Academia.edu no longer supports Internet Explorer.
To browse Academia.edu and the wider internet faster and more securely, please take a few seconds to upgrade your browser.
The Case for Varying Standards of Proof
Gustavo RibeiroRelated Papers
This Essay focuses not on how fact-finders process evidence but on how they apply the specified standard of proof to their finding. The oddity that prompts speculation is that, in noncriminal cases, the common law asks only that the fact appear more likely than not, while the Civil Law seems to apply the same high standard in these cases as it does in criminal cases. As a psychological explanation of the cognitive processes involved, some theorists posit that the bulk of fact-finding is an unconscious process, powerful but dangerous, which generates a level of confidence against which the fact-finder could apply the standard of proof. But this foggy confidence-based theory fails because standards of proof should, and factfinders arguably do, concern themselves with probability rather than confidence. Psychology also cannot explain the divide between the common law and the Civil Law because the real explanation likely lies in the different goals that the two procedural systems pursue...
2013 •
This Essay focuses not on how fact-finders process evidence but on how they apply the specified standard of proof to their finding. The oddity that prompts speculation is that, in noncriminal cases, the common law asks only that the fact appear more likely than not, while the Civil Law seems to apply the same high standard in these cases as it does in criminal cases. As a psychological explanation of the cognitive processes involved, some theorists posit that the bulk of fact-finding is an unconscious process, powerful but dangerous, which generates a level of confidence against which the fact-finder could apply the standard of proof. But this foggy confidence-based theory fails because standards of proof should, and factfinders arguably do, concern themselves with probability rather than confidence. Psychology also cannot explain the divide between the common law and the Civil Law because the real explanation likely lies in the different goals that the two procedural systems pursue...
2010 •
During the course of legal proceedings, evidentiary material is analyzed and evaluated in order to make a final judgement whether the responsible party has discharged the onus of proof. The existence of a standard of proof against which the presiding officer can measure the evidence submitted consequently plays a pivotal role. This standard of proof (bewysmaatstaf) represents the standard of guilt in legal science and has also been described as a standard of conviction. The standard of proof does not pertain to the inherent qualities of evidentiary material, but rather to the degrees of conviction of the presiding officer in a particular case. The function of the standard of proof is furthermore to provide presiding officers with a guideline/yardstick to measure the degree of conviction that the general public believe the presiding officer should have over the correctness of all the factual conclusions in the particular proceedings. In this article, the standard of proof in law will be discussed from a comparative point of view; different standards of proof from different jurisdictions will be considered and juxtaposed against similar standards used in the natural sciences.
The International Journal of Evidence & Proof
The civil standard of proof—what is it, actually2016 •
Unlike common law, Continental European civil law does not strictly distinguish between the standards of proof applicable in civil and criminal matters, respectively. In civil law countries such as Germany and Switzerland, judges are supposed to use the same (high) standard of 'full conviction' in both criminal and civil cases. This study is the first to look at the standard of proof actually used by judges and judicial clerks in a civil law country (Switzerland). It is shown that the standard actually used differs from the one normatively prescribed. No difference between the estimated decision threshold for members of the courts and members of the general population is found. The results suggest that the standard of proof actually employed by judges of a civil law system is not much different from the common law's 'preponderance of the evidence' standard, despite the doctrinal insistence to the contrary.
Academics have never quite understood the standards of proof or, indeed, much about the theory of proof. Their formulations beget probabilistic musings, which beget all sorts of paradoxes, which in turn beget radical reconceptions and proposals for reform. The theoretical radicals argue that the law needs some basic reconception such as recognizing the aim of legal proof as not at all a search for truth but rather the production of an acceptable result, or that the law needs some shattering reform such as greatly heightening the standard of proof on each part of the case to ensure a more-likely-than-not overall result.This Article refutes all those baroque re-readings. It shows that the standards of proof, properly understood on the law’s own terms without a probabilistic overlay, work just fine. The law tells fact finders to compare their degree of belief in the alleged fact to their degree of contradictory disbelief. Following that instruction resolves mathematically the paradoxes...
Could it be right to convict and punish defendants using only statistical evidence? In this paper, I argue that it is not and explain why it would be wrong. This is difficult to do because there is a powerful argument for thinking that we should convict and punish defendants using statistical evidence. It looks as if the relevant cases are cases of decision under risk and it seems we know what we should do in such cases (i.e., maximize expected value). Given some standard assumptions about the values at stake, the case for convicting and punishing using statistical evidence seems solid. In trying to show where this argument goes wrong, I shall argue (against Lockeans, reliabilists, and others) that beliefs supported only by statistical evidence are epistemically defective and (against Enoch, Fisher, and Spectre) that these epistemic considerations should matter to the law. To solve the puzzle about the role of statistical evidence in the law, we need to revise some commonly held assumptions about epistemic value and defend the relevance of epistemology to this practical question.
Jurisprudence
Standards of proof as competence norms2022 •
In discussions of standards of proof, a familiar perspective often emerges. According to what we call specificationism, standards of proof are legal rules that specify the quantum of evidence required to determine that a litigant's claim has been proven. In so doing, they allocate the risk of error among litigants (and potential litigants), minimizing the risk of certain types of error. Specificationism is meant as a description of the way the rules actually function. We argue, however, that its claims are either mistaken or at a minimum deeply misleading, especially when it comes to standard of proof rules (SPRs) that contain indeterminate formulas, as is typical. As against specificationism, we argue that SPRs are best understood as rules that confer competence to decide whether a given standard has been metaccording to whatever vague or inchoate interpretation (if any) of the rule in question triers of fact implicitly or explicitly employ. We call this the competence-norm approach.
This short paper provides an argument for clarifying the standard of proof used in criminal courts within the US and other nations. I do not suggest lowering or increasing the standard, at least not explicitly. Whether it is decided that the Blackstone ratio is acceptable or a different one it should be a bright line rather than the vague standard we currently have in the United States. Using empirical evidence, the paper shows the massive variety of answers people give when faced with phrases such as ‘beyond a reasonable doubt.’ Because of this I argue that the instruction given should be a specific percentage before you can return with a guilty verdict.
Journal of Applied Philosophy
In Defence of Reasonable Doubt2017 •
In criminal trials the state must establish, to a particular standard of proof, the defendant's guilt. The most widely-used and important standard of proof for criminal conviction is the " beyond a reasonable doubt " standard. But what legitimates this standard, rather than an alternative? One view holds the standard of proof should be determined or justified—at least in large part—by its consequences. In this spirit, Laudan uses crime statistics to estimate risks the average citizen runs of being violently victimised and falsely convicted. He argues that since the former risk is higher, and the aggregate harms are worse, the standard of proof should be substantially lowered. He presents a formula for calculating the preferred standard. In this paper I outline various ways Laudan's uses of crime statistics are flawed, and explain how he substantially overestimates risks of victimhood and underestimates costs of false convictions. I also explain why his formula is mistaken, and illuminate consequences Laudan neglects. I conclude that, even if consequences determine the appropriate standard of proof, Laudan's arguments fail to show the standard is too high. I conclude by suggesting that the inadequacies of Laudan's reasoning might be good news for consequence-based justifications of the standard of proof. ________ 1. The Standard of Proof The state sometimes performs actions against the wishes or interests of individual citizens. They might stop and search an individual, charge him with a crime, incarcerate him, impose a restraining order or curfew, remove child custody rights, or order him to pay costs or fees, such as child support. Standards of proof are levels of confidence required to perform these actions legitimately. Before the state may perform these actions they must determine, to that confidence level, certain facts to be the case. 1 Employing public, clearly understood standards of proof serves a range of functions, including bolstering citizens' trust in the system and protecting individuals from undue loss of liberty. There are several standards of proof in use across the world. They vary in demandingness according to the seriousness of the state's action—weaker standards are employed for less severe actions—and the consequences risked by not acting. Formal standards include the " reasonable suspicion " standard in the US, which is required for a brief stop and search. 2 To satisfy this standard the police officer must have a specific, articulable, and individualised suspicion that crime is afoot before they conduct a brief search of an individual. The " preponderance of evidence " standard—usually understood as requiring that the claim is more likely than not—is operative in many Western democracies for most civil and family court cases. 3 The " clear and convincing evidence " standard is even more demanding. It is
San Diego Law Review
Volume 56 | Issue 1
Article 7
3-1-2019
The Case for Varying Standards of Proof
Gustavo Ribeiro
Follow this and additional works at: https://digital.sandiego.edu/sdlr
Part of the Law Commons, and the Legal Theory Commons
Recommended Citation
Gustavo Ribeiro, The Case for Varying Standards of Proof, 56 San Diego L. Rev. 161 (2019).
Available at: https://digital.sandiego.edu/sdlr/vol56/iss1/7
This Article is brought to you for free and open access by the Law School Journals at Digital USD. It has been accepted for inclusion in San Diego Law
Review by an authorized editor of Digital USD. For more information, please contact digital@sandiego.edu.
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
The Case for Varying Standards of Proof
GUSTAVO RIBEIRO*
TABLE OF CONTENTS
I.
II.
III.
IV.
V.
VI.
VII.
INTRODUCTION ........................................................................................ 162
BACKGROUND: THE ERROR-DISTRIBUTION FUNCTION OF
STANDARDS OF PROOF ............................................................................. 165
A PROBLEM AND A PROPOSAL .................................................................. 174
NORMATIVE ARGUMENTS IN FAVOR......................................................... 183
A. Welfare Considerations ................................................................. 183
B. Fairness Considerations ................................................................ 189
C. Distributive Considerations ........................................................... 194
A POSITIVE ARGUMENT IN FAVOR............................................................ 198
OBJECTIONS AGAINST AND REPLIES ......................................................... 207
A. High Administrative Costs and Unpredictable Effects................... 207
B. Cognitive Limitations..................................................................... 212
C. Already Varying De Facto Standards of Proof.............................. 216
CONCLUSION ........................................................................................... 219
*
© 2019 Gustavo Ribeiro. Boston University School of Law. Thank you
to Ronald J. Allen, Scott Brewer, Ali Butler, Hasan Dindjer, Rachel Herdy, Louis Kaplow,
Gary Lawson, Nadav Orian Peer, Andrea J. Schweitzer, Alex Stein, Alex Whiting, and
William Twining for their helpful comments and suggestions on previous drafts and to the
San Diego Law Review team for the terrific editorial work. For useful discussions, I am
grateful to participants at the Byse Workshop, the S.J.D. Colloquium, and the Law and
Philosophy Society at Harvard Law School, the Jurisprudence Discussion Group at the
University of Oxford, and the Legal Theory Seminar Series at the University of
Edinburgh.
