpretive assumptions of its era. The 1964 Congress wrote because of without imagining mixed-motive doctrine. The 1991 Congress wrote a precise mixed-motive rule for one provision. The courts must then apply the 1991 rule to the 1964 and 1967 language, deciding how far the new layer extends into the old. That is why the patchwork looks the way it does: not because anyone designed four different causation standards, but because one amendment landed in stratified text and the courts mapped its boundaries provision by provision. A reader who understands the stratification understands why the standards differ, and also why future amendments will likely produce new boundary disputes rather than clean unifications.

There is a practical coda to the methodological debate. Whichever theory of interpretation one prefers, the working lawyer and the informed reader must operate with the law as the majorities stated it: motivating factor for Title VII status claims, but-for for Title VII retaliation and ADEA claims, contested for ADA claims. Jury instructions in federal employment trials reflect exactly this division, and appellate courts reverse verdicts where the wrong standard was charged. The division reaches its most demanding form in the multi-claim trial, where a jury may be instructed on motivating factor for a status count, but-for for a retaliation count, and but-for again for an age count, all arising from the same employment decision. The dissents’ invitation to Congress remains open, and Congress has repaired Court decisions in this field before, most notably in 1991 itself, but unless and until it does, the patchwork is the law, and the explanation for the patchwork is the 1991 amendment’s limited reach.

Exhaustion and limitations: the gates before the merits

Before any forum applies a causation standard, the plaintiff must satisfy the administrative and timing requirements that Congress attached to each statute. These requirements are unforgiving, and they differ by statute in ways that trap the unwary.

The EEOC charge requirement

Title VII, the ADA, and the ADEA all require a worker to file a charge with the Equal Employment Opportunity Commission before bringing a private lawsuit. The charge must be filed within 180 days of the alleged discriminatory act, extended to 300 days in jurisdictions where a state or local antidiscrimination law exists and an agency is authorized to grant or seek relief. The extension reflects the work-sharing arrangements between the federal commission and state and local fair-employment agencies: a charge filed with one is deemed filed with the other. The trap lies in assuming the extension applies where no qualifying state law exists; in those jurisdictions the 180-day limit is absolute. Filing mechanics and agency process are covered in the companion guide to Title VII EEOC enforcement.

The Equal Pay Act is the exception. Workers alleging unequal pay for equal work may file suit directly in court without filing any agency charge, and the commission accepts voluntary charges that do not extend the court limitations period. The exception reflects the pay statute’s origins in wage-and-hour enforcement rather than the commission’s charge-processing system. The clock generally starts on the date of the discrete act, such as a firing or demotion, not on the date the worker discovers its motivation, and courts enforce these deadlines strictly, with only narrow equitable exceptions.

The charge defines the scope of the later lawsuit: courts generally hold that a plaintiff may sue only on claims that were raised in the charge or that would reasonably have grown out of the agency’s investigation of it. A charge alleging race discrimination does not preserve a later claim of sex discrimination. After a charge is filed, the commission notifies the employer, investigates, and attempts conciliation where it finds cause to believe discrimination occurred. If conciliation fails, the commission may sue in its own name or issue the right-to-sue notice that starts the worker’s clock. Most charges end without a commission lawsuit, since the agency’s docket far exceeds its litigation resources, so the right-to-sue notice is the typical outcome and the private lawsuit the typical enforcement vehicle.

The deferral mechanics add a further wrinkle. When a charge is filed with the commission in a deferral jurisdiction, the commission holds the charge and defers to the state or local agency for initial processing, typically for sixty days, before beginning its own investigation. The charge is deemed filed with both agencies simultaneously, which is what triggers the 300-day federal period. Workers sometimes misunderstand deferral as an extension of time to act; it is instead a routing rule that preserves the filing while the agencies coordinate, and the underlying deadlines do not pause during the deferral window.

What happens inside the agency matters too. After a charge is filed, the commission notifies the employer, investigates, and attempts conciliation where it finds cause to believe discrimination occurred. The employer’s response to the charge shapes what follows. Upon notification, the employer typically submits a position statement rebutting the allegations and producing supporting documents, which becomes part of the investigative file. The commission may request further information, interview witnesses, or attempt mediation through its separate mediation program. A cause finding triggers mandatory conciliation efforts before any lawsuit; a no-cause finding ends the agency’s involvement and leads to the right-to-sue notice. Either way, the paper generated during the charge stage, the worker’s allegations, the employer’s defenses, and the investigator’s notes, often becomes the opening evidentiary record in the later case.

The 90-day suit period has its own mechanics. The clock runs from receipt of the notice, and courts apply a presumption that receipt occurs a few days after mailing when the worker cannot prove otherwise. Equitable tolling is available in theory but granted sparingly: courts require active misleading by the defendant or extraordinary circumstances beyond the worker’s control, and ordinary attorney miscalculation does not qualify. The strictness reflects a legislative judgment that employment disputes should move quickly once the agency step ends, but it punishes workers who misunderstand the notice’s significance, another way the procedural architecture prices claims before the merits are reached.

The Equal Pay Act runs on its own clock. Under 29 U.S.C. 255(a), suit must be commenced within two years after the cause of action accrued, extended to three years for willful violations. No agency charge extends or tolls that period. The three-year extension for willful violations rewards proof of the employer’s mental state. Willfulness in this context means the employer knew its pay practices were unlawful or acted with reckless disregard for whether they were, a showing typically built from internal communications, ignored complaints, or a failure to investigate after being put on notice. The extra year matters because pay discrimination often emerges gradually: a worker may not recognize the disparity until years of paychecks reveal the pattern, and the willfulness extension can rescue claims the two-year limit would otherwise bar. The FMLA follows a parallel structure under 29 U.S.C. 2617(c): two years, or three for willful violations. The pattern across statutes is consistent: Congress gave workers measured time to act, courts enforce the stated periods exactly as written, and the merits of a strong claim cannot cure a missed deadline.

What is lost if the 90-day window after a right-to-sue notice closes?

For Title VII and the ADA, the plaintiff must file suit within 90 days of receiving the notice of right to sue under 42 U.S.C. 2000e-5(f)(1). After the 90-day period expires, the right to sue on that charge is lost. The ADEA uses a related but distinct rule tied to the charge date.

Federal-sector workers: a parallel track

Workers employed by the federal government travel a different procedural road, and the differences are large enough to merit separate treatment. Federal employees do not file charges with the EEOC in the first instance. Under 29 C.F.R. Part 1614, a federal worker who believes she has been discriminated against must first contact an EEO Counselor at her agency within 45 days of the discriminatory action, a deadline far shorter than the private sector’s 180 days. The counselor attempts informal resolution, ordinarily within 30 days, or the worker may elect alternative dispute resolution where the agency offers it. If the matter is not resolved, the worker may file a formal complaint with the agency, which must investigate. After the investigation, the complainant may request a hearing before an EEOC administrative judge or an immediate final decision from the agency, and only after the agency process concludes may she file suit in federal court. Age claims are the exception within the exception: a federal worker alleging age discrimination may proceed directly to court after giving the Commission notice of intent to sue.

The federal-sector track illustrates the same theme that organizes this article: procedure as the gatekeeper of substance. The 45-day counselor deadline is the shortest fuse in federal employment law, and workers who learn of it too late lose the administrative remedy regardless of the merits. The election between the EEO process, a negotiated grievance procedure, and Merit Systems Protection Board appeal rights forces early strategic choices with binding consequences. And the requirement that only claims raised in pre-complaint counseling may be alleged in the later formal complaint reproduces, in compressed form, the private sector’s rule that the charge defines the lawsuit. A federal worker and a private-sector worker with identical grievances thus face different clocks, different forums, and different sequencing, before either reaches a judge.

The continuing-violation doctrine and the Ledbetter override

The distinction matters enormously in harassment cases, where the actionable wrong often emerges from a pattern no single incident would support. Morgan also clarified that the filing period runs from each discrete act even when the acts are related, rejecting the looser continuing-violation theories some circuits had applied. Practitioners therefore calendar each adverse action separately while building hostile-environment claims around the timely anchor act.

Pay discrimination raised a special timing problem that Congress addressed directly. Because unequal pay is often discovered long after the pay decision was made, workers argued that each paycheck carrying the discriminatory rate should restart the charge-filing clock. In Ledbetter v. Goodyear Tire and Rubber Co., decided in 2007, the Court held 5-4 that the clock runs from the original pay-setting decision, not from later paychecks, a rule that made many pay claims untimely. Congress overrode the decision in the Lilly Ledbetter Fair Pay Act of 2009, the first statute signed by President Barack Obama, providing that an unlawful employment practice occurs each time compensation is paid pursuant to a discriminatory decision, so that each paycheck restarts the 180- and 300-day periods. The episode shows the exhaustion gate being adjusted by the legislature: the clock that filters claims is itself a policy choice, and Congress has shown it will reset the clock when the Court’s reading filters too aggressively.

What is the continuing-violation doctrine?

The doctrine governs when the limitations clock starts for repeated misconduct. Under National Railroad Passenger Corp. v. Morgan (2002), each discrete act such as a firing starts its own clock, while a hostile work environment claim is timely if at least one contributing act falls within the filing period.

The arbitration line

If causation determines how hard a claim is to prove, arbitration determines where the proof happens and, in practice, whether it happens at all. The Federal Arbitration Act of 1925 was written for commercial disputes between merchants. Over three decades, the Supreme Court extended it to the employment relationship, and employers responded by making arbitration agreements a standard condition of hiring. The result is a procedural channel that carries the great majority of workplace disputes.

