A Century of Shifting Burdens

In 1898, James Bradley Thayer published A Preliminary Treatise on Evidence at the Common Law. It is not a book many practicing lawyers have read, and not one many need to. Yet a great part of what we do each day in motion practice was first named with precision in its pages. Thayer separated two things that ordinary speech still runs together under the single word "burden." One is the burden of persuasion: the risk of non-persuasion, fixed at the pleadings, which does not shift. The other is the burden of production: the obligation to come forward with evidence, which may pass from one party to the other as a proceeding unfolds. And he gave us the image of the presumption as a bursting bubble, a device that shifts only the burden of production and vanishes once contrary evidence appears.

The conversation did not end with Thayer. John Henry Wigmore refined the vocabulary in his Treatise on Evidence, published in editions from 1904 to 1940, and coined the phrase "risk of non-persuasion." Edmund Morgan, in a series of Harvard Law Review articles between 1933 and 1956, challenged the orthodoxy, arguing that presumptions reflecting important social policies should shift the burden of persuasion as well. When Federal Rule of Evidence 301 took effect in 1975, it adopted Thayer's approach, Congress having rejected the Advisory Committee's Morgan-style proposal. Morgan's view did not disappear; the Institute's working taxonomy records that approximately twelve states adopted it.

I begin with this history because it is easy to forget that the frameworks we now invoke by a case name were once arguments. McDonnell Douglas, Celotex, the rule of reason, the tiers of scrutiny: each began as an answer to a contested question, and each carries within it a judgment about who should bear the risk of being wrong. The question before us is not whether we can recite these frameworks. Every competent lawyer can. The question is whether we understand what we have inherited well enough to use it faithfully, and whether we will take the trouble to pass it on intact.

What a burden-shifting framework is

Stripped to its essentials, a burden-shifting framework is an ordered allocation of responsibility under uncertainty. It says which party must speak first, what that party must show, what showing moves the obligation to the other side, and, beneath all of that, which party loses if at the end the matter remains in doubt. The last question is answered by the burden of persuasion, and on Thayer's account it does not move. What moves is the burden of production, the obligation to put something before the tribunal at each stage.

This distinction, once grasped, explains a remarkable amount. It explains why the employer's burden at the second step of McDonnell Douglas is a burden to articulate a legitimate reason, not to prove it. It explains why Rule 301 provides, in terms, that a presumption does not shift the burden of persuasion, "which remains on the party who had it originally." And it explains why, at summary judgment, what a moving party must show depends so heavily on which party would bear the burden of persuasion at trial.

The inheritance, family by family

The Institute's working taxonomy speaks of thirty-two distinct frameworks spanning employment discrimination, constitutional law, civil procedure, evidence, administrative law, antitrust, securities regulation, criminal procedure, and more. Its master table runs somewhat longer, and it includes, alongside the living frameworks, one the Supreme Court has overruled and one that the taxonomy itself later explains shifts no burden at all. I note this not in criticism but because it illustrates the point of this essay: an inventory of an inheritance must be kept, and kept honestly, or it will quietly mislead those who rely on it.

Consider the principal families, roughly in the order of their arrival.

Tort. The oldest line runs to Byrne v. Boadle, decided in England in 1863, which gave us res ipsa loquitur: an inference of negligence where the event does not ordinarily occur without negligence, the instrumentality was within the defendant's exclusive control, and the plaintiff did not contribute. California carried the idea further. Ybarra v. Spangard (1944) shifted the burden to every defendant who had control over an unconscious patient. Summers v. Tice (1948) shifted the burden on causation to defendants where the plaintiff could not identify which of them caused the harm. Barker v. Lull Engineering (1978) placed on the defendant, under the risk-utility test for design defect, the burden of proving that benefits outweigh risks. Sindell v. Abbott Laboratories (1980) required each defendant to prove it did not manufacture the product that caused the injury. The taxonomy rightly adds that the majority of jurisdictions treat res ipsa as a permissive inference only, leaving the burden of proof with the plaintiff throughout.

Antitrust. Standard Oil Co. v. United States (1911) established the rule of reason. United States v. Socony-Vacuum Oil Co. (1940) supplied the per se rule, relieving the plaintiff of proving effects in price-fixing cases. United States v. Philadelphia National Bank (1963) created a structural presumption for mergers. The quick look doctrine followed in NCAA v. Board of Regents (1984) and California Dental Ass'n v. FTC (1999). And Ohio v. American Express Co. (2018) and NCAA v. Alston (2021) clarified the modern three steps: the plaintiff proves substantial anticompetitive effects, the defendant proffers procompetitive justifications, and the plaintiff shows a less restrictive alternative.