161
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
I. INTRODUCTION
The current evidence rules group cases that ought to remain separate for
a myriad of reasons. Most dramatically, we assign the same standard of
proof to drastically different cases under the justification that we accept,
or should accept, the same error-distribution for those cases.1 To justify
the use of the “preponderance of the evidence” standard in most civil
litigation cases, courts and scholars frequently repeat that—in the majority of
civil cases—society is, or should be, indifferent to errors in favor of plaintiffs
or defendants.2 The inescapability of the beyond “reasonable-doubt standard”
in criminal law is defended as embodying a constant defendant-friendly
error-distribution across all criminal adjudication.3 However ubiquitous, these
justifications are twice mistaken.
First, they are normatively mistaken. There are important welfare, fairness,
and distributive considerations in favor of varying standards of proof across
different types of cases. Analyses based on welfare considerations tell us that
if the benefit from deterring harmful acts is greater than the loss from
chilling benign acts, then we have a strong argument that reducing the
standard of proof might be socially desirable; whereas if the reverse is
true, an increase might be preferable.4 Under this view, the socially optimal
1. See, e.g., Santosky v. Kramer, 455 U.S. 745, 745–46 (1982) (“In any given
proceeding, the minimum standard of proof tolerated by the due process requirement
reflects not only the weight of the public and private interests affected, but also a societal
judgment about how the risk of error should be distributed between the litigants.”);
Addington v. Texas, 441 U.S. 418, 423 (1979) (“The standard serves to allocate the risk
of error between the litigants and to indicate the relative importance attached to the
ultimate decision.”); In re Winship, 397 U.S. 358, 369–70 (1970) (Harlan, J., concurring)
(“[E]ven though the labels used for alternative standards of proof are vague and not a very
sure guide to decisionmaking, the choice of the standard for a particular variety of
adjudication [reflects] a very fundamental assessment of the comparative social costs of
erroneous factual determinations.”).
2. Addington v. Texas, 441 U.S. 418, 423 (1979) (“Since society has a minimal
concern with the outcome of [monetary disputes between private parties,] plaintiff’s
burden of proof is a mere preponderance of the evidence. The litigants thus share the risk
of error in roughly equal fashion.”).
3. Winship, 397 U.S. at 363 (“The reasonable-doubt standard plays a vital role in
the American scheme of criminal procedure. It is a prime instrument for reducing the risk
of convictions resting on factual error. The standard provides concrete substance for the
presumption of innocence—that bedrock ‘axiomatic and elementary’ principle whose
‘enforcement lies at the foundation of the administration of our criminal law.’” (quoting
Coffin v. United States, 165 U.S. 432, 453 (1895)).
4. See, e.g., Louis Kaplow, Burden of Proof, 121 YALE L.J. 738, 855–56 (2012)
[hereinafter Kaplow, Burden of Proof]; sources cited infra note 81. Alternatively, for a
discussion against the idea of a monolithically social and an argument that there are only
competing interests, values, and preferences, in constant combat, differentially aided by
law and legal processes, see, for example, Oliver Wendell Holmes, Jr., Privilege, Malice,
162
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
standard occurs where these two opposing tendencies—deterrence of harmful
acts, on the one hand, and chilling of benign acts, on the other—offset each
other. Most importantly, we have good reason to believe these two forces
vary drastically in different cases. This suggests the optimal standard also
varies with different types of cases.
Taking fairness as the guiding consideration, there is a strong argument that
people have a right to procedures that assign the appropriate importance to
the risk of being unjustly found liable or guilty.5 If this risk varies in
different cases because, for instance, we value the injustices that follow
from wrongful convictions or acquittals differently—such as capital punishment
versus minor misdemeanors—it should follow that people have a right in
different cases to different legal procedures that weigh the risk of an unjust
verdict. Standards of proof are perhaps the most salient example of legal
procedures with that effect.6
Distributive considerations push us to understand the levels of the
standards of proof as part of people’s background legal entitlements that
influence the societal distribution of wealth and income.7 This suggests
we should not ignore the disparate effects on the distribution of wealth
and income that a system that assigns the same standard of proof across
the board has on differently situated plaintiffs and defendants. We also
gain an important insight for programmatic projects. Legal reforms intended
to alter the distribution of wealth or income among different social groups
can utilize standards of proof to reach their objectives.
Second, the justifications for grouping different cases under the same
standard of proof is positively mistaken. Research on jury decision-making
suggests that, under certain circumstances, jurors tend to reduce the de
facto standard of proof necessary for conviction.8 We should adjust our
and Intent, 8 HARV. L. REV. 1, 3 (1894); Karl N. Llewellyn, A Realistic Jurisprudence—
The Next Step, 30 COLUM. L. REV. 431, 461–62 (1930).
5. See RONALD DWORKIN, A MATTER OF PRINCIPLE 3, 119–21 (1985); HO HOCK
LAI, A PHILOSOPHY OF EVIDENCE LAW: JUSTICE IN THE SEARCH FOR TRUTH 213, 223–24
(2008).
6. See DWORKIN, supra note 5, at 84–88, 92–93.
7. See, e.g., Duncan Kennedy, Distributive and Paternalist Motives in Contract
and Tort Law, with Special References to Compulsory Terms and Unequal Bargaining
Power, 41 MD. L. REV. 563, 563, 565 (1982) [hereinafter Kennedy, Distributive and
Paternalist]; Duncan Kennedy, Cost-Benefit Analysis of Entitlement Problems: A Critique, 33
STAN. L. REV. 387, 422–23 (1981) [hereinafter Kennedy, Cost-Benefit Analysis].
8. See, e.g., DENNIS J. DEVINE, JURY DECISION MAKING: THE STATE OF THE SCIENCE
39–40 (2012); REID HASTIE ET AL., INSIDE THE JURY 231 (1983).
163
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
discourse to match reality. Not only is a system with greater variation in the
number of standards of proof normatively desirable but it is also consistent
with how jurors decide cases. Just like we weigh different circumstances
and cases through sanctions, liability, and procedural and evidentiary rules,
we can, and should, also express these differences by using different standards
of proof in different types of cases.
This proposal will undoubtedly face objections. One might be concerned
about the potentially high administrative costs associated with a system
of varying standards of proof.9 Although an important concern for any
policy proposal, we should be careful not to overestimate this point. A
related objection pertains to the difficulty of predicting the dynamic effects
of a system with varying standards of proof.10 Once a new standard is in
place, people will modify their behavior to conform to the new standard.
As such, we would need to predict avoidance and exploitation efforts and
continuously adjust our standards. But these continuous adjustments
would require an immense amount of information. For objectors, the twin
problems of high costs and unpredictable dynamic effects will most likely
offset any potential gain resulting from a system with varying standards of
proof. These two objections touch on a more general point about the
function of standards serving as mechanisms for distributing errors.
To make sure a specific distribution of errors actually occurs in our
society, we need information concerning, among other things, “the
distribution of truly guilty and truly innocent [people] who go to trial.”11
This information, however, is prohibitively costly, or impossible, to obtain.
One way to change our understanding of how standards can satisfactorily
distribute errors is by lowering our expectations about our capacity to
verify whether we have reached an appropriate error-distribution. We
can also see standards of proof as prescriptive generalizations with a
relevant underlying justification—namely to achieve a given error distribution.
Similar to legal rules, standards of proof are suboptimal generalizations.
But unlike legal rules, this suboptimality does not come from over- and
9. See KEVIN M. CLERMONT, STANDARDS OF DECISION IN LAW: PSYCHOLOGICAL
LOGICAL BASES FOR THE STANDARD OF PROOF, HERE AND ABROAD 264–68 (2013);
Ronald J. Allen & Alex Stein, Evidence, Probability, and the Burden of Proof, 55 ARIZ.
L. REV. 557, 580–81 (2013).
10. See Kaplow, Burden of Proof, supra note 4, at 809, 855.
11. LARRY LAUDAN, TRUTH, ERROR, AND CRIMINAL LAW: AN ESSAY IN LEGAL
EPISTEMOLOGY 73 (2006) (citing Ronald J. Allen, The Restoration of In Re Winship: A
Comment on Burdens of Persuasion in Criminal Cases After Patterson v. New York, 76
MICH. L. REV. 30, 47 n.65 (1997) [hereinafter Allen, Winship]); Ronald J. Allen, Rationality,
Algorithms and Juridical Proof: A Preliminary Inquiry, 1 INT’L J. EVIDENCE & PROOF 254,
260 (1997) [hereinafter Allen, Juridical Proof]; Michael S. Pardo, Second-Order Proof
Rules, 61 FLA. L. REV. 1083, 1088 (2009).
AND
164
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
under-inclusiveness. Rather, it comes from constraints on our ability to
verify whether the targeted error distribution has been achieved.
Another objection I address is the claim that the legal system already
adjusts the distribution of error across different types of cases through
other legal mechanisms, such as adding or removing causes of action for
certain subsets of cases. Insofar as these other instruments impact how
easy it is to impose liability on potential defendants, they can cause the de
facto standard of proof to vary. After considering the effects of altering
the de facto standard using a procedural or substantive strategy, I conclude
that the procedural strategy is either superior or equally preferable to the
substantive strategy, which gives us a good reason to prefer the former.
This suggests it will often be best to alter the error-distribution to a socially
desirable ratio through procedural instruments.
This Article is divided into five additional parts. Part II discusses the
two functions usually attributed to standards of proof. Special focus is
placed on the standards serving as a mechanism to distribute the risk of
factual error between plaintiffs and defendants. Part III suggests that once
we seriously consider this function of standards, we can conclude that we
group cases under a restricted number of standards. This Article argues
for a system with a greater number of standards of proof that inform legal
decision-making. Part IV considers welfare, fairness, and distributive
arguments in favor of varying standards of proof. Part V presents a positive
argument in favor of the proposal. Lastly, Part VI responds to potential
objections.
II. BACKGROUND: THE ERROR-DISTRIBUTION FUNCTION OF
STANDARDS OF PROOF
Standards of proof are a tricky evidentiary mechanism. Although they
are ubiquitous in legal decision-making, there is widespread disagreement
about how exactly to understand them.12 This lack of understanding greatly
12. Empirical research suggests considerable confusion regarding standards exists
in practice, both with respect to articulation and communication of the standards of proof
by trial judges and to comprehension and application by the juries. See, e.g., Joel D.
Lieberman, The Psychology of the Jury Instruction Process, in 1 PSYCHOLOGY IN THE
COURTROOM: JURY PSYCHOLOGY 129, 129, 132, 139–40 (Daniel A. Krauss & Joel D.
Lieberman eds., 2009) (reviewing the existing literature); see also CLERMONT, supra note
9, at 113; Dorothy K. Kagehiro, Defining the Standard of Proof in Jury Instructions, 1
PSYCHOL. SCI. 194, 196, 198 (1990); Federico Picinali, Is “Proof Beyond a Reasonable
Doubt” a Self-Evident Concept? Considering the U.S. and the Italian Legal Cultures
165
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
affects our capacity to articulate and communicate standards in a precise
and defensible manner. Existing formulations are often described as obscure,
incoherent, or worse.13 This disagreement, in turn, reveals a lack of familiarity
with the functions standards are meant to serve.