The statute’s origin explains the controversy over its extension. Congress passed the Federal Arbitration Act in 1925 to enforce arbitration agreements between merchants of roughly equal bargaining power who preferred private dispute resolution to crowded courts. Nothing in the 1925 debates contemplated the modern employment application, in which arbitration clauses appear in take-it-or-leave-it hiring paperwork presented to individual workers. The dissenters in the employment cases returned to this history repeatedly, arguing that the Court had repurposed a commercial statute for a domain Congress never considered. The majorities answered that the Act’s text draws no distinction between commercial and employment contracts once the section 1 exemption is read narrowly, and that the judiciary’s role is to enforce the statute as written rather than to update it for changed circumstances.

Gilmer and the arbitrability of statutory claims

Gilmer v. Interstate/Johnson Lane Corp., 500 U.S. 20 (1991), established the foundational premise. The plaintiff, a securities manager, had signed a registration agreement containing an arbitration clause and later alleged age discrimination. The Court, 7-2, held that the ADEA claim could be subjected to compulsory arbitration under the FAA pursuant to the parties’ agreement. The majority reasoned that statutory claims may be arbitrated absent a congressional intent to preclude arbitration or an inherent conflict between arbitration and the statute’s purposes, and it found neither in the ADEA. The dissenters argued that the ADEA’s enforcement scheme contemplated judicial resolution and that arbitration’s limited procedures were inadequate for vindicating statutory rights. Gilmer’s premise, that agreeing to arbitrate does not surrender substantive rights but only changes the forum, became the doctrinal engine for everything that followed. Gilmer also preserved a boundary: the Court noted that an individual subject to an arbitration agreement remains free to file a charge with the EEOC, preserving the agency’s independent enforcement role.

Gilmer also planted the seed of the doctrine’s main internal limit. The majority said arbitration is permissible so long as the prospective litigant can effectively vindicate the statutory right in the arbitral forum, a reservation that later litigants invoked against arbitration terms imposing prohibitive costs or one-sided procedures. Courts have applied that reservation sparingly, striking only the most extreme terms while enforcing the mine run of agreements. The reservation’s narrow construction is one reason the arbitration line moved so steadily in one direction: the Court announced a limiting principle in Gilmer and then rarely found it satisfied.

Gilmer’s own facts show how far the doctrine traveled. The plaintiff was a securities manager whose arbitration obligation came from a stock-exchange registration agreement, a specialized industry arrangement, not a general employment contract. The Court’s reasoning, however, was not confined to the securities industry: statutory claims are arbitrable absent congressional preclusion, a principle that applied equally to any employment agreement once Circuit City brought those agreements under the FAA. The industry-specific origins of the doctrine thus gave way to a general rule covering the entire economy.

Circuit City: the FAA reaches most employment contracts

The FAA contains an exemption, in 9 U.S.C. section 1, for contracts of employment of seamen, railroad employees, or any other class of workers engaged in foreign or interstate commerce. For decades, some courts read that language to exclude employment contracts generally. Circuit City Stores, Inc. v. Adams, 532 U.S. 105 (2001), rejected that reading 5-4. Justice Kennedy’s majority held that the residual clause covers only transportation workers, applying the interpretive canon that general words following specific ones are limited to the same class. The dissents argued that Congress in 1925 legislated against a background in which arbitration of employment disputes was controversial, and the exemption’s plain terms covered employment contracts broadly. The warning did not carry the day, and after Circuit City, the question was no longer whether employment arbitration agreements were enforceable in principle but how far employers could push their terms.

The practical stakes of the interpretive dispute were enormous. Under the dissenters’ reading, employment contracts as a category would have fallen outside the FAA, leaving arbitration of workplace claims to state law and individual negotiation. Under the majority’s reading, a single federal statute governed nearly every employment arbitration agreement in the country, preempting the state-law variation that had previously limited their spread. Employers responded accordingly: once Circuit City confirmed that a standard-form clause would be enforced under federal law, arbitration provisions migrated from executive contracts into offer letters and handbooks across the economy.

Are employment arbitration agreements enforceable?

Most employment arbitration agreements are enforceable under the Federal Arbitration Act. Since a 2001 decision read the Act’s employment exemption narrowly and a 1991 decision held statutory claims arbitrable, courts routinely compel individual arbitration of workplace disputes, and the Act preempts state rules that single out arbitration agreements for disfavored treatment.

Concepcion and the savings clause

AT&T Mobility LLC v. Concepcion, 563 U.S. 333 (2011), addressed that state-law resistance. California courts had treated class-action waivers in arbitration agreements as unconscionable under state contract law. The Court, 5-4, held that the FAA preempted that rule. Justice Scalia’s majority reasoned that requiring classwide procedures interferes with the fundamental attributes of arbitration as an individualized, informal proceeding. The dissent, by Justice Breyer, argued that the FAA’s savings clause preserves generally applicable contract defenses such as unconscionability, and that the majority had elevated a judicially created policy favoring arbitration over the statute’s text. After Concepcion, state unconscionability doctrines could no longer serve as a reliable check on arbitration terms, and the enforceability question shifted to federal labor law. The savings clause still preserves the defenses that apply to contracts generally, fraud and duress among them, but after Concepcion a state rule whose practical operation is hostility to arbitration cannot survive even when phrased as a neutral contract principle.

The California rule the Court struck down had a concrete origin. In Discover Bank v. Superior Court (2005), the California Supreme Court held that class-action waivers in consumer arbitration agreements were unconscionable when the alleged wrongdoing involved small individual damages, reasoning that such waivers functioned as exculpatory clauses for widespread small-scale misconduct. The Concepcion majority treated that reasoning as exactly the kind of state policy the FAA displaces: a rule that conditions enforcement on the availability of class procedures interferes with arbitration’s fundamentally bilateral character. The underlying dispute illustrated why class procedures mattered. The plaintiffs alleged a small-dollar consumer overcharge on a cell-phone contract, the kind of claim no rational consumer would arbitrate individually given the filing costs. The majority’s response was that the FAA’s preference for bilateral arbitration does not yield to state policies favoring aggregation, however sympathetic the underlying claim.

The Federal Arbitration Act’s command that arbitration agreements are valid, irrevocable, and enforceable carries an exception, and the exception generated a second front in the arbitration wars. Section 2 of the Act preserves the agreements save upon such grounds as exist at law or in equity for the revocation of any contract. That savings clause means that generally applicable contract defenses, fraud, duress, and unconscionability among them, can invalidate an arbitration agreement just as they can invalidate any contract. The litigation of the 1990s and 2000s therefore featured a running battle over whether state unconscionability doctrine could be used to strike down employment arbitration agreements, particularly agreements imposed as a condition of employment on workers with no bargaining power and no realistic alternative.

The difficulty for challengers was that the Supreme Court distinguished between contract defenses that apply to all contracts and rules that disfavor arbitration in practice. A state rule that says arbitration agreements, and only arbitration agreements, are suspect is preempted. A state unconscionability doctrine that applies evenhandedly to all contracts survives, but only if the court applies it evenhandedly, without smuggling in a special hostility to arbitration. In practice, the distinction proved hard for challengers to navigate, because many of the features that made employment arbitration agreements feel unfair, one-sided arbitrator selection, distant forums, fee provisions, shortened limitations periods, were attacked through unconscionability, and courts applying the Supreme Court’s framework often severed the offending terms and enforced the remainder rather than voiding the agreement. The result was a body of law in which arbitration agreements were policed at the margins for the most lopsided terms but the core commitment to arbitrate was almost always preserved.

Costs supplied one of the few successful term-level challenges. Arbitration is not free: arbitrators charge daily or hourly fees, administrative fees apply, and a worker who must pay thousands of dollars to pursue a claim may be priced out of the forum altogether. In Green Tree Financial Corp. v. Randolph, decided in 2000, the Court held that large arbitration costs could preclude a litigant from effectively vindicating federal statutory rights, while placing on the party resisting arbitration the burden of showing a likelihood of prohibitive costs. Lower courts applying Randolph struck down or severed fee provisions that required workers to share arbitrator fees they could not afford, and many employers responded by drafting agreements in which the employer pays all arbitration costs. The fee question thus became one of the standard checklist items in drafting and challenging employment arbitration agreements: who pays, how much, and whether the allocation would prevent the worker from bringing the claim at all. Like unconscionability generally, the doctrine policed the margins without disturbing the core commitment to arbitrate.

The class-waiver fight in 2015: an unresolved split

The sharpest unresolved question in April 2015 concerned agreements that require individual arbitration and waive class or collective proceedings. The National Labor Relations Board took the position in D.R. Horton, Inc. (2012) that such waivers violated section 7 of the National Labor Relations Act, which protects employees’ right to engage in concerted activity for mutual aid or protection. The Board reasoned that filing a class or collective action is concerted activity, so an agreement waiving that right as a condition of employment was unlawful. Employers won the opposite result in several federal courts of appeals, which held that the FAA required enforcement of the waivers and that nothing in the labor statute displaced the arbitration act. The Fifth Circuit, in D.R. Horton, Inc. v. NLRB (2013), rejected the Board’s position and enforced the waiver, while the Board continued to adhere to its own rule in cases before it. The result was a genuine circuit split: the enforceability of a class waiver depended on where the case was filed, and no Supreme Court decision had resolved the conflict.

What was the class-waiver question in 2015?

In 2015, the answer was unresolved. The labor board held in 2012 that class-action waivers violated federal labor law’s protection of concerted activity, while employers won the opposite result in several federal appeals courts. The split meant enforceability depended on the circuit, and no Supreme Court decision had settled the conflict.