Constitutional scrutiny. From footnote four of United States v. Carolene Products Co. (1938) came the idea that certain classifications and rights warrant more searching review. Under strict scrutiny the government bears the burden of proving a compelling interest pursued by narrowly tailored means. Craig v. Boren (1976) established intermediate scrutiny, and United States v. Virginia (1996) demanded an "exceedingly persuasive justification." Rational basis review presumes validity and places on the challenger the burden of negating every conceivable basis, though in City of Cleburne v. Cleburne Living Center, Romer v. Evans, and United States v. Windsor the Court examined actual rather than hypothetical reasons where animus was evident. Most recently, New York State Rifle & Pistol Association v. Bruen (2022) placed on the government, at its second step, the burden of demonstrating historical analogues for a firearms regulation.

Employment discrimination. McDonnell Douglas Corp. v. Green (1973), which the taxonomy describes as the template for burden-shifting across several doctrinal areas, set out three steps: the plaintiff establishes a prima facie case, the employer articulates a legitimate, nondiscriminatory reason, and the plaintiff proves pretext. Texas Department of Community Affairs v. Burdine (1981) made clear that only the burden of production shifts, and St. Mary's Honor Center v. Hicks (1993) and Reeves v. Sanderson Plumbing Products (2000) defined what proof of pretext requires. Around this core grew the mixed-motive framework of Price Waterhouse v. Hopkins (1989), codified and modified by the Civil Rights Act of 1991; the pattern-or-practice framework of Teamsters v. United States (1977); and the cat's paw doctrine of Staub v. Proctor Hospital (2011). The Court has also drawn boundaries. Gross v. FBL Financial Services (2009) held that the mixed-motive framework does not apply to age discrimination claims, and University of Texas Southwestern Medical Center v. Nassar (2013) required but-for causation for Title VII retaliation.

Criminal law. In re Winship (1970) requires the prosecution to prove the elements of an offense beyond a reasonable doubt. Patterson v. New York (1977) held that a state may place on the defendant the burden of proving an affirmative defense by a preponderance of the evidence. Under the federal subjective test for entrapment in Jacobson v. United States (1992), the defendant shows inducement, and the government must then prove predisposition beyond a reasonable doubt.

Summary judgment. In 1986 the Celotex trilogy (Celotex Corp. v. Catrett, Anderson v. Liberty Lobby, and Matsushita Electric Industrial Co. v. Zenith Radio Corp.) made summary judgment under Rule 56 a regular instrument of case disposition rather than a disfavored device. A movant that does not bear the burden of persuasion at trial may discharge its initial burden by pointing out the absence of evidence for the nonmoving party's case; the burden of production then passes to the nonmovant to identify specific facts showing a genuine dispute. The Institute's strategy guide on the motion records that approximately thirty-nine states have adopted the federal standard, Florida becoming the thirty-ninth on May 1, 2021. California has not. Under Aguilar v. Atlantic Richfield Co. (2001) and Code of Civil Procedure section 437c, a moving defendant must either show that the plaintiff does not have and cannot reasonably obtain needed evidence, or present affirmative evidence negating an element.

Qualified immunity, securities, and immigration. Harlow v. Fitzgerald (1982) adopted the objective "clearly established law" standard; Saucier v. Katz (2001) imposed a mandatory two-step sequence, which Pearson v. Callahan (2009) abandoned. Under the Private Securities Litigation Reform Act of 1995, as construed in Tellabs, Inc. v. Makor Issues & Rights (2007), a plaintiff must plead facts giving rise to a strong inference of scienter; Basic Inc. v. Levinson (1988) presumes reliance in efficient markets, and Halliburton Co. v. Erica P. John Fund (2014) allows defendants to rebut that presumption with price impact evidence at class certification. In immigration, the government must prove deportability by clear, unequivocal, and convincing evidence (Woodby v. INS, 1966), while an asylum applicant must establish a well-founded fear of persecution (INS v. Cardoza-Fonseca, 1987), a lower showing than the more-likely-than-not standard for withholding of removal (INS v. Stevic, 1984).