Standards of proof are thought to serve essentially two related functions.14
First, they are decision-making thresholds—standards tell us whether the
available evidence in favor or against a particular evidentiary hypothesis
offers enough evidential support for us to take a given proposition as
proven for some particular purpose.15 Consider medical drug testing as
an example.16 Scientists and public health policymakers face a difficult
tradeoff when deciding whether to release drugs to the public. On the one
hand, they want to make drugs publicly available as quickly as possible to
prevent the spread of diseases. On the other hand, they want to be confident
that the drug is effective and does not cause severe side effects. The problem
is that, to be sure of a drug’s effectiveness and safety, scientists and
policymakers must expend time and resources that might not be available.
To resolve this tension between speed and safety, scientists and
policymakers usually establish a decision-making threshold.17 If the
available data offers some level of support for the proposition that the drug
is effective and safe, then they are willing to accept the proposition that
Towards the Understanding of the Standard of Persuasion in Criminal Cases, GLOBAL
JURIST, 2009, at 1, 11, 26 (finding empirical studies that show widespread misunderstanding of
the importance of beyond a reasonable doubt); Frederick Schauer, Slippery Slopes, 99
HARV. L. REV. 361, 370–73 (1985) (discussing the challenges involved in formulating and
communicating precise linguistic formulations of legal principles); David U. Strawn &
Raymond W. Buchanan, Jury Confusion: A Threat to Justice, 59 JUDICATURE 478, 478–
80 (1976) (suggesting that after instructions only 50% of jurors understand that the
defendant did not have to present any evidence of his innocence and that the state had to
establish the defendant’s guilt beyond reasonable doubt).
13. See LAUDAN, supra note 11, at 61, 64.
14. The discussions that follow are premised on the idea that the applicable standard
of proof has significant implications for how the guilty and the innocent fare in the judicial
system. Given that an overwhelming majority of cases settle before trial, see, for example,
Daniel Epps, The Consequences of Error in Criminal Justice, 128 HARV. L. REV. 1065,
1114 (2015), one might wonder whether standards are merely a sideshow. One wellknown answer is to contend that standards are still important because settlements and pleas
occur in the shadow of trial outcomes. See Stephanos Bibas, Plea Bargaining Outside the
Shadow of Trial, 117 HARV. L. REV. 2463, 2464–66 (2004); Oren Gazal-Ayal & Avishalom
Tor, The Innocence Effect, 62 DUKE L.J. 339, 383–86, 388–96 (2012); Alec Walen, Proof
Beyond a Reasonable Doubt: A Balanced Retributive Account, 76 LA. L. REV. 355, 401–
02 (2015). This article proceeds on this assumption.
15. See LAUDAN, supra note 11, at 64.
16. See, e.g., MELISSA LEACH & JAMES FAIRHEAD, VACCINE ANXIETIES: GLOBAL
SCIENCE, CHILD HEALTH AND SOCIETY 17 (2007).
17. See id.
166
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
the drug is effective and safe.18 In other words, scientists and policymakers
establish a minimal threshold of evidential support that the evidence must
reach to consider a given proposition as proven. Essentially, they establish
a standard of proof.
A second, related, function of standards of proof courts and scholars
often highlight—and the one I focus on—is that of distributing errors.19
To understand this different, yet connected, function let us start with a
question: Why would anyone want to establish a standard of proof higher
than more probable than not? After all, if a proposition is probably more
true than not, then, under epistemic justification, we are at least prima
facie rationally justified in believing that proposition.20 Why require
something more demanding than what rationality allows?
To answer this question, let us return to the drug testing example.
Suppose you oversee a massive governmental drug-development program.
You collect data from the most prestigious laboratories, and the results
clearly seem to show that drug X is only probably more effective and safer
than not. Under normal conditions—other than a catastrophic epidemic—
would you authorize the drug program based on this evidence alone? I
believe most people would not. Most people weigh the cost of the error
of releasing a drug that is ineffective or unsafe more heavily than they do
the cost of the error of delaying the release of a safe and effective drug.
Weighing different errors differently is not a phenomenon specific to
public health situations. Something similar takes place in the legal system.
Because we cannot completely eliminate errors from an adjudicative
system with limited resources, society must decide which errors are more
serious and worth spending more resources to avoid. Standards of proof
are the best example of legal mechanisms primarily concerned with this
18.
19.
See id.
See supra note 1; see also MICHAEL O. FINKELSTEIN, QUANTITATIVE METHODS
IN LAW: STUDIES IN THE APPLICATION OF MATHEMATICAL PROBABILITY AND STATISTICS TO
LEGAL PROBLEMS 67 (1978); LAUDAN, supra note 11, chs. 2–3; ALEX STEIN, FOUNDATIONS
OF EVIDENCE LAW 138 (2005) (“[E]vidential rules and principles affiliating to [the AngloAmerican systems of evidence] have a single all-important function: allocation of the risk
of error.”); Richard S. Bell, Decision Theory and Due Process: A Critique of the Supreme
Court’s Lawmaking for Burdens of Proof, 78 J. CRIM. L. & CRIMINOLOGY 557, 563, 579–
81 (1987); Michael Tigar, The Supreme Court, 1969 Term, 84 HARV. L. REV. 1, 158 n.13
(1970) (“[T]he reasonable doubt standard seeks to assure that erroneous acquittals of the
guilty are far more common than the erroneous convictions of the innocent.”).
20. See generally, e.g., EARL CONEE & RICHARD FELDMAN, EVIDENTIALISM: ESSAYS
IN EPISTEMOLOGY (2004).
167
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
task. If we could compute the relative costs and benefits of all accurate
and erroneous convictions and acquittals in every decision, we might be
able to use this cost ratio to determine the height of the standard.21
One way to visualize this function of standards is through graphs like
the one depicted in Figure 1 below.22 Such illustrations show the relations
between the standard of proof and the distribution of the types of errors
we expect to occur. The horizontal axis represents a spectrum of probabilities
that fact-finders, either judges or jurors, assign to defendants’ “apparent guilt.”
At one end are cases that tend to go to trial with very weak evidence—for
instance, cases involving private, solo conduct with low externalities and
low probability of detection. At the other end are cases that tend to go to
trial with very strong evidence against defendants, such as public daylight
conduct with many witnesses, externalities, and high detection rates. The
vertical axis is the relative frequency or proportion of cases that exhibit
apparent guilt.23
We are interested here in two distributions: one representing truly innocent
defendants and another representing truly guilty defendants.24 Each
distribution represents the likely values of apparent guilt for the two types
of defendants.25 It is important to clarify at the outset a few important
assumptions. Absent a highly dysfunctional legal system, we should expect
that truly guilty defendants will, on average, have higher apparent guilt than
truly innocent defendants. As a consequence, the distribution of truly guilty
defendants is represented to the right of the truly innocent distribution on
the horizontal spectrum. However, the exact location of each distribution
21. LAUDAN, supra note 11, at 68.
22. See infra p. 169. For examples of other works using similar diagrams to represent
the relation between standards of proof and the distribution of the risk of error between
parties, see LAUDAN, supra note 11, 66–70; Ronald J. Allen, Burdens of Proof, 13 LAW,
PROBABILITY & RISK 195, 201, 204–06, 211–12 (2014); Bell, supra note 19, at 570–82;
Michael L. DeKay, The Difference Between Blackstone‐Like Error Ratios and Probabilistic
Standards of Proof, 21 LAW & SOC. INQUIRY 95, 101, 105, 113, 120, 123–25 (1996).
23. This makes the area under each curve equal to 1—representing 100% of the
data—and explains why both distributions are shown with similar sizes.
24. Some might object to the use of the terms guilty and innocent. Under this view,
guilty and innocent represent ascriptions that only come at the end of a specifically legal
procedure. See, e.g., Zenon Bankowski, The Value of Truth: Fact Scepticism Revisited, 1
LEGAL STUD. 257, 265 (1981). To not confuse readers familiar with the literature, I use
these terms simply to adopt the same terminology used by other works on this topic. One
could use easily alter the names of both curves without harm to the lessons that follow.
25. Another way to understand these distributions is as conditional distributions
reflecting the likely values of apparent guilt—given that the defendant is truly innocent or
truly guilty. DeKay, supra note 22, at 101–02. We can then derive the unconditional
probabilities—for instance, that a randomly selected defendant is acquitted and truly
innocent—from the conditional probabilities. See id. All one must do is to adjust the
conditional probabilities to reflect the mix of truly guilty and innocent defendants
appearing before the court. See id.
168
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
[VOL. 56: 161, 2019]
Varying Standards of Proof
SAN DIEGO LAW REVIEW
along the x-axis is not crucial for our purposes. Also, although the graph
depicts both distributions using normal curves, normality is not an
essential feature of this analysis. “Other unimodal distributions [can] yield
similar results . . . .”26 Another important assumption is that the distributions
overlap. This is because of significant variability within each group. Some
truly innocent defendants have a higher apparent guilt than some truly guilty
defendants.
igure 1
Truly Innocent
Truly Guilty
Figure 1 does not show how many findings will be correct or mistaken,
nor does it show how the errors will be distributed. We can illustrate
setting up a standard of proof by drawing a line perpendicular to the
horizontal axis. That line marks how strong the admitted evidence supporting
a conclusion about liability must be to legally justify a conviction or
finding of liability. More specifically, it shows how strong the admitted
evidence supporting the case of the party carrying the burden of persuasion
must be. Figure 2 represents a possibility.
26.
Id. at 102.
169
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
igure 2
Standard of Proof
Truly Innocent
Truly Guilty
This mode of representation is visually appealing because the areas of
these regions are closely related to the probabilities and frequencies of the
four possible outcomes of a verdict. The area under the “Truly Guilty”
curve and to the right of the standard of proof line represents the probability
that a truly guilty individual will be convicted at trial. The area under the
“Truly Guilty” curve to the left of the standard—the small light gray
area—indicates the probability of a truly guilty individual being mistakenly
acquitted at trial. Conversely, the probability that a truly innocent individual
goes to trial and is mistakenly convicted is shown by the larger dark gray
area under the “Truly Innocent” curve to the right of the standard line. For
all other values to the left of that point, the individual is correctly found
not liable.
Presumably, we want to minimize the likelihood that a truly innocent
individual is mistakenly found liable. Holding other features of the legal
system constant, such as level of enforcement, sanctions, et cetera, one
way we can attempt to reduce mistaken liability verdicts is by increasing
the minimal level of evidential support required for a liability conclusion.
Graphically, this means moving Figure 2’s standard of proof line further
to the right. This would reduce the dark gray area under the “Truly Innocent”
curve, translating to a reduction in the probability of truly innocent defendants
mistakenly found liable.27 Figure 3 illustrates this.