Epic Systems (2018): a post-horizon development

Readers encountering later developments should note the explicit date wall. In a decision issued May 21, 2018, after this article’s reference date, the Supreme Court decided Epic Systems Corp. v. Lewis, 584 U.S. 497. The Court, 5-4, held that employment arbitration agreements requiring individualized proceedings and waiving class and collective actions are enforceable under the FAA, that section 7 of the labor act does not render them unlawful, and that the FAA’s savings clause does not supply a contrary result. Justice Gorsuch’s majority emphasized the FAA’s instruction that arbitration agreements be enforced according to their terms; Justice Ginsburg’s dissent argued that the labor act’s protection of concerted activity is a substantive federal right that individual-arbitration waivers effectively nullify. The underlying claims in the consolidated cases were wage claims under the Fair Labor Standards Act, not discrimination claims, but the holding’s reasoning extends to employment arbitration agreements generally. That 2018 resolution postdates this article’s 2015 reference date and is presented here only as a later development.

The EEOC’s independent track

The arbitration line has a boundary, and the boundary is the government. In EEOC v. Waffle House, Inc., 534 U.S. 279, decided January 15, 2002, the Court held that an arbitration agreement between an employer and an employee does not bar the EEOC from pursuing victim-specific judicial relief, such as back pay, reinstatement, and damages, in an enforcement action. The agency was not a party to the contract, the Court reasoned, and it has independent statutory authority to bring suit. The decision means that an arbitration agreement binds the worker who signed it but does not bind the government agency charged with enforcing the statute. The Commission brings a small number of cases relative to the volume of charges it receives, selecting matters with broader significance, so most workers bound by arbitration agreements will never see an agency lawsuit vindicating their individual claim. But the track matters structurally: the arbitration line never fully closed the courthouse door, because the courthouse remained open to the public enforcer. The Commission itself once opposed mandatory arbitration as a condition of employment in a 1997 policy statement, then rescinded the statement as the case law rejected its premises, an enforcement agency adjusting its institutional position to the courts’ construction of the statutes.

The empirical record: Colvin and the caveats

Alexander Colvin’s 2011 study of American Arbitration Association filings from 2003 to 2007 found an employee win rate of 21.4 percent, using a deliberately broad definition that counted any damages award however small, with higher-paid employees winning more often: 22.7 percent for those earning under 100,000 dollars, 31.4 percent for 100,000 to 250,000 dollars, and 42.9 percent above 250,000 dollars. Colvin’s later work with Mark Gough, covering roughly 6,500 cases, found a 19.1 percent win rate and a mean award of 25,929 dollars across all rulings, with a median of 48,670 dollars among employee wins. The same research documented a repeat-player effect: employees won 14.5 percent of cases against repeat employers versus 28.8 percent against one-shot employers, and 10.8 percent when the same employer-arbitrator pairing recurred versus 20.7 percent without it. Self-representation compounded the gap: unrepresented employees’ win odds fell 46 percent and average awards 47 percent. A 2015 Economic Policy Institute report by Katherine Stone and Colvin compared forums directly, finding arbitration win rates at 59 percent of the federal-court rate and 38 percent of the state-court rate, with updated figures of 19.1 percent in arbitration versus 29.7 percent in federal court.

The repeat-player finding invites a structural explanation, though Colvin himself cautioned that the data cannot prove the mechanism. Employers who arbitrate repeatedly learn which arbitrators rule favorably and can strike unfavorable ones from the lists providers supply, while one-shot employees choose blind. Arbitrators, in turn, depend on repeat business: a provider whose arbitrators never rule for employers may find employers taking their clauses elsewhere. Whether these incentives consciously affect outcomes or operate subtly, the structural asymmetry is built into the forum in a way it is not in courts, where judges do not depend on the parties for future assignments. Defenders of arbitration respond that repeat employers also develop better compliance and settlement practices, which could explain the gap without arbitrator bias, another causal question the data cannot resolve.

A related doctrine, effective vindication, deserves mention because readers encounter it in commentary. The idea is that an arbitration agreement cannot be enforced if it prevents the effective vindication of federal statutory rights, for example by imposing costs so high that no rational claimant would proceed. In American Express Co. v. Italian Colors Restaurant, decided in 2013, the Court rejected that argument as applied to a class waiver in the antitrust context, holding 5-3 that the fact that an individual claim would be economically infeasible to arbitrate alone does not make the waiver unenforceable. The dissent argued that the decision allowed wrongdoers to insulate themselves from liability for small-dollar wrongs. Italian Colors was not an employment case, but its reasoning closed off one of the last doctrinal avenues for challenging class waivers, and it formed part of the legal background against which the employment waiver fight played out through 2015.

The caveats are as important as the numbers. Arbitrated cases are not a random sample: the disputes that reach arbitration differ systematically from those filed in court by claim type, salary level, and employer size, so win-rate gaps cannot be read as pure forum effects. Settlements before award go unobserved in the filing data, and settlement behavior may differ across forums. The definition of a win matters enormously: Colvin’s 21.4 percent figure uses the broadest possible definition, and narrower definitions counting only substantial recoveries would lower the rate. Earlier research on smaller datasets often concluded that arbitration outcomes were roughly similar to litigation outcomes, so the literature is genuinely mixed rather than one-sided. Business groups have fielded competing studies that labor-side researchers criticize for selective presentation, and the self-representation gap may reflect weaker cases and lower-income workers selecting into self-representation rather than arbitration procedure alone. Defenders of arbitration add that the forum is faster than litigation, that filing costs are typically lower, and that lower costs make it economically feasible to pursue smaller claims that no contingency-fee lawyer would take to court. This article reports the mixed record with its methodological caveats rather than adopting either the neutral-efficiency or the denial-of-rights framing.

What does the empirical research show about arbitration outcomes?

The research documents lower employee win rates and smaller awards in arbitration than in court, with studies finding win rates near one fifth of cases and mean awards around twenty six thousand dollars. But the literature is genuinely mixed: case selection effects, unobserved settlements, and competing studies mean the gaps cannot be read as pure forum effects.

A 2018 post-horizon finding

One further finding requires a date wall. In a 2018 study, Colvin reported that only about 1 in 10,400 employees subject to mandatory arbitration files a claim, a filing rate 35 to 80 times lower than in federal and state courts. That figure is dated 2018 and belongs to the post-horizon period; it is reported here as a later development, not as part of the pre-2015 record.

Why most claims never reach a court

The arbitration line explains the article’s central empirical claim: the great majority of employment claims never reach a jury because the forum has moved. When arbitration agreements became a standard hiring condition, the default forum for workplace disputes shifted from public courts to private proceedings. Class waivers, where enforced, compounded the shift by removing the aggregation mechanism that made small-value claims economically viable for private attorneys. Private attorneys working on contingency evaluate the expected recovery against the cost of representation, and a claim worth a few thousand dollars in an individual arbitration rarely clears that bar. The result is not only that filed claims move forums but that many potential claims are never filed at all, a suppression effect the win-rate studies cannot capture because they observe only filed cases.

The class action line

If arbitration keeps individual claims out of court, the class action is the device by which plaintiffs aggregate many small or dispersed claims into one case, and the Supreme Court’s 2011 decision in Wal-Mart Stores, Inc. v. Dukes reshaped what plaintiffs must show to use it. These employment decisions belong to a larger body of rulings surveyed in the companion guide to the Supreme Court’s civil rights act decisions, which carries the doctrine of the civil rights title itself.

Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011), decertified the largest employment class ever certified. The plaintiffs sought to certify a nationwide class of some 1.5 million current and former female Wal-Mart employees, across roughly 3,400 stores, alleging that the company’s policy of delegating pay and promotion decisions to local managers produced systemic sex discrimination.

Rule 23’s structure explains why the commonality holding mattered so much. A class seeking certification must first satisfy four prerequisites: numerosity, commonality, typicality, and adequacy of representation. It must then fit one of the rule’s three categories: subsection (b)(1) for incompatible-standards cases, subsection (b)(2) for classes seeking primarily injunctive or declaratory relief, and subsection (b)(3) for damages classes, which additionally require predominance of common questions and superiority of the class device. Employment classes historically proceeded under (b)(2), which carries no predominance requirement, on the theory that systemic discrimination called for systemic injunctive relief with back pay as incidental. Dukes attacked the entry gate, commonality, which every class must pass regardless of category.

Commonality after Dukes

Rule 23(a)(2) requires that class members’ claims share common questions of law or fact. The Court, 5-4, held that the Dukes class failed that requirement. Justice Scalia’s majority held that a common contention must be of such a nature that it is capable of classwide resolution, which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke. The plaintiffs offered statistical evidence of disparities, anecdotal accounts of discrimination, and a sociologist’s testimony about corporate culture, but the majority found no glue holding together the reasons for thousands of discretionary decisions by individual managers. Without evidence of a uniform policy or decision-making practice, the claims could not be resolved together.

The expert evidence illustrated the methodological stakes. The plaintiffs offered a sociologist’s social-framework testimony that the company’s strong corporate culture, combined with decentralized discretionary decision-making, made employment decisions vulnerable to stereotyping. The majority treated that testimony as insufficient to supply the required glue, reasoning that a general theory of cultural vulnerability could not substitute for proof tying specific decisions together. The dissent viewed the testimony as exactly the kind of common proof the rule contemplates, arguing that a uniform policy of delegating discretion within a uniform culture was itself the common contention. The disagreement was thus not only about the legal standard but about what counts as evidence of a common practice, a question that continues to divide lower courts in certification fights.