Administrative law. Skidmore v. Swift & Co. (1944) gave agency views weight according to their power to persuade. Motor Vehicle Manufacturers Ass'n v. State Farm (1983) required agencies to examine the relevant data and give a reasoned explanation. Chevron U.S.A., Inc. v. NRDC (1984) placed on challengers, in effect, the burden of showing an agency's reading of an ambiguous statute unreasonable. In June 2024, Loper Bright Enterprises v. Raimondo overruled Chevron, restoring the courts' independent judgment as to a statute's best reading. Skidmore now serves as the primary framework, and the major questions doctrine of West Virginia v. EPA (2022) requires clear congressional authorization for agency action of vast economic and political significance. The consequences of that change are the subject of the essay that follows this one.

What the inheritance teaches

Set these families side by side and a pattern appears that no single framework discloses. They differ enormously in subject, yet they are built from the same few materials: Thayer's two burdens, an ordered sequence of showings, and a judgment about which party is better placed to bear the cost of uncertainty. In rational basis review that judgment favors the legislature; in strict scrutiny, the individual. In McDonnell Douglas it asks the employer only to articulate a reason, while never relieving the plaintiff of the ultimate burden. These are not technical accidents. They are the considered conclusions of generations of judges and lawyers about how a fair tribunal should proceed when it cannot know everything.

The second lesson is less comfortable. The inheritance is not fixed. Saucier gave way to Pearson. The mixed-motive framework of Price Waterhouse was modified by statute and then confined by Gross and Nassar. The taxonomy records that Dobbs v. Jackson Women's Health Organization (2022) overruled both Roe v. Wade and Planned Parenthood v. Casey, so that abortion regulations now receive rational basis review and the burden rests with the challenger. Chevron, which governed for forty years, is gone. A lawyer who learned these frameworks in school and has not revisited them may cite, with complete confidence, a framework the courts have withdrawn.

Where inheritors go astray

The working taxonomy catalogs the errors that recur in briefing, and I commend the full list to every practitioner. A few deserve mention here, because each is a failure of stewardship rather than of intelligence.

The most common is to write that "the burden shifts" without saying which burden, inviting a court to assume that the burden of persuasion has moved when only the burden of production has. A second is to treat every multi-factor test as a burden-shifting framework; the four factors of Winter v. NRDC (2008) for a preliminary injunction remain the movant's burden throughout, and nothing shifts. A third is to apply mixed-motive analysis to age discrimination or retaliation claims, where the Court has said it does not belong. A fourth is to go on invoking Chevron after Loper Bright, or the Casey undue burden standard after Dobbs. A fifth is to describe Rule 301 as shifting the burden of proof, when its text says precisely the opposite.

A sixth, recorded in the strategy guide, is to carry the federal Celotex approach into California state court, where Aguilar requires more of a moving defendant. Such a motion may cite only accurate authority and still misstate the movant's burden in the forum where it is filed. The Institute's library calls this a burden mismatch and treats it at length in Right Law, Wrong Stage.

None of these errors involves a fabricated case. Each involves a real framework applied at the wrong stage, to the wrong burden, or in the wrong court, or a framework applied after the courts have withdrawn it. That is what makes them dangerous. They survive a check of citations because the citations are genuine.

The inheritance and the new instruments

It would be incomplete to close without acknowledging that the profession now drafts, in part, with the assistance of machines. The Institute's program in adversarial reasoning asks how such systems should model the other side's best case and allocate burdens. The question is not a narrow technical one. A system that cannot say which burden it is describing, whose burden it is, at what stage, in which court, and under which version of a framework has not mastered the vocabulary Thayer set out more than a century ago. It will reproduce, and at speed, the very confusions the taxonomy catalogs.

The remedy is not to distrust such instruments by reflex. It is to hold them to the standard we should hold ourselves: precision about the burden, fidelity to the forum, and currency of the framework.

A duty of stewardship

The burden of persuasion, Thayer taught, is fixed and does not shift. I would suggest that a lawyer's responsibility for the frameworks he or she invokes is of the same character. It cannot be shifted to a treatise, a template, a junior colleague, or a tool.

That responsibility has a practical form. State which burden you mean, every time. Identify the party that bears it and the stage at which you are speaking. Confirm that the framework you cite remains the law, and that it is the law of your forum rather than another's. And teach the distinction between production and persuasion to those who come after you, for otherwise they will learn "burden" as a single word and carry the error into every brief they write.

We are heirs to a body of doctrine built by people who thought hard about fairness under uncertainty. The least we owe them, and the clients who depend on us, is to use what they built with the care they brought to building it.