27. This point seems intuitively correct. Raising the standard of proof will likely
increase the percentage of truly guilty individuals that are convicted out of the entire set
170
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
Standard of Proof
Truly Innocent
Truly Guilty
Note, however, that with the same stroke we would also reduce the
likelihood of correctly assigning liability to a truly guilty individual. In
other words, when determining the optimal standard, the errors of false
positives versus the errors of false negatives are traded off. The greater
the likelihood that we will correctly assign liability to truly guilty individuals,
of convicted individuals. This is particularly true if we hold constant the frequency with
which cases come before the court—keeping the “Truly Innocent” and “Truly Guilty”
curves intact as Figures 2 and 3 illustrate. However, depending on the empirical assumptions
we start with, we might also expect the exact opposite effect. The problem is that there is
a good dose of endogeneity in this kind of analysis—the number of innocent and guilty
individuals that go to trial seems to be a function of, among other things, the level of the
standard of proof. With a higher standard of proof, a conviction is more difficult, which
lowers the expected costs of sanctions for individuals—some of whom might become
more likely to commit the act that brings them to court, mistakenly or not. The question
then becomes whether we have reason to believe we will see a corresponding increase in
the frequency of cases that come before the court. If more cases happen to go to trial as a
result of an increase in the standard of proof, this might reduce the chances that a higher
percentage of individuals convicted are truly guilty, the opposite of what seems intuitively
correct. The situation, then, seems to be as follows: On the one hand, we have the effect
that an increase in the standard of proof increases the chances that a higher percentage of
individuals convicted are truly guilty. On the other hand, the fact that we should expect
the frequency for both cases to increase might reduce this likelihood. The problem is that
these two effects might run in opposite directions and if so, we would have a hard time
knowing which effect is stronger and thus more likely to prevail. See Kaplow, Burden of
Proof, supra note 4, at 789–805.
171
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
the greater the likelihood that we will mistakenly assign liability to truly
innocent individuals. On the flipside, lowering the standard of proof—
moving the line in Figures 2 or 3 to the left—increases the likelihood that
fact-finders will correctly assign liability to truly guilty individuals while
concurrently increasing the likelihood that truly innocent individuals will
mistakenly be found liable. In either direction, there is a trade-off.28
If we value one type of error much more than the other, why not simply
set the standard at the highest level possible to eliminate virtually all
chances of convicting an innocent? The main problem with this proposal
is that it ignores the costs associated with both types of errors. If the
standard is set too high, we risk making it nearly impossible to convict not
only the truly innocent, but also the truly guilty. Setting too high a standard
has nontrivial consequences, including: (1) losses in deterrence, potentially
causing an increase in crime rates; (2) losses in feelings of justice, because
a wrong that was committed is not remedied; and (3) losses in trust in the
legal system, because victims are denied retribution.29 This important point
is often overlooked. Legal scholarship and practice tend to concentrate
solely on the costs following wrongful convictions. Consequently, it has
ignored—or grossly underestimated—the costs following wrongful acquittals.
These costs are real, and there is certainly a limit to how much we will accept
to prevent the flipside of convicting an innocent. Once society concludes
these costs are too high, it can adjust the applicable standard of proof to
bring about a more desirable tradeoff. This is ultimately a matter of public
policy and choices must inevitably be made. There are no pre-determined
socially desirable ratios.30
28. This trade-off is partially the result of a significant variability within each
group. Some truly innocent defendants appear guiltier than some truly guilty defendants.
DeKay, supra note 22, at 101. This is visually represented by the overlap of the two
curves. Because appearance of guilt is imperfectly related to true guilt, fallible decisionmakers will invariably fail to correctly identify all innocent or guilty defendants. See id.
The level of variability between both groups—the overlap size—has great importance.
Anything that makes truly guilty defendants appear guiltier or makes innocent defendants
appear less guilty moves the curves apart and increases accuracy. See, e.g., DeKay, supra
note 22, at 101–02.
29. See Ronald J. Allen & Larry Laudan, Deadly Dilemmas, 41 TEX. TECH L. REV.
65, 68, 73 (2008); see also Ronald J. Allen & Larry Laudan, Deadly Dilemmas II: Bail
and Crime, 85 CHI.-KENT L. REV. 23, 25–27 (2010); Ronald J. Allen & Larry Laudan,
Deadly Dilemmas III: Some Kind Words for Preventive Detention, 101 J. CRIM. L. &
CRIMINOLOGY 781, 801–02 (2011).
30. Even after including the costs of wrongful acquittal, the picture is still incomplete.
It is not enough to weigh only the costs of erroneous decisions, whether convictions or
acquittals. Accurate convictions and acquittals must also be weighed in reaching any
decision about the proper standard of proof. See, e.g., Ronald J. Allen, How to Think About
Errors, Costs, and Their Allocation, 64 FLA. L. REV. 885, 889 (2012); DeKay, supra note
22, at 95; Laurence H. Tribe, Trial by Mathematics: Precision and Ritual in the Legal
172
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
The error-distribution function of standards should not be overstated,
however. That standards of proof serve as mechanisms for distributing
errors is not to say standards are the only force at play in determining the
error-distribution. The relationship between standards and the distribution of
different types of errors is actually more complicated than it first seems.
The precise allocation of errors also depends on factors other than the
standard, including the frequency of truly guilty and truly innocent that go
to trial and the probability distribution of the cases that get to trial where
the admitted evidence is stronger than the applicable standard.31 In other
words, the error-distribution also depends on how the “Truly Innocent”
and “Truly Guilty” curves look.32
One last point before getting into the arguments for varying standards
of proof. It is commonly thought that the function of legal fact-finding is
primarily an epistemic one of getting the facts right.33 But after studying
evidence, one quickly notes that the epistemic function of fact-finding cannot
be the whole story. Still, accuracy remains a prime goal of the evidentiary
system. Therefore, a worthy question is: How does the error distribution
function of standards connect to the epistemic function of fact-finding?
One hypothesis is that the level of the standard is directly correlated to
the level of accuracy in fact-finding.34 So, if we raise the former, we also
raise the latter. No matter how intuitive this hypothesis might sound, it is
not quite right. The confusion results mostly because certain changes in the
legal system can affect both accuracy and standards. Consider the effects of
providing additional resources for indigent defendants’ counsels, or similarly
reducing their caseload. One expected result is improving the overall
Process, 84 HARV. L. REV. 1329, 1381–82 (1971). We should set standards with our eyes
on the relative costs and benefits of both accurate and erroneous convictions and acquittals.
31. See Allen, Winship, supra note 11, at 47 n.65 (“Without knowing the distribution of
guilt probabilities of factually innocent and guilty defendants, we cannot know the actual
effect of choosing one standard of proof over another.”). For concrete examples, see
FINKELSTEIN, supra note 19, at 59–60. But see Bell, supra note 19, at 574–75 (conveying
different results concerning preponderance of the evidence, but no clear and convincing
evidence and beyond a reasonable doubt).
32. See supra Figures 1–3. For a more in-depth examination, see infra Section VI.A.
33. See Tehan v. United States, 382 U.S. 406, 416 (1966) (“The basic purpose of a
trial is the determination of truth . . . .”).
34. For discussions about the value of accuracy in fact-finding, see, for example,
Kaplow, Burden of Proof, supra note 4, at 741, 746; Louis Kaplow, The Value of Accuracy in
Adjudication: An Economic Analysis, 23 J. LEGAL STUD. 307, 307–08 (1994) [hereinafter
Kaplow, Accuracy in Adjudication]; Jack B. Weinstein, Some Difficulties in Devising
Rules for Determining Truth in Judicial Trials, 66 COLUM. L. REV. 223, 228–29 (1966).
173
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
quality of evidence admitted at trial. For example, counsels for indigent
defendants will have more time to prepare better defenses and will be able
to hire investigators and forensics experts. On the one hand, this might
correlate with more accurate fact-finding. Better evidence should provide
more support for evidentiary propositions that are more likely to be true.
On the other hand, increasing resources for counsels might also cause a
de facto increase in the level of the standards of proof.35 Because defendant
counsels will be able to prepare better defenses, we should expect that on
average it will be harder to convict defendants. This is an example of a change
that has the potential to affect both accuracy and the level of standards.
We can also envision changes in the legal system that could affect one but
not the other. Suppose that in addition to increasing the resources available
to indigent defendants’ counsels, we simultaneously increase resources
available to prosecutors, effectively restoring the prior de facto standard.
These examples indicate the importance of separating accuracy and the
level of the standard is not merely conceptual. Although both work together
in many instances, we can design changes that would affect one but not
the other. This point is important for my purposes. Below, I argue that
the legal system should have more flexibility to impose varying standards
of proof. This means sometimes increasing, but also sometimes lowering,
the applicable standard. Regardless, it should be clear from the outset that
lowering the standard does not necessarily mean a decrease in accuracy.
III. A PROBLEM AND A PROPOSAL
This Part argues that there is a problem with the rules currently governing
standards of proof. We group cases under a very restricted number of
standards. This problem requires a solution. The solution I explore in this
Article requires moving in the direction of a system with a greater number
of standards of proof to inform legal decision-making.
Civil courts often repeat they do not have a reason to favor error for one
party over the other.36 This perceived equality of the relative importance
of errors is the driving force behind the preponderance of the evidence
standard in almost all civil cases in the United States.37 It is also a common
justification for a specific formulation of that standard—namely that the
decision-maker should decide if the plaintiff’s case is “more likely than
35. Kaplow, Accuracy in Adjudication, supra note 34, at 356–57.
36. See, e.g., Addington v. Texas, 441 U.S. 418, 423 (1979).
37. See, e.g., Ronald J. Allen, Burdens of Proof, Uncertainty, and Ambiguity in
Modern Legal Discourse, 17 HARV. J.L. & PUB. POL’Y 627, 633–34 (1994); Ronald J.
Allen, The Error of Expected Loss Minimization, 2 LAW, PROBABILITY & RISK 1, 3 (2003);
Pardo, supra note 11, at 1088–89; Lawrence B. Solum, Procedural Justice, 78 S. CAL. L.
REV. 181, 312 (2004).
174
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
not,” which is often articulated as having a greater than 50% probability.38
By setting the standard that way, under certain assumptions, the preponderance
rule roughly aims to divide the risk of error equally among the parties.39
There are significant exceptions, however. These include certain types
of civil lawsuits in which errors favoring plaintiffs are more significant or
costly than errors favoring defendants. Consequently, the risk of error is
asymmetrically skewed against plaintiffs. In these cases, the “clear and
convincing evidence” standard applies to one or more elements of the plaintiff’s
claim. Examples include fraud,40 civil commitment,41 deportation,42
denaturalization,43 termination of parental rights,44 decisions to terminate
life,45 and proof of malice in defamation cases involving public figures.46
However, these exceptions are restricted in number and scope. The
general belief is that in civil cases—in which the applicable standard is
38. Richard A. Posner, An Economic Approach to Legal Procedure and Judicial
Administration, 2 J. LEGAL STUD. 399, 408 (1973). It is important to note that a preponderance of
the evidence standard does not mean that only roughly half of the verdicts will be accurate.
