Burden of proof.

• Author: Kaplow, Louis

ARTICLE CONTENTS INTRODUCTION I. OPTIMAL EVIDENCE THRESHOLD A. Framework 1. Context 2. Evidence Threshold B. Analysis II. CONVENTIONAL CONCEPTIONS COMPARED A. Conventional Burden of Proof B. Comparison 1. Analysis 2. Comment on Information Requirements C. Elaboration 1. Relationship Between Level of Evidence Threshold and Ex Post Likelihood 2. Alternative Ex Post Objectives D. Comparison Revisited: Theory Versus Practice III. INTERACTIONS BETWEEN THE BURDEN OF PROOF AND OTHER FEATURES OF THE LEGAL SYSTEM A. Enforcement Effort B. Sanctions 1. Level 2. Nonmonetary Sanctions C. Accuracy IV. ALTERNATIVE METHODS OF ENFORCEMENT A. Monitoring and Auditing B. Investigation C. Regulation of Future Conduct 1. Optimal Evidence Threshold 2. Conventional Conceptions Compared 3. Mixed Cases D. Selection of Cases for Adjudication 1. Public Enforcement 2. Private Enforcement CONCLUSION INTRODUCTION

Systems of adjudication base outcomes on whether the strength of available evidence satisfies a designated burden of proof. Similar modes of decisionmaking are employed in many important nonlegal settings, such as when medical treatments are selected in light of test results and other diagnostics. The stringency of the proof burden determines how error is allocated between mistakes of commission--improper assignment of liability--and mistakes of omission--improper exoneration.

This Article explores how to set the evidence threshold (1) in the manner that best advances social welfare. (2) It seeks less to displace entrenched views than to fill a vacuum, for prior scholarship devotes surprisingly little attention to the rationale for how stringent proof burdens should be. In the United States, civil litigation ordinarily is governed by a preponderance of the evidence rule, under which the plaintiff must establish that it is more likely than not that the defendant is liable; (3) in criminal trials, guilt must be established beyond a reasonable doubt. (4) Neither requirement is particularly controversial, and perhaps for this reason many treatises and texts on the law of evidence and on civil and criminal procedure provide little or no explanation for these rules. By contrast, somewhat greater attention has been devoted to the production burden (5) (how much evidence a party must present in order to avoid losing by default) and tiebreaking (6) (which party wins in a civil case when the factfinder believes there is a fifty percent chance that each side is correct)--subjects that are not the focus here.

When the question is broached directly, commonly proffered justifications are as unsatisfying as they are brief. In the civil context, the primary argument for the preponderance rule seems to be a lack of any apparent reason to do otherwise, a view often expressed by noting the absence of the special features of the criminal context. (7) Some authorities go so far as to express society's negligible concern as to the outcome, (8) which if pressed might lead one to wonder why civil litigation should exist at all (or why the authors bother to produce elaborate texts and treatises on how it should be conducted, while others write extensively on what substantive rules should govern). It is hard to avoid the conclusion that the strong attraction of the 50% requirement is substantially attributable to its being a powerful focal point, some of its power deriving from there being no other focal points--besides 0% and 100%, neither of which has any appeal. (9)

For criminal cases, the reasoning is more readily identified, mention often being made of the high stakes that make false convictions particularly problematic. (10) Even here, elaboration is rare and the conclusions are not obvious: High stakes make erroneous acquittals more troublesome as well; note that multiplying the consequences on both sides of a balance by a common factor has no effect on which way the scale tips. (11) Moreover, as will be elaborated in the body of this Article, stricter proof burdens can, in plausible settings, increase rather than decrease the number of false convictions, and the presence of higher social costs of sanctions likewise has ambiguous implications regarding whether the proof burden should be higher or lower. (12) The point of these observations is not that existing evidence thresholds are too high or too low in either the civil or criminal settings but rather that current thinking--actually, fairly old thinking that has been repeated but not much reconsidered--provides an insufficient basis for addressing the question. (13)