The dissent, by Justice Ginsburg, argued that the majority had smuggled the predominance inquiry of Rule 23(b)(3) into the commonality inquiry of Rule 23(a)(2), raising the bar beyond what the rule’s text requires, and maintained that a company’s uniform delegation of discretionary authority could itself supply the common question. The dissent’s deeper objection was institutional. Large employers had increasingly decentralized personnel decisions, and the majority’s rule, in the dissent’s view, rewarded that decentralization with immunity from aggregate challenge: the more diffuse the decision-making, the harder the class. The majority’s answer was that Rule 23 requires proof, not inference, and that statistical disparities plus anecdotes cannot substitute for evidence of a common policy. The exchange captures a persistent tension in aggregate litigation between the difficulty of proving systemic wrongs and the rigor the rules demand before allowing them to be proved together.

The vote counts attach to each holding separately. The 5-4 division attached to the commonality holding; on the separate question whether back-pay claims could be certified under Rule 23(b)(2), the Court was unanimous that they could not, because the monetary relief sought was not incidental to injunctive or declaratory relief. After Dukes, plaintiffs seeking both structural relief and money must generally satisfy the more demanding Rule 23(b)(3) requirements, including predominance and superiority, or split the case.

The certification tightening continued two years later. In Comcast Corp. v. Behrend (2013), the Court held 5-4 that a damages class under Rule 23(b)(3) must show its damages model measures only the harm attributable to the liability theory. Although Comcast was an antitrust case, lower courts applied its damages-model discipline to employment classes, adding a second front to the certification fight. The result was a pincer: (b)(2) classes faced the heightened commonality gate, and (b)(3) classes faced predominance plus the damages-model gate.

The practical effect was to make large employment classes substantially harder to certify. Plaintiffs must identify a specific employment practice, applied uniformly, that permits classwide proof, rather than pointing to decentralized discretion plus statistical disparities. Combined with the arbitration line, the effect compounds: individual arbitration agreements channel claims out of court, and Dukes raises the bar for the claims that remain. A worker with a modest-value claim faces a forum that requires individual proceedings and a certification standard that resists aggregation, which is why this article locates the decisive variable in the forum rather than the doctrine.

In the years after Dukes, the lower courts applied the tightened commonality standard to employment classes with results that generally narrowed certification. Plaintiffs who could point to a uniform policy, a biased test, or centralized decision-making continued to certify classes, while plaintiffs challenging decentralized discretionary systems faced the glue requirement Dukes had announced and frequently failed it. The decision thus sorted employment class actions by the structure of the employer’s decision-making: centralized systems remained vulnerable to aggregation, decentralized systems largely did not. That sorting effect is one of the decision’s most important practical consequences, because it gave employers’ organizational design a direct bearing on their class-action exposure.

The dissent’s response is part of the record and deserves equal presentation. Justice Ginsburg, writing for four, argued that the majority had converted commonality from a threshold requirement into a merits inquiry, demanding at the certification stage the very proof of a uniform discriminatory policy that trial was meant to test. The dissent credited the plaintiffs’ evidence, statistical disparities across regions and job categories, anecdotal accounts, and expert testimony about a strong corporate culture of gender stereotyping, as sufficient to show that the delegation of discretion itself was the common practice being challenged. And the dissent closed with the institutional point that had organized employment law since 1991: when the Court construes the civil rights statutes narrowly, Congress can and has responded, inviting legislative repair. Whether Congress would take up that invitation for Rule 23 was, in 2015, an open question.

What must plaintiffs show to certify an employment class action?

After a 2011 decision, plaintiffs seeking to certify an employment class must identify a common contention capable of classwide resolution, meaning its truth or falsity resolves an issue central to every claim in one stroke. Discretionary decisions by local managers, without evidence tying those decisions together, do not satisfy the commonality requirement.

The supervisor line: Vance and the employer defense

A different procedural question determines who the employer answers for. Under Title VII, an employer is vicariously liable for harassment by a supervisor that culminates in a tangible employment action, such as firing or demotion, with no defense. But where the supervisor’s harassment creates a hostile work environment without a tangible employment action, the employer may assert the affirmative defense recognized in two 1998 decisions, Faragher v. City of Boca Raton, 524 U.S. 775, and Burlington Industries, Inc. v. Ellerth, 524 U.S. 742. The defense has two elements, and the employer must prove both: that it exercised reasonable care to prevent and correct promptly any harassing behavior, generally through a complaint procedure, and that the plaintiff employee unreasonably failed to take advantage of any preventive or corrective opportunities the employer provided or to avoid harm otherwise.

Harassment by a coworker who is not a supervisor is judged under a negligence standard: the employer is liable only if it knew or should have known of the harassment and failed to act. Everything therefore turns on who counts as a supervisor, and the circuits had split between two definitions. The EEOC, joined by the Second and Fourth Circuits, treated as supervisors those employees with authority to direct the victim’s daily work. Other circuits required authority over tangible employment decisions. In Vance v. Ball State University, 570 U.S. 421, decided in 2013, the Court adopted the narrower definition 5-4. Justice Alito’s majority opinion held that an employee is a supervisor for purposes of vicarious liability only if empowered by the employer to take tangible employment actions against the victim, meaning the power to effect a significant change in employment status such as hiring, firing, failing to promote, reassignment with significantly different responsibilities, or a decision causing a significant change in benefits. The Court expressly rejected the EEOC’s broader definition. Justice Ginsburg’s dissent argued that the majority’s definition ignored how authority actually operates in modern workplaces, where the person who controls scheduling and daily oversight wields the power that makes harassment inescapable even without the formal power to fire.

The rejected alternative had real support. The Equal Employment Opportunity Commission’s enforcement guidance defined supervisor to include anyone with authority to direct another’s daily work activities, a standard the Second and Fourth Circuits had adopted. Under that definition, shift leads and team coordinators with scheduling power counted as supervisors, substantially expanding vicarious liability. The Vance majority found the standard unworkable because daily-work direction exists on a spectrum with no principled cutoff, while the dissent found the majority’s cutoff arbitrary because it excluded the very workers whose power enables harassment. The dissent’s workplace-reality argument drew on the record of how harassment typically operates: the harasser is often a shift lead, crew chief, or team coordinator who controls schedules, assignments, and daily working conditions without formal hire-and-fire authority. Under the majority’s rule, such a harasser is judged under negligence, requiring the victim to prove the employer knew or should have known, a showing that fails when the employer can claim ignorance of shop-floor dynamics.

The negligence standard itself has teeth that litigants sometimes underestimate. An employer that receives a complaint and does nothing, that knows of a harasser’s history and takes no preventive steps, or whose complaint mechanism is a dead letter can be found negligent on the paper record. The standard also creates a documentation incentive symmetric to the affirmative defense: employers that log complaints, investigate promptly, and discipline offenders build the record that defeats negligence, while workers who report early and in writing build the record that establishes it.

The supervisor line is procedural in the sense that matters here: it allocates the burden of proof and determines which liability standard applies before anyone asks whether the harassment was severe or pervasive enough to be actionable. The defense’s first element is usually shown through a written anti-harassment policy, a complaint mechanism that does not require reporting to the harasser, and prompt investigation of complaints. The second element turns on the worker’s conduct: a worker who never reports harassment that the policy invited her to report will struggle to defeat the defense, while a worker who reported and was ignored can show her failure to use the procedure was reasonable. Courts apply the defense case by case, and the fact-intensive nature of both elements means the defense is often resolved at summary judgment or trial rather than on the pleadings.

Who counts as a supervisor under Title VII?

Under a 2013 decision, an employee counts as a supervisor for Title VII harassment liability only when empowered to take tangible employment actions against the victim, such as hiring, firing, failing to promote, or reassigning with significantly different responsibilities. A worker who merely directs another’s daily tasks does not qualify.

Summary judgment: where most cases end

Arbitration diverts the great majority of claims from court, but among the claims that remain, most end at summary judgment rather than trial. The standard, drawn from Federal Rule of Civil Procedure 56 and the Supreme Court’s 1986 trilogy, asks whether any genuine dispute of material fact requires a jury; if the paper record could not persuade a reasonable jury for the plaintiff, the judge decides the case. Employment disputes are unusually susceptible to this disposition because they turn on documented reasons: performance reviews, attendance records, emails, and the employer’s articulated justification create a record that judges can evaluate without witness credibility determinations.

The modern standard dates to three 1986 decisions, Celotex Corp. v. Catrett, Anderson v. Liberty Lobby, and Matsushita Electric Industrial Co. v. Zenith Radio, which together made summary judgment more available by clarifying that the moving party need not produce evidence negating the opponent’s claim. Employment defendants benefited substantially: the trilogy gave judges tools to weed out claims built on speculation, stray remarks, or subjective belief unsupported by the record. Plaintiffs’ advocates have long argued that the trilogy imported a paper-trial mentality into a system meant to preserve jury resolution of credibility disputes, particularly in discrimination cases where intent is the central question and direct evidence is rare.

The summary-judgment stage is where the causation standards do their heaviest work. Under the motivating-factor standard, a plaintiff who produces evidence that the forbidden trait was one input among several can often survive, because a jury could find the standard met. Under but-for causation, the plaintiff must produce evidence from which a jury could find the action would not have occurred without the forbidden consideration, a showing that fails when the employer’s documented legitimate reasons stand unrebutted. The McDonnell Douglas pretext stage thus becomes the whole case: if the plaintiff cannot raise a genuine dispute about the employer’s stated reason, the claim ends on paper. This dynamic explains why discovery focuses so intensely on comparator evidence, shifting explanations, and remarks suggesting bias; each is an attempt to build the record that survives the motion.