Id. This is because an increase or decrease in either
the plaintiff’s stakes or the effectiveness of his litigation expenditures . . . will
induce the plaintiff to spend at a higher rate than the defendant, and vice versa.
If the allegations essential to one party’s claim are in fact true, ordinarily it will
be easier for him to prove them than for his opponent to disprove them, assuming
they spend the same amount of money on the trial.
Id. at 431. “[T]he effectiveness of the expenditures of the party with the meritorious claim
will be high relative to the effectiveness of his opponent’s expenditures. This will induce
the first party to spend more heavily on litigation than the second.” Id. at 431–32. But
within the errors, we should expect an equal number of plaintiffs and defendants. Id.
at 431–35.
39. See, e.g., Grogan v. Garner, 498 U.S. 279, 286 (1991) (“[T]he preponderanceof-the-evidence standard results in a roughly equal allocation of the risk of error between
litigants . . . .”). This is roughly—not exactly—because the defendant wins if there is a
tie. See id. One justification for this is based on an idea that there is a certain value to the
status quo and that, when a plaintiff brings a legal claim, it disrupts the status quo, so the
plaintiff should bear the risk of error when the evidence is in equipoise. See Herman &
MacLean v. Huddleston, 459 U.S. 375, 390 (1983). But see CLERMONT, supra note 9, at
17–18 (dismissing this as no more justified than believing the defendant disrupts the status
quo).
40. See 12 JOSEPH T. MCLAUGHLIN & THOMAS D. ROWE, JR., MOORE’S FEDERAL
PRACTICE § 60.43 (3d ed. 1997).
41. E.g., Addington v. Texas, 441 U.S. 418, 427, 431 (1979).
42. E.g., Woodby v. INS, 385 U.S. 276, 277 (1966).
43. E.g., Schneiderman v. United States, 320 U.S. 118, 123 (1943).
44. E.g., Santosky v. Kramer, 455 U.S. 745, 750, 769 (1982).
45. E.g., Cruzan v. Mo. Dep’t of Health, 497 U.S. 261, 265 (1990).
46. E.g., N.Y. Times Co. v. Sullivan, 376 U.S. 254, 279–80, 285–86 (1964).
175
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
preponderance of the evidence—we do not have a reason to favor errors
for one party over the other.47 Otherwise, the background assumptions in
most civil trials include “that the litigants’ resources and abilities are
approximately equal” and that in half the cases plaintiffs and defendants
are each right about the disputed issues.48 These assumptions, however,
are fundamentally flawed.
Consider the following hypothetical cases. On the one hand, a breach
of contract dispute between Amazon and FedEx. On the other hand, a
janitor suing Wal-Mart for racial discrimination. Now ask yourself: In the
janitor’s case, is it realistic to expect there to be an equality of resources
and abilities between the parties? Assume further that Wal-Mart has been
guilty for not paying overtime and other types of compensations and
benefits before. Some might question whether the janitor case should be
subject to the same error-distribution as the breach of contract case between
two large corporations. This, however, is what occurs when we enforce
the same standard of proof in both cases. These two examples illustrate
situations in which the desirable error-distributions are likely to vary.
This, in turn, suggests that the current evidentiary system fails to account
properly for the relevant differences between different types of cases.
Cases might also differ for reasons other than economic disparity.
Parties can have disparate knowledge of the relevant facts.49 In a world
with symmetrical information and low transaction costs, our assumptions
about parties’ equality of resources and abilities might not carry much weight.
However, in a world with informational and economic asymmetry and
where discovery costs can be extraordinary, these assumptions become
crucial. Ultimately, applying a rule that is based on assumptions that fail
to hold in the real world can lead to unanticipated results. Instead of
having errors equally distributed between parties, different litigants can
end up carrying a larger share of the risk than what is socially desirable.
This problem is not unique to civil cases. In fact, in criminal law, the
situation is even more extreme. There, not only are there widely diverse
cases under only one standard of proof—beyond reasonable doubt—but
also the Supreme Court has an elevated requirement to constitutional
status. In Winship—and other cases addressing the Due Process Clauses
of the Fifth and Fourteenth Amendments as a necessary element of a
47. See, e.g., Addington v. Texas, 441 U.S. 418, 427, 431 (1979).
48. Bell, supra note 19, at 570. But see generally Marc Galanter, Why the “Haves”
Come Out Ahead: Speculations on the Limits of Legal Change, 9 LAW & SOC’Y REV. 95
(1974) (showing how “repeat players” can litigate cases and ultimately shape the development
of law in more favorable fashions than “one shotters”).
49. Ronald J. Allen, A Reconceptualization of Civil Trials, 66 BOS. U.L. REV. 401,
427 (1986).
176
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
legally justified conviction in criminal adjudication—the Court made
clear that the prosecution must establish every key element of their case
beyond a reasonable doubt.50 Justice Frankfurter went so far as to say that
the standard of proof in criminal cases plays “a vital role in the American
scheme of criminal procedure.”51
Behind the acquittal-friendly character of the beyond a reasonable
doubt standard is the idea that mistaken convictions entail significantly
higher costs than mistaken acquittals.52 The traditional view is that, because
of this idea, the beyond a reasonable doubt standard needs to be set high
across the entire criminal justice system.53 Once we look closer, however,
the idea that we value all mistaken convictions equally sounds normatively
inadequate and descriptively inaccurate. Normatively, heavyweight
considerations push us towards the desirability of enforcing different error
distributions among substantially distinct types of cases.54 Descriptively,
social science research on jury decision-making suggests jurors already
change decision-making thresholds in determining whether to condemn
or acquit defendants based on a set of perceived relevant differences.55
Consider a recently decided case where the Massachusetts Supreme
Judicial Court reviewed which standard of proof the Massachusetts Sex
50. 397 U.S. 358, 363 (1970); see also Allen, Winship, supra note 11, at 39, 48–49.
This principle actually goes back to the eighteenth century. See BARBARA J. SHAPIRO,
BEYOND REASONABLE DOUBT AND PROBABLE CAUSE: HISTORICAL PERSPECTIVES ON THE
ANGLO-AMERICAN LAW OF EVIDENCE 79 (1991); Bruce A. Antkowiak, Judicial Nulification,
38 CREIGHTON L. REV. 545, 560 (2005). The reasonable doubt standard applies across the
board in the criminal justice system; it applies to misdemeanors and felonies, all degrees
of offenses, and even in cases without a jury. See United States v. Randolph, 93 F.3d 656,
660 (9th Cir. 1996) (first citing Jackson v. Virginia, 443 U.S. 307, 319 (1979); and then
citing United States v. Mayberry, 913 F.2d 719, 720 (9th Cir. 1990)).
51. Winship, 397 U.S. at 363 (“The reasonable-doubt standard plays a vital role in
the American scheme of criminal procedure. It is a prime instrument for reducing the risk
of convictions resting on factual error. The standard provides concrete substance for the
presumption of innocence—that bedrock ‘axiomatic and elementary’ principle whose
‘enforcement lies at the foundation of the administration of our criminal law.’” (quoting
Coffin v. United States, 156 U.S. 432, 453 (1895))).
52. See SHAPIRO, supra note 50, at 20–25; Erik Lillquist, Recasting Reasonable
Doubt: Decision Theory and the Virtues of Variability, 36 U.C. DAVIS L. REV. 85, 89 (2002);
Anthony A. Morano, A Reexamination of the Development of the Reasonable Doubt Rule,
55 B.U. L. REV. 507, 509, 519 (1975).
53. This view is repeated by those on both sides of the legal spectrum. See Posner,
supra note 38, at 411, 413; see also RICHARD A. EPSTEIN, FORBIDDEN GROUNDS: THE CASE
AGAINST EMPLOYMENT DISCRIMINATION LAWS 225 (1992).
54. See infra Part IV.
55. See infra Part V.
177
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
Offender Registry Board (SORB) had to satisfy to classify a convicted sex
offender under sex offender registry laws.56 The plaintiff, a convicted sex
offender, had been previously classified as having a moderate risk of reoffense by a preponderance of the evidence—the prevailing standard of
proof according to state precedent at the time.57 The court agreed with the
plaintiff that recent amendments to the sex offender registry laws and other
developments made the preponderance standard inadequate and concluded:
[D]ue process requires that a sex offender’s risk level be proved by clear and
convincing evidence. The risk classification that SORB must make now has
consequences for those who are classified that are far greater than was the case
when we decided Doe No. 972. The preponderance standard no longer adequately
protects against the possibility that those consequences might be visited upon
individuals who do not pose the requisite degree of risk and dangerousness.58
The important aspect of that case, for our purposes, is how the court
reached its decision to increase the required standard of proof. Although
the court noted it previously held that “the ‘possible injury to sex offenders
from being erroneously overclassified’ was ‘nearly equal’ to ‘any harm to
the State from an erroneous underclassification,’” it emphasized that
significant revisions to sex offender registry laws and other developments
imposed extra burdens on registered offenders without providing additional
protections.59 First, amendments had expanded the number of offenses
that required registration.60 According to the court, a higher number of
56. See Doe v. Sex Offender Registry Bd. (Doe 380316 II), 41 N.E.3d 1058, 1071–
72 (Mass. 2015). SORB is responsible for establishing and maintaining a “central
computerized registry of all sex offenders required to register” pursuant to state legislation.
MASS. GEN. LAWS ch. 6, § 178D (2018).
The file on each sex offender required to register [includes]:
(a) the sex offender’s name, aliases used, date and place of birth, sex, race,
height, weight, eye and hair color, social security number, home address,
any secondary addresses and work address and, if the sex offender works
at or attends an institution of higher learning, the name and address of the
institution;
(b) a photograph and set of fingerprints;
(c) a description of the offense for which the sex offender was convicted or
adjudicated, the city or town where the offense occurred, the date of conviction
or adjudication and the sentence imposed;
(d) any other information which may be useful in assessing the risk of the sex
offender to reoffend; and
(e) any other information which may be useful in identifying the sex offender.
Id.
57. Doe 380316 II, 41 N.E.3d at 1060.
58. Id. at 1061.
59. Id. at 1064 (quoting Doe v. Sex Offender Registry Bd. (Doe 380316 I), 697
N.E.2d 512, 520 n.14 (1998)).