To further motivate the inquiry, consider the application of conventional conceptions of the burden of proof in adjudication to the context of medical decisionmaking. (14) Would we say that, because society per se--as distinct from the patient--has no special stake in most medical treatment decisions, the rule for whether to have surgery, receive chemotherapy, or be subject to some other procedure should be based on a fifty percent probability? Or that, if the stakes are high (say, intrusive surgery for a very serious medical condition), it is better to let ten diseased patients go untreated than to mistakenly undertake surgery on one who did not need it? Or, perhaps the opposite, that it is better to let ten nondiseased patients suffer side effects than to let one diseased individual go untreated? Such formulations are confused, indeed absurd. Surely in setting the treatment threshold we should weigh, along with the likelihood that the patient truly has the condition, the expected benefit of the treatment to those who need it (would they otherwise die? what are the side effects and other costs? are there alternatives?) and the expected cost to those who in fact do not. And if we were observing doctors and patients making medical treatment decisions, we would expect to see them balancing the consequences of the possible outcomes--treatment and abstention--for the truly sick and for those who only appear to be ill, factoring in the likelihoods of each.

Legal systems, like medical systems, are of great importance. Civil and criminal law underlie social order, playing an essential role in facilitating economic activity, ensuring public safety, and otherwise promoting social welfare. Accordingly, decision criteria for adjudication--the setting of proof burdens in various legal contexts--should rest on a stronger foundation than age-old dicta. They should instead be grounded in explicit analysis that attends to the consequences of legal outcomes: correct and mistaken imposition of liability as well as proper and erroneous exoneration.

The pertinent effects are familiar in rough terms even though they have not systematically informed discourse on how proof burdens ought to be set. For individuals who might commit harmful acts, the prospect of liability produces deterrence whereas the possibility of mistaken absolution dilutes it. Deterrence of wrongful conduct is central not only to controlling crime but also to inducing individuals to perform contracts, comply with environmental and health regulations, avoid careless activity, and interact honestly in the marketplace. The control of harmful behavior is, of course, the raison d'etre for the legal system, and it is crucial to consider how the system's ability to achieve this objective is affected by how high the burden of proof is set.

The strength of the burden of proof also determines the magnitude of the legal system's major negative consequence aside from direct operational costs: namely, erroneous assignments of liability. A nontrivial likelihood that those who commit benign acts might be found liable will tend to chill desirable behavior. Although chilling receives less explicit attention in general treatments of the legal system and in discussions of the burden of proof, it is nevertheless a significant consideration in most areas of substantive law. Just as the anticipation of proper assignment of liability encourages mutually beneficial contractual interaction, concerns for mistakes may discourage it. Misapplication of antitrust law may chill aggressive competition or useful collaboration; with securities law, costs of raising capital may be inefficiently elevated; with medical malpractice, doctors may avoid high-risk patients or treatments and otherwise engage in defensive medicine. (15)

Part I of this Article investigates in a basic setting how to set the evidence threshold optimally in light of deterrence and chilling effects. (16) Slightly reducing the evidence threshold will increase the probability that individuals who contemplate the commission of harmful acts would expect to be subject to sanctions, and likewise for those who contemplate benign acts. (17) The extent of each effect depends on the nature of the evidence in a manner that is described. It also is necessary to determine how many harmful and benign acts are deterred and chilled, respectively, on account of these increases in expected sanctions. This magnitude is shown to depend on how many individuals have private benefits for the two types of acts that approximately equal the corresponding expected sanctions. Finally, for given numbers deterred and chilled, the effects on social welfare need to be assessed. In both cases, when acts are discouraged, individuals' private benefits from these acts are forgone; in the case of harmful acts, society is spared their negative consequences. The calculus for determining the optimal evidence threshold is, on reflection, conceptually straightforward and in accord with welfare-based intuitive reasoning. As a practical matter, however, the factors are many, their magnitudes undoubtedly vary greatly across contexts, and ascertaining the pertinent quantities is likely to be difficult.

Part II juxtaposes this welfare-based analysis with conventional thinking about the burden of proof, taking standard formulations on their face, in idealized form, until the closing...