One evidentiary doctrine illustrates the motion’s filtering function. Courts routinely discount isolated biased remarks by non-decisionmakers, or remarks remote in time from the challenged action, as stray remarks insufficient to show pretext. A single offensive comment by a coworker with no role in the decision rarely survives, while repeated comments by the decisionmaker close in time to the action may. The doctrine’s application is fact-sensitive and sometimes screens out evidence a jury should weigh, but it remains a standard tool by which judges keep the pretext inquiry tethered to the decision at issue.

The sovereign immunity map

The Constitution’s Eleventh Amendment immunizes states from private damages suits in federal court, but Congress may abrogate that immunity when it acts under a valid grant of power. The Amendment’s text speaks only of suits against a state by citizens of another state, but the Court since Hans v. Louisiana (1890) has read it to embody a broader principle of sovereign immunity that also bars suits by a state’s own citizens. Congress may override that immunity, but only when acting under a constitutional provision that grants the power to do so. Abrogation requires a clear statement in the statutory text, and as a result the modern immunity fights are fought almost entirely on the second question, whether Congress acted under a valid source of power.

For the employment statutes, the relevant grant is section 5 of the Fourteenth Amendment, which authorizes Congress to enforce the amendment’s guarantees. The governing test comes from City of Boerne v. Flores (1997), which held that Congress’s section 5 power is remedial rather than substantive: Congress may enforce the Fourteenth Amendment as the Court interprets it but may not redefine its substance. Legislation must therefore show congruence and proportionality between the constitutional injury Congress documented and the statutory remedy it imposed. The test’s bite depends on the level of constitutional scrutiny the underlying classification receives. Race and sex classifications trigger heightened scrutiny, so Congress could more easily document unconstitutional state conduct and justify broad remedies. Age and disability classifications trigger only rational-basis review, so Congress faced a steeper climb: it had to show a pattern of state conduct that was unconstitutional even under deferential review.

Title VII: abrogation valid

Fitzpatrick v. Bitzer, 427 U.S. 445 (1976), upheld Congress’s abrogation for Title VII. The Court held that the Eleventh Amendment and the principle of state sovereignty are limited by section 5, and that the 1972 amendments extending Title VII to state and local government employees were a valid exercise of the enforcement power. State employees may therefore sue states in federal court for employment discrimination and recover money damages, including back pay. Title VII sits on the favorable side of the map because Congress documented the constitutional problem and tailored the remedy to it.

The decision’s context matters. The 1972 amendments extended Title VII to state and local government employees over a record of public-sector discrimination, and the Court treated that record as sufficient to support abrogation for a statute enforcing the core equal-protection guarantee against race and sex discrimination. Fitzpatrick also established the principle the later cases apply: the Eleventh Amendment yields to a valid exercise of the section 5 power, so the immunity question is always a question about the statute, never about sovereign dignity in the abstract.

The ADEA: abrogation invalid

Kimel v. Florida Board of Regents, 528 U.S. 62 (2000), reached the opposite result for the age statute. The Court, 5-4 on the dispositive portions, held that the ADEA was not a valid exercise of the section 5 power. Congress had not identified a widespread pattern of unconstitutional age discrimination by states, and age classifications receive only rational-basis review under the Equal Protection Clause, so the broad prohibition Congress enacted was not congruent and proportional to any documented constitutional violation. The ADEA’s purported abrogation was therefore invalid, and private individuals cannot recover money damages from states under the age statute. The same Congress that validly abrogated for race and sex discrimination in employment could not do so for age, because the underlying constitutional baseline differs.

The dissenters argued that Congress had in fact documented extensive age discrimination by states, pointing to the legislative record of hearings and state practices, and that the majority applied the congruence test with a selectivity that discounted that evidence. The majority answered that much of the record concerned private-sector conduct or reflected rational state personnel policies, such as mandatory retirement for public-safety workers, which the Constitution permits. The exchange illustrates how the congruence inquiry turns on characterizing the legislative record, a task on which reasonable judges divide.

The ADA: split by title

Board of Trustees of the University of Alabama v. Garrett, 531 U.S. 356 (2001), held 5-4 that Congress did not validly abrogate state immunity for Title I of the ADA, the employment title. The majority concluded that Congress had identified no sufficient pattern of unconstitutional disability discrimination in public employment, so the accommodation mandate was not a congruent and proportional response. State employees therefore cannot recover damages from states for violations of the ADA’s employment provisions.

The dissenters contended that the majority had understated the record of state employment discrimination against people with disabilities, including evidence of exclusion from public workplaces and stereotyped assumptions about capability. The majority responded that the record showed societal discrimination broadly rather than unconstitutional state action specifically, and that the ADA’s sweeping accommodation mandate exceeded any documented constitutional violation. As with Kimel, the disagreement centered on how to read the legislative record and how demanding the congruence test should be.

Tennessee v. Lane, 541 U.S. 509 (2004), carved a narrow exception for Title II, the public-services title, and the exception must be kept narrow. The Court, 5-4, held that Title II validly abrogated immunity only as applied to the class of cases implicating the fundamental right of access to the courts. Congress had documented disability discrimination implicating due-process rights, including courthouse access, and the reasonable-accommodation remedy was congruent and proportional in that specific context. Lane does not validate Title II abrogation generally; it validates it for court access. A state employee’s Title I employment claim against a state remains barred under Garrett even after Lane.

The case’s facts explain the narrow holding. George Lane, a paraplegic criminal defendant, had to crawl up two flights of courthouse stairs to reach the courtroom for his own hearing, and Beverly Jones, a court reporter, alleged she lost work because courthouses were inaccessible. Congress had documented such exclusion from the justice system specifically, which is why the Court could find congruence and proportionality for court access while leaving the broader Title II question for another day. The as-applied technique, validating abrogation for one class of cases without deciding others, became the template for later refinements.

The FMLA: split by type of leave

The leave statute produced the most finely divided map, because the answer turns on which kind of leave was taken. Nevada Department of Human Resources v. Hibbs, 538 U.S. 721 (2003), upheld abrogation 6-3 for the FMLA’s family-care provision, leave to care for a child, spouse, or parent. Chief Justice Rehnquist’s majority reasoned that Congress had targeted a documented pattern of sex-based stereotyping in states’ administration of leave benefits, and that the gender-neutral family-care entitlement was a congruent and proportional remedy. State employees may recover money damages for family-care violations.

The Hibbs lineup itself was notable: Chief Justice Rehnquist, the Court’s leading skeptic of section 5 abrogation, wrote the majority opinion, joined by four justices more sympathetic to congressional power. The alignment signaled that the stereotyping record was strong enough to persuade even the congruence test’s toughest grader, which is why Hibbs remains the high-water mark for employment abrogation. The hearings record gave the majority its material. Congress heard testimony that state employers routinely denied men family-care leave while penalizing women who took it, reflecting the assumption that caregiving was women’s work. The record included state personnel manuals and testimony from workers in both directions, men discouraged from caregiving and women channeled into it. That two-sided stereotyping evidence is what distinguished Hibbs from Kimel and Garrett: Congress had documented not just differential treatment but a sex-based classification system in leave administration, giving the gender-neutral remedy a direct target.

Coleman v. Court of Appeals of Maryland, 566 U.S. 30 (2012), held that the self-care provision, leave for the employee’s own serious health condition, did not validly abrogate immunity. The judgment was 5-4, but the opinion structure requires precision: Justice Kennedy’s opinion was a plurality, joined by three colleagues, with Justice Scalia concurring in the judgment only and Justice Thomas concurring separately. The plurality reasoned that, unlike family care in Hibbs, the self-care provision was not directed at an identified pattern of gender-based discrimination by states, so it was not congruent and proportional section 5 legislation. The leave statute’s full framework is covered in the companion guide to the Family and Medical Leave Act of 1993.

The opinion lineup in Coleman deserves attention because it shows how fragile the FMLA map is. Justice Scalia concurred in the judgment only, on the ground that Hibbs itself was wrongly decided and should be overruled; his vote supplied the fifth for the judgment while rejecting the plurality’s reasoning. Justice Thomas wrote separately to emphasize the limits of the section 5 power. The result is a 5-4 judgment resting on a four-justice plurality, meaning the self-care holding commands a majority only for its outcome, not its rationale. Justice Ginsburg’s dissent argued the distinction was artificial, contending that women take more family and medical leave overall and that denying self-care abrogation perpetuated the very stereotypes Hibbs condemned. The plurality’s answer, that congruence requires a documented pattern of unconstitutional conduct rather than a plausible theory, left the FMLA split in place.

The map’s logic is worth stating plainly. Abrogation does not turn on how serious the underlying discrimination is, nor on how clearly the statute prohibits it. It turns on whether Congress, acting under the Fourteenth Amendment, documented a pattern of unconstitutional state conduct and responded with a proportionate remedy. One qualification keeps the map honest. The immunity cases bar private suits for money damages against states, but they do not bar suits for prospective injunctive relief against state officials under Ex parte Young (1908), nor do they bar the federal government itself from suing states to enforce these statutes. The map describes who may recover damages from the state treasury, not who is entirely without remedy.

Can state employees sue their employer for damages?