60. Id.
178
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
offenses increased the risk of errors in the SORB’s task of classifying
specific offenders’ “risk of re-offense.”61
Second, sex offenders faced increasingly rigorous reporting requirements
and harsher sanctions for failing to meet such requirements.62 For
instance, offenders were required to register secondary addresses, the names
and addresses of the educational institutions they attend, and any new
addresses ten days prior to moving as well were subjected to “intensive
parole conditions.”63 If an offender failed to meet such requirements, he
could be fined; however, if a judge determined imprisonment was a more
appropriate sanction for failing to comply with registration requirements,
the judge was bound to impose a mandatory minimum sentence of no less
than six months.64
Third, the court noted how offenders faced increasing difficulties based
on their registered sex offender status.65 These included limited ability to
find work and housing, denial of access to federal social programs, as well
as a vast range of other restrictions, such as a ban on “engaging in ice cream
truck vending, regardless of whether their offense involved harm to a child”
and a ban on living in a nursing home.66
Fourth, the court focused on the fact that information about registered
offenders was broadly disseminated.67 For instance, both level two and
level three sex offenders’ information is posted online.68 For the Justices,
“[w]here previously the time and resource constraints of local police departments
set functional limits on the dissemination of registry information, the
Internet allows for around-the-clock, instantaneous, and worldwide access
to that information–a virtual sword of Damocles.”69 This meant that if—
after being classified as a level two or three offender and consequently
having personal information posted online—an offender was later reclassified
to level one such that the offender was no longer required to post his
information online, the damage done to the offender’s personal right would
61.
62.
63.
64.
65.
66.
67.
68.
69.
Id.
Id. at 1065.
Id.
Id. at 1065–66 (citing MASS. GEN. LAWS ch. 6, § 178H (2010)).
See id. at 1070.
Id. at 1066.
Id. at 1064.
See Moe v. Sex Offender Registry Bd., 6 N.E.3d. 530, 538–39 (Mass. 2014).
Doe 380316 II, 41 N.E.3d at 1067 (citing Moe, 6 N.E.3d at 536–37).
179
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
be difficult to remediate.70 This is particularly true because information
posted online is hardly, if ever, truly withdrawn from the internet. Lastly,
the court also raised a concern about whether the SORB’s guidelines for
classifying offenders accurately reflected current scientific knowledge.71
In light of these factors, and after considering the effects of a new
standard of proof in terms of more false negatives, the court determined
the distribution of errors under the preponderance standard was tilted too
sharply in favor of the government. For the court, such offenders “should
not be asked to share equally with society the risk of error.”72 A higher
standard of proof was necessary to preserve their due process rights and
restore what the court saw as the proper balance of the distribution of
risk.73
Other cases might also benefit from higher standards of proof. In death
penalty cases, the costs associated with wrongful conviction might be so
high as to justify a standard of proof requiring “nearly absolute certainty.”74
Where the repercussions from the conviction go beyond the formal sentence,
the relative disutility that flows from an erroneous conviction may be
sufficiently high so that the resulting standard of proof may be even more
than beyond a reasonable doubt.75 Consider also cases involving lowlevel criminality, such as traffic offenses or other minor misdemeanors
that would not result in any jail time; in such cases, the cost of an erroneous
conviction seems lower compared to felony charges with long mandatory
minimum sentences.
On the flipside, crimes where the costs from an erroneous acquittal are
relatively high may require lower standards of proof. Consider a case
involving a charge of terrorist acts where the defendant is a known public
supporter of such practices and associated groups.76 Here, the risk of harm
70. Id. (quoting Robert Kirk Walker, Note, The Right to Be Forgotten, 64 HASTINGS
L.J. 257, 259–60 (2012)).
71. See id. (citing Doe v. Sex Offender Registry Bd. (Doe 68549), 18 N.E.3d 1081,
1092 (Mass. 2014)). This case is made even more interesting because it was decided at
the same time as the federal government was requiring many universities around the
country to lower the standard of proof in sexual harassment and sexual violence cases to
preponderance of the evidence pursuant to Title IX, the federal law that prohibits gender
discrimination in educational institutions that receive federal education funds. See
CATHERINE E. LHAMON, OFFICE FOR CIVIL RIGHTS, U.S. DEP’T OF EDU., QUESTIONS AND
ANSWERS ON TITLE IV AND SEXUAL VIOLENCE 5, 26 (2014) (“The school must use a
preponderance-of-the-evidence . . . standard in any Title IX proceedings . . . .”). The contradictory
nature of both state-sanctioned recommendations in closely related cases is puzzling.
72. Doe 380316 II, 41 N.E.3d at 1071 (quoting Addington v. Texas, 441 U.S. 418,
427 (1979)).
73. See id. at 1072.
74. Lillquist, supra note 52, at 91.
75. See id.
76. For similar hypotheticals, see id.; Walen, supra note 14, at 421.
180
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
following from an erroneous acquittal seems higher than in traditional
nonviolent drug offenses.77 We can also imagine cases in which the benefits
of an accurate acquittal might be low. For instance, there may be a case
in which there is ample evidence that a particular defendant is engaged in
drug trafficking, car theft, or stock manipulation, but there is relatively
little evidence that the defendant committed the act for which the defendant
is being charged. Society often accepts, or even demands, that more resources
be spent on law enforcement to fight certain crimes, under the justification
that fighting those crimes is more valuable than fighting others.78 Likewise,
society may decide the risks of adjudicative errors are more or less costly
depending on the type of offense under consideration.
These claims might sound controversial. I suspect this is partially a
result of the strong gravitational pull the current system exerts on our
judgments. Let us consider a thought experiment: Imagine you are part
of a special committee designing an evidentiary system from scratch. In
particular, you are a member of a subcommittee charged with establishing
the standards of proof applicable across different civil and criminal cases.
You understand your task involves assessing the different significances
and costs associated with false positives and false negatives—as well as
benefits associated with accurate results—in different types of cases. It is
implausible that such a committee would arrive at a unanimous decision
that the proper ordering of costs and benefits is the same in virtually all
cases. With that in mind, we can turn to our current evidentiary system
and ask whether the current state of affairs with a fixed, very restricted
number of standards applicable across drastically different cases is really
justified.79 This Article offers a different argument in support of answering
77. See Lillquist, supra note 52, at 162.
78. See, e.g., Talia Fisher, The Boundaries of Plea Bargaining: Negotiating the
Standard of Proof, 97 J. CRIM. L. & CRIMINOLOGY 943, 953 (2007) (arguing for the
desirability of a negotiable criminal standard of proof between prosecutors and defendants);
Note, Winship on Rough Waters: The Erosion of the Reasonable Doubt Standard, 106
HARV. L. REV. 1093, 1095 (1993) (“[A]s society’s interest in crime control changes, society’s
assessment of the proper balance between erroneous convictions and erroneous acquittals
may change too.”).
79. Another interesting finding emerges from this quick thought experiment. It is
not only difficult to justify a restricted number of standards of proof in civil and in criminal
cases but it also becomes hard to justify a categorical distinction between the standards for
civil and criminal cases. See, e.g., LAI, supra note 5, at 215–16 (“For all these reasons, it
is difficult to justify a categorical difference in the standard of caution for civil and criminal
cases; the standard in both contacts should be determined on the same broad principle.
There must be as many ‘standards of proof’ as there are material differences in the circumstances
181
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
no to this question. We have good reasons to want different error distributions
across varying circumstances.80
I do not want to suggest society and the current legal system are
completely oblivious to relevant differences between circumstances and
cases. In fact, we constantly express our weighting of those differences
through law. We assign higher penalties for what we believe are more
serious crimes and lower penalties for less severe crimes. We hold defendants
liable for higher damages—sometimes even exceeding plaintiffs’ losses—
when their conduct is considered more blameworthy or socially costly.
We have different regulations aimed at leveling the playing field between
disparately situated members of society—labor law, consumer protection,
antitrust. Even after a suit is filed, we express our different weighting of
circumstances and parties through a myriad of evidentiary and procedural
regulations—who carries the burdens of production and persuasion, exclusionary
rules, rules of discovery, jurisdictional rules. To the extent these regulations
make guilty or liable verdicts difficult to reach, they also have a potential
impact on the distribution of errors in our society. Thus, we can see these
regulations as mechanisms through which society already reflects the
different weighing of the costs of errors in different situations, even if most
legal rules are not justified concerning their potential impact on errordistribution.
An important question then follows: If society already reflects its views
about the costs of errors in different situations through existing legal
mechanisms, why should we discuss how to distribute errors with standards
of proof? The answer is twofold. First, scholars and practitioners seldom
acknowledge the impact of other legal rules in the distribution of errors.
Second, even when we see discussions about standards of proof as a
mechanism for distributing errors, we do not actually see discussions
about how errors should be distributed. Instead, we uncritically accept
that errors should be distributed similarly to vastly different situations as
if we weigh the relative costs and significance of those situations similarly.
of cases—or, more accurately, there should only be one standard, a variant one. . . . Since
criminal cases differ in both the gravity of charges and the range of punishment, there is
no basis for applying to all of them a uniform standard [of proof].” (footnote omitted));
Issachar Rosen-Zvi & Talia Fisher, Overcoming Procedural Boundaries, 94 VA. L. REV.
79, 86 (2008).
80. This proposal is not unprecedented in American legal scholarship. See, e.g.,
LAUDAN, supra note 11, at 55 (“Where standards of proof are concerned, I am not
convinced that one size fits all.”); Kaplow, Burden of Proof, supra note 4, at 786 n.86
(“The . . . optimal evidence thresholds and the evidence threshold . . . will vary greatly by
context, even at fairly refined levels.”); Walen, supra note 14, at 417 (“[W]hether one is a
consequentialist or deontologist, one has good reason not to cling artificially to the idea
that only one [standard of proof] must be used for all of criminal law.”).
182
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
But once we consider other legal rules, we acknowledge how different those
situations are and enact a long list of regulations that reflect our sensitivity
toward those differences. At that point, however, discussions about the
impact of error distributions are nowhere to be found. We, therefore, end
up never really discussing, or even considering, the impact of legal rules
on the distribution of errors in our society and what the socially desirable
distributions actually are. We need to bring such discussions to the
surface. And what better time to have these than when thinking about the
very mechanism that many scholars and practitioners alike agree has errordistribution as a main function?
IV. NORMATIVE ARGUMENTS IN FAVOR
This Part surveys normative arguments in favor of varying standards of
proof. It starts by describing recent influential works on standards of
proof by consequentialists legal scholars. It then argues that once we
follow the logic of their positions to its ultimate conclusions, we must
consider seriously different standards for different types of cases. It then
discusses how the same is true for arguments based on fairness considerations.
These share similar conclusions with welfare-based consequentialists
arguments, even though they begin with different premises. Finally, this
Part explores how arguments premised on considerations regarding the
division of legal entitlements in our society also support moving to a
system with varying standards of proof.
A. Welfare Considerations
The topic of standards of proof has recently regained currency as a subject
among economically-minded legal scholars.81 One of their focuses has
81. See generally, e.g., Michael L. Davis, The Value of Truth and the Optimal
Standard of Proof in Legal Disputes, 10 J.L. ECON. & ORG. 343 (1994); Dominique
Demougin & Claude Fluet, Deterrence Versus Judicial Error: A Comparative View of
Standards of Proof, 161 J. INSTITUTIONAL & THEORETICAL ECON. 193 (2005); Edward K.
Cheng, Reconceptualizing the Burden of Proof, 122 YALE L.J. 1254 (2013); Kaplow,
Burden of Proof, supra note 4; Louis Kaplow, Information and the Aim of Adjudication:
Truth or Consequences?, 67 STAN. L. REV. 1303 (2015); Louis Kaplow, Optimal Proof
Burdens, Deterrence, and the Chilling of Desirable Behavior, 101 AM. ECON. REV. 277
(2011); Thomas J. Miceli, Optimal Prosecution of Defendants Whose Guilt Is Uncertain,
6 J.L. ECON. & ORG. 189 (1990); A. Mitchell Polinsky & Steven Shavell, Legal Error,
Litigation, and the Incentive to Obey the Law, 5 J.L. ECON. & ORG. 99 (1989); Daniel L.
Rubinfeld & David E.M. Sappington, Efficient Awards and Standards of Proof in Judicial
183
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
been on how to set the levels of standards to maximize social welfare.82
More specifically, these scholars have presented thought-provoking analyses
of how the optimal level of standards depends on a range of elements that
are bound to vary in different situations. My goal is to show that, regardless
of specific criticisms or departures from more traditional accounts, many
of the key elements of these projects support a system of varying standards
of proof.
Louis Kaplow has delivered what is perhaps the most comprehensive
recent welfare analysis of standards of proof.83 For that reason, this
section draws heavily from his framework. Kaplow’s analysis is based on
the idea that the most relevant effects of setting the standard of proof at
different levels are the changes in the costs associated with “deterrence
and chilling” in society at large.84 Although deterrence and chilling refer
ultimately to the same phenomenon—the impact of legal rules on primary
behavior—the use of different terms is meant to capture the distinction
between what Kaplow refers to as “harmful acts”—acts with negative net
social benefits—and “benign acts”—acts with positive or zero net social
benefits.85
The essential premise is that the prospect of sanctions will influence the
behavior of individuals thinking about committing harmful or benign
acts.86 More precisely, individuals with an opportunity to commit a harmful
or benign act are disincentivized when the private benefit an individual
expects to gain from committing the act is lower than the sanction an
Proceedings, 18 RAND J. ECON. 308 (1987). For earlier examples, see generally Alan D.
Cullison, Probability Analysis of Judicial Fact-Finding: A Preliminary Outline of the
Subjective Approach, 1 U. TOL. L. REV. 538 (1969); John Kaplan, Decision Theory and
the Factfinding Process, 20 STAN. L. REV. 1065 (1968); Posner, supra note 38. This type
of work is not without its critics. See, e.g., Walen, supra note 14, at 423 (arguing
consequentialist approaches to standards failed to “give the distinction between innocence
and guilt its proper moral significance”).
82. See generally Kaplow, Burden of Proof, supra note 4. It is important to note
that Kaplow’s account differs drastically from the formulation of standards in Part II. For
him, the account focused on the distribution of errors between parties suffers from the
defect of not taking into consideration all the elements included in his welfare analysis.
See generally id. It is out of the scope of this article to arbitrate between different formulations
of standards of proof. My objective in Part II was to introduce the reader to a widespread
way of formulating standards of proof according to scholars and practitioners alike. My
goal with introducing Kaplow’s welfare analysis in this Part IV is to highlight how it also
leads to the conclusion that a system with varying standards of proof is normatively desirable.
I take no stance on which formulation is preferable.
83. See generally id.
84. Id. at 746, 746 n.16.
85. Id. at 747.
86. See generally id.
184
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
individual expects to receive for committing the act.87 That is, even in
deciding whether to commit an act that is currently lawful, individuals
weigh the expected private benefit against the expected private costs of
committing that act.88
Ideally, only harmful acts would be deterred, while benign acts would
not be chilled. The fact that some benign acts can be confused with
harmful acts makes this result difficult. One important question for our
purposes is what happens to the levels of deterrence and the chilling effect
of different acts when we change the applicable standard of proof?
Asking this question will allow us to understand better some of the costs
and benefits involved in the choice of standards. Recall the discussion
above about the trade-off involved when we alter the level of the standards
of proof.89 We saw that if we lower the standard, that is, move the line in
Figures 2 or 3 to the left, we increase the likelihood that the legal system
will correctly assign liability to truly guilty individuals that go to trial while
at the same time increasing the likelihood that truly innocent individuals
will mistakenly be sanctioned.90
Likewise, when we lower the standard, holding other features of the
legal system constant, we increase the chances of punishing both harmful
and benign acts, which, in turn, increases the expected sanction for both
types of acts.91 As a result, harmful acts will be deterred while more benign
acts will be chilled.92 When harmful acts are deterred, society benefits;
when benign act are chilled, society loses. Thus, we must compare the
aggregate deterrence benefits to the overall chilling costs.93 The important
question from a welfare-based analysis is whether the net benefit from
these effects is positive or negative. If the benefit resulting from the
deterrence of harmful acts is greater than the loss resulting from the chilling
of benign acts, then reducing the level of standard is desirable.94 Whereas
if the reverse is true—that is, if the loss outweighs the benefits—not only
would such reduction be socially detrimental but also an increase in the
87. See generally id. This is the same regardless of whether the sanction is correctly
or mistakenly imposed on the person. Id. at 749–50.
88. Id.
89. See supra pp. 171–72.
90. See supra pp. 171–72.
91. Kaplow, Burden of Proof, supra note 4, at 763.
92. Id.
93. Id.
94. Id.
185
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
level of the standards would be “socially desirable.”95 The optimal standard
of proof for welfare maximization is given by the point where these two
opposing forces reach an equilibrium.96
To appreciate fully the point that the optimal standard varies according
to the act under consideration, we must break down the two forces
explained in the last paragraph into its defining components. Let us start
with the deterrence effect. The level of deterrence following a change in
the level of the standard depends primarily on three factors: “the increase
in expected sanctions, . . . the concentration of marginal harmful acts,”
and the difference between social costs and the private benefits of marginal
harmful acts.97
First, the change in expected sanctions. This is a product of three other
factors: the (1) likelihood that the competent authorities will identify the
harmful act, (2) likelihood that the individual will be subject to sanctions,
and (3) magnitude of the sanctions imposed.98 A change in the level of
the applicable standard directly impacts the second factor—the likelihood
that the individual will be subject to sanctions.99 The exact impact that a
change in the level of the standard has on this factor can be determined
from Figures 2 or 3. Depending on whether we increase or lower the
standard—move the line to the right or move it to the left, respectively—
the likelihood that the individual will be subject to sanctions will be equal
to the area under the harmful acts curve.
We can then calculate the impact of a change in the level of the standard
by determining the difference between the new area under the curve after
we changed the standard and the area under the curve before we moved
the line. For example, as one raises the level of the standard, we move the
point representing the standard to the right in Figures 2 or 3. As a result,
parts of the area under the harmful acts curve that used to be to the right
of the standard line will now be to the left. That the new area under the
harmful acts curve to the right of the standard line is now smaller than the
area under the curve to the right of where the line used to be means that
the likelihood that the individual will be subject to sanctions is lower. We
can then compute the increase in the expected sanction by multiplying the
magnitude of the sanctions imposed by the likelihood that (1) the competent
authorities will identify the harmful act and (2) the individual will be subject
to sanctions. Assuming both likelihoods stay constant, an increase in the
95.
96.
97.
98.
99.
186
Id.
Id.
Id. at 764.
Id. at 754.
Id.
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
level of the standard generally means the expected sanction for harmful
acts is lower.
The second component that will help determine the deterrence benefits
is the “concentration of marginal harmful acts.”100 This is determined by
“how often private benefits from [those] acts fall [within the] range” of
the difference of the increase in the expected sanction.101 Together with
the first component, this provides the change in the number of individuals
who will be deterred because of a legal change, such as adjusting the level
of the applicable standard of proof.102 To reach that figure, we must consider
which specific acts will—or will not—be deterred as we alter the standard.103
The acts that will be deterred “will be those that generate private benefits”
lower than the new expected sanction after lowering the standard, minus
those acts which were already deterred under the old expected sanction.104
The third and last factor determining deterrence effects of changes in
the standard is “the net social gain per deterred act.”105 This is equal to the
“difference between social harm . . . and the private benefit” of a marginal
harmful act.106 The size of the social harm will depend on the context: “for
littering, it will be small, but for discharging highly toxic chemicals into
the water supply, it will be large.”107 And, the benefit of a marginal harmful
act “equals the expected sanction.”108 “From the social harm per act, we
[still] need to subtract each act’s private benefit.”109
The exercise is not over yet. “Against th[e] deterrence benefit[s], we
must weigh the chilling cost[s].”110 These costs are equal to “the increase
in the expected sanction[s] for benign acts weighted by the concentration
of marginal benign acts.”111 This “product is then multiplied by the net
cost per act that is chilled.”112 The analysis is analogous to the deterrence
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
Id. at 764.
Id. at 765.
Id. at 765–66.
Id. at 765–67.
Id. at 765.
Id. at 766.
Id.
Id.
Id.
Id.
Id. at 768.
Id.
Id.
187
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
effects discussion above, so I will leave to the reader the task of filling in
the blanks.
The optimal standard will be where the marginal “deterrence benefit
just equals the chilling costs.”113 At that equilibrium, society does not
have an incentive to increase or lower the applicable standard. Certainly,
this calculation in the real world is highly complex. Much depends on
many empirical facts that are likely to vary quite drastically in context.
The optimal level of the standard is determined by a multitude of factors,
which can take on a vast range of values. The important takeaway for this
Article is that we have reason to believe deterrence and chilling costs will
be vastly different among different types of acts.114 The conclusion that
follows is that the optimal level of the standard also varies according to
different acts. As Kaplow puts it:
In some cases, the optimal evidence threshold might be quite high. For example, if
the act is not very harmful, there are many benign acts that look like the harmful
act, and the legal system does not discriminate well between them, then chilling
costs will exceed deterrence benefits except possibly at very high evidence
thresholds. On the other hand, if the act is quite harmful, few benign acts are
likely to be confused with harmful acts, and it is otherwise difficult to generate a
high expected sanction (perhaps stage one only identifies a small fraction of harmful
acts), then deterrence benefits will exceed chilling costs until the evidence threshold
is reduced substantially.115
The fact that the optimal standard varies for different acts gives us a
reason to question whether the current system of evidentiary rules, in which
different acts are lumped together under the same standard, is really the best
system for welfare maximization analyses. It does not take much institutional
imagination to envision a system that tracks the optimal level of the standard
more closely than our current system does.