It depends on the statute. State employees may sue their employer for damages under Title VII, the FMLA’s family-care provision, and the ADA’s court-access applications, because the Court found each a valid exercise of Congress’s enforcement power. They may not recover damages under the ADEA, ADA Title I, or the FMLA’s self-care provision.

Damages and remedies: what a victory is worth

The forum and the proof standard determine whether a plaintiff wins; the remedial scheme determines what winning is worth. Here again the statutes diverge, and the 1991 Act is the hinge.

The 1981a caps

Before 1991, Title VII authorized back pay and equitable relief but not compensatory or punitive damages. Section 102 of the Civil Rights Act of 1991, codified at 42 U.S.C. 1981a, added compensatory and punitive damages for intentional discrimination, subject to per-complaining-party caps that scale with employer size: $50,000 for employers with more than 14 and fewer than 101 employees, $100,000 for more than 100 and fewer than 201, $200,000 for more than 200 and fewer than 501, and $300,000 for more than 500 employees. Compensatory damages cover future pecuniary losses and nonpecuniary harms such as emotional pain, suffering, inconvenience, mental anguish, and loss of enjoyment of life; punitive damages require proof of malice or reckless indifference and are unavailable against government employers.

The compensatory categories repay attention because they define what the caps actually limit. Future pecuniary losses include the out-of-pocket costs discrimination causes beyond lost wages, such as medical expenses from stress-related illness or the costs of a job search. Nonpecuniary harms cover the dignitary and emotional injuries that often exceed the economic ones in the worker’s lived experience but that the pre-1991 statute left uncompensated entirely. The caps apply to the sum of these categories plus punitive damages, so a worker with severe emotional harm and a modest wage loss hits the same ceiling as a worker with the reverse profile, a compression the tiered structure does not relieve. Back pay and other equitable relief under the pre-1991 remedial section are excluded from the caps.

The caps were the price of enactment. A broader 1990 bill that would have authorized uncapped damages was vetoed, and the 1991 compromise that replaced it scaled damages to employer size to answer the objection that unlimited jury awards would bankrupt small businesses. The tiered structure, from $50,000 to $300,000, reflects that negotiation rather than any theory of optimal deterrence. Business groups had opposed uncapped damages as inviting runaway jury awards, while civil rights advocates insisted that back pay alone gave employers no reason to change behavior. The tiered caps split the difference by employer size, and the malice-or-reckless-indifference standard for punitive damages added a fault filter on top of the caps. The structure that emerged was less a theory of deterrence than a legislative settlement, which explains its rough edges: the thresholds are round numbers chosen for administrability, not calibrated to any measured optimal penalty. The punitive-damages standard comes from Kolstad v. American Dental Association (1999), which held that malice or reckless indifference turns on the employer’s knowledge that its actions may violate federal law, not on egregiousness alone. And because the caps have never been adjusted for inflation since 1991, their real value has eroded steadily, a point commentators across the spectrum have noted.

How do the 1991 damages caps change with employer size?

Yes, for the statutes that authorize compensatory and punitive damages. Under the 1991 damages provision, the combined award per complaining party is capped from $50,000 for employers with 15 to 100 workers up to $300,000 for employers with more than 500. The ADEA authorizes no compensatory or punitive damages, only back pay, doubled for willful violations.

The age, pay, and leave remedial menus

The age statute incorporates the wage-and-hour remedial scheme rather than the 1991 damages regime. A successful ADEA plaintiff may recover back pay, front pay or reinstatement, and injunctive relief, plus liquidated damages, generally computed by doubling the back-pay award, when the violation was willful, meaning the employer knew or showed reckless disregard for the statute’s requirements. Compensatory and punitive damages are not available under the ADEA at all. The practical gap is significant: an older worker who proves willful age discrimination recovers economic losses doubled, while a worker who proves intentional race discrimination recovers economic losses plus capped non-economic and punitive damages. The willfulness standard for liquidated damages comes from Trans World Airlines, Inc. v. Thurston (1985): the worker must show the employer knew its conduct was prohibited or showed reckless disregard for whether it was. The standard is easier to state than to prove, since it requires evidence about the employer’s state of mind, such as ignored legal advice or a pattern of similar violations. When met, the doubling is automatic, which makes willfulness the most valuable factual issue in many age cases.

Two further remedial doctrines shape what victory is worth. Reinstatement is the preferred equitable remedy, but courts award front pay, the discounted value of future lost earnings, where reinstatement is infeasible because the employment relationship has broken down or the position no longer exists. And every back-pay award is reduced by the worker’s duty to mitigate: the plaintiff must make reasonable efforts to find comparable work, and interim earnings are offset against the recovery. An employer that proves the worker made no effort to mitigate can slash the award substantially, which is why defense counsel subpoena job-search records and plaintiffs’ counsel document every application. The Equal Pay Act’s remedies follow its wage-and-hour ancestry: back pay for the period of the violation, limited to two years before filing or three for willful violations, with liquidated damages in an equal amount where the violation is willful. The FMLA provides lost wages and benefits, interest, liquidated damages equal to the lost wages unless the employer proves good faith, and equitable relief such as reinstatement; it provides no compensatory or punitive damages.

A comparative illustration makes the pricing concrete. Consider three workers at mid-size employers, each losing 60,000 dollars in back pay. The first proves race discrimination under Title VII and shows the decision was motivated by bias but not taken with malice; she recovers the back pay plus capped compensatory damages for emotional harm, within the tier for her employer’s size. The second proves age discrimination under the ADEA and shows the employer knew its conduct violated the statute; he recovers the back pay doubled to 120,000 dollars as liquidated damages, but nothing for emotional distress, which the ADEA does not compensate. The third proves a willful Equal Pay Act violation; she recovers the back pay plus an equal amount as liquidated damages. Same lost wages, three different recoveries, each dictated by the remedial choices of a different Congress in a different decade.

The four-statute procedural table

The findable artifact of this article is the four-statute procedural table, expanded here to five rows to include the leave law whose immunity split the brief requires. Each row answers the six questions the brief specifies: the causation standard, the administrative exhaustion requirement, the limitations period, damages availability and cap, arbitration exposure, and state employee suit availability. Readers comparing the rows may keep the comparison in a legislation study notebook for reference alongside the statute articles.

Statute	Causation standard	Administrative exhaustion requirement	Limitations period	Damages availability and cap	Arbitration exposure	State employee suit availability
Title VII, Civil Rights Act of 1964 (as amended 1972, 1991)	Motivating factor for status discrimination, 42 U.S.C. 2000e-2(m); but-for for retaliation, Nassar (2013)	EEOC charge required	180 days to file charge, 300 where a state or local agency exists; 90 days to sue after right-to-sue notice	Back pay plus compensatory and punitive damages capped at 50,000 to 300,000 dollars by employer size, 42 U.S.C. 1981a; no punitive damages against governments	Yes; individual agreements enforceable; class-waiver validity unresolved in 2015	Yes; valid abrogation, Fitzpatrick v. Bitzer (1976)
Age Discrimination in Employment Act of 1967	But-for, Gross (2009); burden of persuasion never shifts	EEOC charge required	180 days to file charge, 300 where state law covers age; suit from 60 days after charge filing to 90 days after EEOC completes action	Back pay plus liquidated (double) damages for willful violations; no compensatory or punitive damages	Yes; ADEA claims held arbitrable in Gilmer (1991)	No damages suits against states; invalid abrogation, Kimel (2000)
Title I, Americans with Disabilities Act of 1990	Contested in 2015; Seventh and Sixth Circuits applied but-for (Serwatka 2010; Lewis 2012); Supreme Court silent; other circuits uncommitted	EEOC charge required	180 days to file charge, 300 where a state or local agency exists; 90 days to sue after right-to-sue notice	Same 1981a caps as Title VII: 50,000 to 300,000 dollars by employer size	Yes; FAA employment line applies	No damages suits under Title I; invalid abrogation, Garrett (2001)
Equal Pay Act of 1963	Affirmative-defense structure: employer must prove the pay differential rests on seniority, merit, production, or a factor other than sex	No EEOC charge required	2 years to sue, 3 for willful violations, 29 U.S.C. 255(a); back pay limited to 2 years before filing, 3 for willful	Back pay plus equal liquidated damages for willful violations; no compensatory or punitive damages	Yes in principle; wage claims fall under the FAA employment line	No Supreme Court decision had sustained a damages suit against a state under the Act by the end of 2015
Family and Medical Leave Act of 1993	Entitlement structure: interference turns on entitlement and denial; retaliation on adverse action for exercising rights	No EEOC charge; private suit or Labor Department complaint	2 years to sue, 3 for willful violations, 29 U.S.C. 2617(c)	Lost wages and benefits, interest, liquidated damages equal to lost wages absent good faith, equitable relief; no compensatory or punitive damages	Yes in principle; FAA employment line applies	Family-care leave: yes, Hibbs (2003); self-care leave: no, Coleman (2012, 5-4 plurality)

Reading one row shows how the columns interact. Take Title VII retaliation: the causation cell demands but-for proof, the exhaustion cell requires a timely commission charge, the limitations cell allows 90 days after the right-to-sue notice, the damages cell offers the 1981a caps, the arbitration cell warns that an individual agreement will move the forum, and the immunity cell confirms that state employees may recover damages. Change the statute to the ADEA and three cells change: causation stays but-for but under a different decision, damages lose the non-economic categories, and immunity flips to no. The table’s value lies in making those cross-statute differences visible at a glance.

Reading the procedural map

The table’s value lies less in any single row than in how the rows combine for a worker whose situation touches more than one statute, which is the common case rather than the exception. Consider a hypothetical 58-year-old Black woman employed by a state agency who believes she was denied a promotion because of her race, her sex, and her age, and who reported the denial to human resources before being transferred to a less desirable position. Her potential claims span three rows. The Title VII race and sex claims use the motivating-factor standard and permit capped compensatory and punitive damages, and Fitzpatrick preserves her damages suit against the state. The ADEA age claim requires but-for causation, offers only back pay with possible doubling for willfulness, and under Kimel cannot produce a damages judgment against the state at all. The transfer following her internal report adds a Title VII retaliation claim, which under Nassar requires but-for causation even though it sits in the same title as her motivating-factor status claims. She must file an EEOC charge within the applicable deadline or lose the Title VII and ADEA claims alike, and if she signed an arbitration agreement, the forum for the private claims may be an arbitrator while the EEOC retains its independent authority. One set of facts, three statutes, two causation standards, two remedial regimes, and opposite immunity answers: the table exists so that no reader has to hold all of that in memory at once.

For the reader who wants a working checklist, the questions run in order: which statute and which provision supplies the claim, and therefore which causation standard; whether the administrative charge was timely filed and the suit timely brought; whether an arbitration agreement controls the forum and what it says about collective proceedings; whether the claim can be aggregated or must proceed alone; whether the employer is a state and the statute permits a damages suit against it; and what remedy the statute makes available. Answer those six, in that order, and the shape of any workplace dispute comes into view before a single fact about the underlying events is disputed.

Three recurring errors close the account, and each corresponds to one of the article’s lines. The first is assuming one causation standard applies across statutes. It does not. Motivating factor governs Title VII status claims because Congress wrote it into 42 U.S.C. 2000e-2(m) in 1991; but-for governs Title VII retaliation and ADEA claims because Congress did not write it into those provisions; the ADA standard was contested in 2015. The correction is not a different single rule but the abandonment of single rules: name the statute, name the provision, then name the standard. The second error is assuming all claims may be arbitrated or none may. The truth is mixed by design. Most employment claims under most statutes may be arbitrated under individually signed agreements, per Circuit City and Gilmer. But the EEOC may still investigate, file charges, and sue for victim-specific relief in court per Waffle House, and in 2015 the validity of class and collective waivers was an unresolved split between the labor board and the courts. The third error is assuming state employees have identical rights under every statute. They do not: a state employee may recover damages under Title VII and for FMLA family-care leave, may not recover damages under the ADEA, under ADA Title I, or for FMLA self-care leave, and faces an untested question under the Equal Pay Act.

A reader who has followed the four lines will have noticed a recurring move: the dissent urging Congress to repair the majority’s reading. Justice Ginsburg made the invitation explicitly in Nassar and in Dukes, the Gross dissent implied it, and the Coleman dissent framed the plurality’s line-drawing as a legislative question wrongly decided by judges. The invitations matter because Congress has accepted them before: the Civil Rights Act of 1991 was itself a congressional repair of Price Waterhouse and several other late-1980s decisions; the Lilly Ledbetter Fair Pay Act of 2009 was a direct override of Ledbetter v. Goodyear; the ADA Amendments Act of 2008 was a rebuke to a line of Supreme Court decisions that had narrowed the definition of disability, though it left the causation question untouched. At this article’s 2015 reference date, three invitations stood open with no legislative response yet recorded: the Nassar dissent’s call to extend the motivating-factor standard to retaliation claims, the Dukes dissent’s suggestion of a legislative answer to the tightened commonality standard, and the ADA causation split, which awaited either a Supreme Court decision or a statutory amendment. The architecture described in this article is therefore a snapshot of a moving system, accurate in April 2015, with the fault lines for future movement clearly marked.

The forum decides the case. That claim has been given its content: the causation standard that applies depends on which statute and which provision the plaintiff invokes, and the differences trace to Congress’s 1991 choice and the Court’s refusal to extend it; the great majority of disputes never reach a jury because arbitration agreements, upheld across three decades of decisions, have become the default forum; the aggregation mechanism that once made small claims viable was narrowed by the commonality decision; and the availability of damages against state employers depends on a statute-by-statute immunity map that splits even within a single statute. Substance still matters, but in this field it matters last, after the forum, the clock, the proof standard, and the defendant’s amenability to suit have each had their say.

Two cautions close the account. The first concerns the arbitration debate. The outcome research detailed above documents real gaps between arbitration and litigation while leaving their causes disputed, and the dispute matters because the two advocacy framings draw opposite conclusions from the same numbers. The efficiency framing reads lower costs and faster resolution as the point of the forum; the denial-of-rights framing reads lower win rates and awards as the price workers pay for it. The methodological caveats, selection effects, unobserved settlements, and competing studies, mean neither framing can claim the data conclusively. This article reports the findings with their caveats and leaves the characterization to the reader.

The second caution is about use. This article describes the procedural architecture of federal employment law; it does not assess any reader’s situation, predict any outcome, or substitute for advice about a specific dispute. Deadlines run from events readers may not recognize as triggering them, standards vary by circuit on contested questions, and later decisions refine every cell of the table. The map is a starting point for understanding the system, not a guide to navigating an individual case.

Frequently Asked Questions

Q: Why do most workplace discrimination cases go to arbitration?

Most workplace disputes reach arbitration because employers made arbitration agreements a standard condition of hiring after the Supreme Court held the Federal Arbitration Act covers most employment contracts. Circuit City Stores v. Adams (2001) read the Act’s employment exemption narrowly to cover only transportation workers, and Gilmer v. Interstate/Johnson Lane (1991) held that statutory discrimination claims may be arbitrated. Together with the 2011 decision preempting state rules hostile to arbitration, these rulings meant that a signed agreement routinely sends a discrimination claim to a private arbitrator instead of a court. Because employers made such agreements widespread at hiring, the default forum for workplace disputes shifted from public courts to private proceedings.

Q: What did Epic Systems decide about workplace discrimination claims?

This question requires a date wall, because the article’s reference date is April 15, 2015, and Epic Systems Corp. v. Lewis, 584 U.S. 497, was decided May 21, 2018. The Court held 5-4 that employment arbitration agreements requiring individualized proceedings and waiving class and collective actions are enforceable under the Federal Arbitration Act, that section 7 of the National Labor Relations Act does not make them unlawful, and that the Act’s savings clause does not supply a contrary result. The majority emphasized enforcing agreements according to their terms; the dissent argued the labor statute’s protection of concerted activity is a substantive right the waivers effectively nullify. The consolidated cases involved wage claims under the Fair Labor Standards Act, not discrimination claims, though the reasoning extends to employment arbitration agreements generally. In 2015, the class-waiver question was an unresolved circuit split.

Q: What causation standard applies to each workplace discrimination statute?

The standards differ by statute and provision. Status discrimination under Title VII uses the motivating-factor standard Congress wrote into the statute in 1991: the plaintiff prevails by showing race, color, religion, sex, or national origin was a motivating factor, even if other factors also motivated the decision. Retaliation under Title VII requires but-for causation after the 2013 Nassar decision, because the 1991 amendment never touched the retaliation provision. Age claims under the ADEA require but-for causation after the 2009 Gross decision, for the same reason. Disability claims under the ADA sat in a contested position in 2015: the Supreme Court had never ruled, two circuits applied but-for, and other courts left the question open.

Q: Can you bring a workplace discrimination class action?

Class actions remain available in theory but face steep barriers after Wal-Mart Stores, Inc. v. Dukes (2011). Plaintiffs must identify a common contention capable of classwide resolution, meaning its truth or falsity resolves an issue central to every class member’s claim in one stroke; statistical disparities plus decentralized discretionary decisions, without evidence tying those decisions together, do not satisfy commonality. Claims for back pay cannot be certified under Rule 23(b)(2) where monetary relief is not incidental to injunctive relief, a point the Court decided unanimously. Separately, many workers signed arbitration agreements with class waivers, which channel claims into individual proceedings where aggregation is unavailable. Between the certification standard and the arbitration channel, large employment classes have become substantially harder to assemble.

Q: Can state employees bring workplace discrimination claims?

It depends on the statute. State employees may sue their state employer for damages under Title VII, because Fitzpatrick v. Bitzer (1976) upheld Congress’s abrogation of state sovereign immunity, and under the FMLA’s family-care provision, because Nevada v. Hibbs (2003) upheld abrogation for leave to care for a child, spouse, or parent. They may not recover damages under the ADEA, because Kimel v. Florida Board of Regents (2000) found the abrogation invalid; under ADA Title I, because Board of Trustees v. Garrett (2001) found that abrogation invalid; or under the FMLA’s self-care provision, because Coleman v. Court of Appeals of Maryland (2012) held, through a 5-4 plurality opinion, that the self-care abrogation was invalid. Tennessee v. Lane (2004) upheld ADA Title II abrogation only as applied to court access.

Q: What did Wal-Mart v. Dukes do to workplace discrimination cases?

Wal-Mart Stores, Inc. v. Dukes, 564 U.S. 338 (2011), decertified the largest employment class ever certified, a proposed nationwide class of some 1.5 million current and former female employees across roughly 3,400 stores. On commonality, the Court held 5-4 that Rule 23(a)(2) requires a common contention capable of classwide resolution, resolving an issue central to each claim in one stroke, and found discretionary decisions by local managers lacked the necessary unifying proof. The dissent argued the majority imported predominance into commonality. Unanimously, the Court held back-pay claims improper under Rule 23(b)(2) because the monetary relief was not incidental to injunctive relief.

Q: Who counts as a supervisor in workplace discrimination law?

Under Vance v. Ball State University, 570 U.S. 421 (2013), an employee counts as a supervisor for Title VII harassment liability only when empowered to take tangible employment actions against the victim, such as hiring, firing, failing to promote, reassigning with significantly different responsibilities, or causing a significant change in benefits. The Court rejected the broader agency definition under which anyone directing another’s daily work qualified. The distinction controls the liability standard: a supervisor’s harassment with a tangible action triggers vicarious liability with no defense, a supervisor’s harassment without one opens the employer’s Faragher-Ellerth affirmative defense, and a non-supervisor’s harassment is judged under negligence, requiring proof the employer knew or should have known and failed to act.

Q: Are damages capped in workplace discrimination suits?

For the statutes that authorize compensatory and punitive damages, yes. Under 42 U.S.C. 1981a, added by the Civil Rights Act of 1991, the combined compensatory and punitive award per complaining party is capped according to employer size: $50,000 for employers with 15 to 100 employees, $100,000 for 101 to 200, $200,000 for 201 to 500, and $300,000 for more than 500. Punitive damages require malice or reckless indifference and are unavailable against government employers, while back pay and equitable relief fall outside the caps. The ADEA authorizes no compensatory or punitive damages at all, only back pay with liquidated double damages for willful violations. The FMLA similarly limits recovery to lost wages and benefits with liquidated damages absent good faith.

Q: What is the deadline for filing a charge with the EEOC?

A charge must be filed within 180 days of the alleged discriminatory act for claims under Title VII, the ADA, and the ADEA. The deadline extends to 300 days in jurisdictions where a state or local antidiscrimination law exists and an agency is authorized to grant or seek relief, reflecting work-sharing arrangements between the federal commission and local agencies. The clock generally starts on the date of the discrete act, such as a firing or demotion, not on the date the worker discovers its motivation. Missing the deadline usually forfeits the claim before its substance is examined, because courts enforce these limits strictly with only narrow equitable exceptions. The Equal Pay Act requires no agency charge at all, so its two-year court clock runs independently.

Q: What happens after you receive a right-to-sue notice?

For Title VII and ADA claims, receiving the EEOC’s notice of right to sue starts a 90-day countdown to file a lawsuit in court. Once the 90 days expire, the right to sue on that charge is lost, regardless of the claim’s strength. The short window is deliberate: it moves the dispute into court promptly after the agency process concludes. Workers should treat the notice as an urgent deadline rather than a routine letter, because courts apply the 90-day limit strictly. For ADEA claims the structure differs slightly, with suit permitted from 60 days after charge filing until 90 days after the agency completes action, but the lesson is identical. Calendar the deadline on receipt and do not assume extensions are available.

Q: Do Equal Pay Act claims require an EEOC charge?

No. The Equal Pay Act is the exception to the charge-filing rule: workers may file suit directly in federal court without first filing a charge with the Equal Employment Opportunity Commission. The commission accepts voluntary charges, but filing one does not extend the court limitations period. Suit must be commenced within two years after the cause of action accrued, or within three years for a willful violation, under 29 U.S.C. 255(a). The exception reflects the pay statute’s wage-and-hour origins rather than the commission’s charge-processing system. Workers pursuing both pay claims and Title VII claims over the same facts must still file a timely agency charge for the Title VII portion, so the direct-suit option for pay claims does not eliminate the deadline discipline the other statutes demand.

Q: What is the Faragher-Ellerth affirmative defense?

Faragher v. City of Boca Raton and Burlington Industries v. Ellerth, both decided in 1998, give employers an affirmative defense to hostile-work-environment claims where a supervisor’s harassment involved no tangible employment action. The employer must prove two elements: it exercised reasonable care to prevent and correct harassing behavior promptly, typically through a policy and complaint procedure, and the plaintiff unreasonably failed to use the preventive or corrective opportunities provided. The defense is unavailable when the supervisor took a tangible employment action such as firing or demotion, which the law treats as an official act of the enterprise. The doctrine creates incentives for employers to maintain working complaint systems and for employees to use them, since bypassing an available process can defeat an otherwise valid claim.

Q: What is the difference between disparate treatment and disparate impact?

Disparate treatment and disparate impact are the two theories of discrimination liability, and they ask different questions. Disparate treatment asks whether the employer acted with discriminatory intent, treating a worker worse because of a protected trait; the causation standards discussed in this article, motivating factor and but-for, govern this theory. Disparate impact asks whether a facially neutral employment practice, such as a test or height requirement, falls more harshly on a protected group without adequate job-related justification, regardless of intent. The employer may defend an impact claim by showing the practice is job related and consistent with business necessity. Most individual cases proceed as treatment claims, while impact theory powers challenges to uniform practices.

Q: Can the EEOC bring its own lawsuit against an employer?

Yes. The Equal Employment Opportunity Commission may file suit in federal court to enforce Title VII, the ADA, the ADEA, and the Equal Pay Act after investigating a charge and attempting conciliation. Commission suits seek relief for aggrieved individuals and broader equitable remedies, and they are not subject to the arbitration agreements individual workers signed, because the agency never agreed to arbitrate. This public-enforcement channel matters precisely because private claims are so often channeled into arbitration: the commission can pursue pattern cases in court even when the affected workers could not sue individually. The agency’s resources limit how many cases it takes, so commission litigation supplements rather than replaces private enforcement, but its independence from arbitration agreements makes it structurally distinctive.

Q: What does but-for causation mean in plain terms?

But-for causation means the outcome would not have happened without the factor in question. In employment terms, a worker proves but-for causation by showing that the adverse action, such as a firing or demotion, would not have occurred if the forbidden reason had been absent. It is a stricter test than the motivating-factor standard, under which the forbidden reason need only have played a part alongside other reasons. If an employer can show it would have taken the same action for lawful reasons alone, the but-for test fails. The Supreme Court adopted this standard for Title VII retaliation in Nassar in 2013 and for ADEA claims in Gross in 2009, reading the word because in those provisions to carry its ordinary common-law meaning.

Q: What happens if an employer would have made the same decision anyway?

Under the 1991 framework, proving the same decision would have occurred anyway does not defeat liability; it limits the remedy. If the employer proves it would have taken the same action even without the forbidden motive, the plaintiff still establishes an unlawful employment practice under 42 U.S.C. 2000e-2(m), but under 42 U.S.C. 2000e-5(g) the available relief is restricted to declaratory relief, injunctive relief, and attorney’s fees. Damages, back pay, and reinstatement are unavailable. The structure reflects a legislative judgment that establishing a discriminatory motive carries independent legal significance even when the employment outcome would have been identical. Readers sometimes assume the same-decision proof ends the case; under the statute it converts a damages case into a liability-only case.

Q: Can the EEOC sue for a worker who signed an arbitration agreement?

Yes. In EEOC v. Waffle House, Inc., 534 U.S. 279, decided January 15, 2002, the Supreme Court held that an arbitration agreement between an employer and an employee does not bar the EEOC from pursuing victim-specific judicial relief, such as back pay, reinstatement, and damages, in an enforcement action. The agency was not a party to the contract and has independent statutory authority to bring suit in federal court. The arbitration agreement binds the worker who signed it but does not bind the government enforcer. Combined with Gilmer’s holding that a worker under an arbitration agreement may still file an EEOC charge, the decision draws the boundary of the arbitration line: the private claim goes to the arbitrator while the public enforcement action stays in court.

Q: What is congruence and proportionality in state immunity cases?

Congruence and proportionality is the test the Supreme Court uses to decide whether Congress has validly abrogated state sovereign immunity under section 5 of the Fourteenth Amendment. Congress must document a pattern of unconstitutional conduct by the states and tailor the statutory remedy to that pattern. The test explains why the immunity map differs by statute: Title VII’s abrogation survived because race and sex classifications receive heightened constitutional scrutiny and Congress documented state discrimination, while the ADEA’s abrogation failed in Kimel because age receives only rational-basis review and Congress identified no sufficient pattern of unconstitutional state age discrimination. The same test produced the split FMLA result, with family-care leave sustained in Hibbs and self-care leave struck down in Coleman.

Q: Which FMLA leave types let state employees recover damages?

State employees may recover money damages from their state employer for violations of the FMLA’s family-care provision, leave to care for a child, spouse, or parent. In Nevada Department of Human Resources v. Hibbs in 2003, the Court held 6-3 that the provision validly abrogates state sovereign immunity because Congress targeted a documented pattern of sex-based stereotyping in states’ administration of leave benefits. State employees may not recover damages for violations of the self-care provision, leave for the employee’s own serious health condition. In Coleman v. Court of Appeals of Maryland in 2012, a 5-4 plurality held the self-care provision was not directed at an identified pattern of gender-based discrimination by states and therefore was not congruent and proportional enforcement legislation. The leave type determines the answer.

Q: How do liquidated damages work under the ADEA?

The ADEA incorporates the Fair Labor Standards Act’s remedial structure, under which a prevailing plaintiff recovers back pay and, where the violation is willful, liquidated damages generally computed by doubling the back-pay award. A violation is willful when the employer knew its conduct was prohibited or showed reckless disregard for whether it was prohibited. Compensatory and punitive damages are not available under the ADEA, so liquidated damages function as the statute’s enhancement for deliberate violations rather than as compensation for emotional distress or other nonpecuniary harm. The structure means an age-discrimination plaintiff’s recovery is tied to monetary loss, doubled for willfulness, while a Title VII plaintiff with the same back pay may also recover capped compensatory and punitive damages.