This framework is not without its problems. For instance, it potentially
requires high administrative costs and ignores effects of dynamic behavior.116
I deal with this and other objections below. Here, it suffices to say that I
do not endorse every aspect of Kaplow’s account of standards of proof.
My objective in exploring this account was to illustrate one type of reasoning
about evidence law—in general—and standards of proof—in particular—
that stresses the sometimes drastically different private and social costs
associated with different types of conduct and how these costs should weigh
in our thinking about how to regulate behavior.
113. Id. at 769.
114. It is also important to keep in mind that other enforcement mechanisms have
behavioral effects, and the interaction of changes in part of the system might have unpredicted
consequences—think second best-type arguments.
115. Kaplow, Burden of Proof, supra note 4, at 770.
116. Allen & Stein, supra note 9, at 580, 582–84.
188
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
B. Fairness Considerations
In his only published work on procedural and evidentiary issues, Ronald
Dworkin explores whether a community that provides its citizens a right
not to be convicted if innocent but also decides questions of procedure
and evidence according to an ordinary welfare analysis, can be blamed for
inconsistency.117 He starts by addressing the following question: If we
recognize “that people have a right not to be convicted of . . . crime[s]”
they did not commit, does it follow “that people have a right to the most
accurate procedures possible” to prove their guilt or innocence, regardless
of how expensive these might be?118
Suppose cases could be made marginally more accurate if we had a
“best out of three” rule. Under this rule, cases would be tried by up to
three different juries and the party that got favorable verdicts from two juries
would prevail.119 Suppose further that this would cost cases more time
and money. If society decides to continue to have only one trial to save
costs, it will forgo the marginal increase in accuracy. Consequently, courts
will mistakenly convict some innocent defendants that otherwise would
be acquitted under the best out of three rule. Is the decision not to adopt
such a rule unjust against defendants who are denied a reduction in their
risk of being mistakenly convicted? If yes, we would be forced to admit
our justice system is largely unfair because we do not implement all possible
instruments that could increase, however marginal, accuracy regardless of
the instruments’ costs.
Most people would resist conceding this last point. Should we, then,
hold that people at trial have no rights to any particular level of accuracy?
That seems to be our assumption if we left all procedural questions to costbenefit analysis, like the one in the last section. But would that position
be consistent with the initial assumption that people have a right not to be
convicted if innocent? Can we find some middle ground between the
117. See DWORKIN, supra note 5, at 72–80.
118. Id. at 79–80. Note that a right in a Dworkinian sense is quite strong, implying
that arguments of policy are not sufficient to prevent the enforcement of a right. See id. at
72. For instance, “[i]f a prosecutor were to pursue a person he knew to be innocent, it
would be no justification or defense that convicting that person would spare the community
some expense or in some other way improve general welfare.” Id. When a person asks a
court to enforce a right, “that the community would be better off if that right were not
enforced, is not [to be] counted a good argument against him.” Id. at 73.
119. Or we could add the counts on individual cases of actions. The details are not
important for this quick thought experiment.
189
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
absurd claim that people have a right to all procedures that increase
accuracy regardless of costs and the seemingly nihilistic claim that people
lack a right to any level of accuracy?120
Dworkin attempts to answer these difficult questions through the concept
of “moral harm.”121 According to him, we must distinguish between (1) the
harm a person suffers through punishment—pain, frustration, damages to
reputation, loss of income—that we might call “bare harms,” and (2) another
kind of harm a person might suffer whenever the punishment is unjust or
otherwise treated unjustly—whether the person knows, or cares, about
it.122 It is this second type of harm Dworkin refers as “moral harm.”123
We can grasp that idea when we sympathize for someone upon learning
that the person was treated unjustly or cheated, even though we learned
nothing more about how much that person suffered from punishment.124
With the notion of moral harm, we can better understand the supposed
strangeness behind the behavior of a society that recognizes that people
have a right not to be convicted if innocent but decides procedural and
evidentiary policies according to welfare analyses. It seems inconsistent
to acknowledge the right not to be convicted if innocent unless we recognize
moral harm as a special kind of harm against which people must be protected.
The problem is that typical welfare analyses seem to have no place for
moral harm.125
Importantly, even an analysis that has a place for moral harm should
not strive to reduce it to zero. “[W]e do not lead our lives to achieve the
minimum of moral harm at any cost . . . .”126 Rather, we accept weighty
120. Note, it is not true that the right not to be convicted if innocent is meaningless
in a society that does not recognize a right to any particular level of accuracy and decides
its procedural rules solely on the basis of cost-benefit analysis. That right still protects
citizens against unfounded suits.
121. DWORKIN, supra note 5, at 80–90.
122. Id. at 81.
123. Id. Someone might question this distinction by saying it confuses the quantity
of harm someone suffers from a punishment with the different issue of whether that harm
is just. When punishment is unjust, the harm suffered is unjust, but it is confusing to say
the injustice in some way adds to the harm. However, one need not accept this distinction
to acknowledge the substantive point behind the idea of moral harm—that someone suffers
a special injury when treated unjustly. See id. at 80–81.
124. See id. Although Dworkin does not discuss this possibility, it seems plausible
that the concept of moral harm could be modified slightly to cover cases of wrongful acquittals.
125. Dworkin here is referring to “utilitarian calculations” that measure costs and
benefits according to a utilities and disabilities function determined largely by some
psychological states along a pleasure and pain spectrum, even if such function includes
people’s preferences to be punished, or have others be punished, unjustly. Id. at 82. This
is because, to Dworkin, the idea of moral harm is an objective notion. See id. at 80–86.
126. Id. at 86.
190
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
“risks of suffering injustice[s]” to achieve other worthy goals.127 We do
so when we enter into relationships, sign contracts, or trust friends.128
Likewise, society should have a similar approach to moral harm.129 Under
this light, society’s evidentiary rules are embedded in its choices about the
relative importance of different types of moral harm.130
Now we get to the heart of why I believe this discussion provides an
argument for a system with varying standards of proof. Nothing in Dworkin’s
analysis forces us to place the same importance on all types of moral
harms in different cases. Just like we expect that, in a fair society, the
level of sanctions attached to various crimes should be consistent with the
relative importance society places on that crime, we should also expect
that procedural and evidentiary instruments correlate to the relative importance
of risk of moral harm. If we value the injustices that would follow from
wrongful convictions or acquittals differently, then we should weigh moral
harm differently among different cases. According to this framework, we
should have different standards of proof for different situations, and these
standards should reflect society’s choices about the relative importance of
the types of moral harm involved.
There are many different instruments a society might use to alter its
legal procedures to have them reflect the relative weight of moral harms.
One way for society to pay a higher price for the accuracy of its adjudicative
proceedings is to guard against particular kinds of mistakes that involve
greater moral harm. Standards of proof are the clearest instrument through
which this protection can be articulated and achieved. In Dworkin’s words:
[I]n the civil law . . . it is generally assumed that a mistake in either direction involves
equal moral harm. But [the different standard in defamation suits] may represent
some collective determination that it is a greater moral harm to suffer an
uncompensated and false libel than to be held in damages for a libel that is in fact
true.131
In a Dworkinian universe, these considerations acquire additional strength
because people might have a right to procedures that correlate the correct
127. Id.
128. Id.
129. See id.
130. Id. (“[W]e might regard the design of criminal and civil procedures as a fabric
woven from the community’s convictions about the relative weight of different forms of
moral harms, compared with each other, and against ordinary sacrifices and injuries.”).
131. Id. at 89.
191
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
importance to the risk of moral harm they face.132 This is because, for
Dworkin, it is part of the principle that any political decision must treat
all citizens as equals, and no decision may deliberately impose a greater
risk of moral harm on any citizen than that which it imposes on any
other.133 The difficulty, however, is that all members of society do not
face equal risks of suffering the same kinds of moral harms. People are
not equally likely to be drawn into the criminal process though innocent
or equally likely to benefit from the savings gained by choosing a particular
rule of evidence rather than a socially more expensive rule. Economic
inequality and existing prejudices are factors that might make it likely that
members of different segments of society will be more likely to be accused
of certain types of crimes. For example, the poor might be more likely to
be accused of petty thefts and the rich of antitrust violations. Other differences,
such as in temperament and personalities, also make it more likely that
certain types of people will be accused of different crimes. The hot-blooded
or the gun aficionados are more likely to be accused of violent or gun-related
offenses, while the greedy or Wall Street-types are more likely to be accused
of securities violations.
Because of these inequalities, the final weighing of moral harms against
bare harms is likely to be controversial. Minorities, who might face a
higher probability of being drawn into the criminal process even though
innocent, will protest that the level of accuracy provided by the rules at
play is too low, undervaluing the moral harms following from unjust
convictions. The majority, on the other hand, might think the level of
accuracy is too high, and thus overvalues moral harms compared to the
benefits forgone when using taxpayer funds for other purposes.
Overall, this means procedural and evidentiary rules involve important
political decisions. They involve decisions about weighing of moral harms
against bare harms, which crucially affects the probabilities of different
people and social groups. These decisions also involve issues about the
use of taxpayers’ money. It is expected that people will gain differently
from a different use of public funds. In turn, this suggests it is not enough
to fault a society’s procedural and evidentiary system because it imposes
different consequences on different members. Society’s decisions about
its procedural and evidentiary rules can only be accused of being unfair if
these decisions constantly discriminate against independently distinct
132. Id. at 89–90. To say this is a right is to say that once the content of the right is
determined—that is, once the appropriate evaluation of the moral harm is done—society
must protect the right of its citizens to procedures that represent that evaluation, even if
the general welfare suffers as a consequence. See id.
133. Id. at 85.
192
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
groups.134 It is not enough to consider these decisions unfair simply because
they weigh moral harms differently from what other members of the society
would weigh them.
Dworkin also mentions that people have a right to a “consistent”
weighing of the importance of moral harm.135 This consistency is not merely
that like cases should be treated alike. There also should be consistency
with the community’s evaluation of the moral harm at play.136 For Dworkin,
to assure consistency, we must search society’s textual and historical record
to find an interpretation that fits and justifies most—but not necessarily
all—of that record.137 Most of the time, this consistency requirement might
act as a conservative force protecting the accused from changes in the
evaluation of moral harm. But it can also act as an instrument for reform
by identifying existing, and even deeply embedded, procedures as mistakes
that cannot be made consistent with legal and political practices.
How much consistency is required? Should we require the same weight
of moral harm across all cases? I do not think so. Consistency across the
board would require that society value different moral harms as having
the same weight. Yet this is not what we observe. The different use of
standards of proof between civil and criminal cases suggest that society
assigns disparate values to moral harms following from different cases.
However, even within criminal or civil cases, we find evidence for various
evaluations of moral harms through procedural and evidentiary mechanisms
that make conviction more or less likely as well as increase or decrease
the interval between the minimum and maximum sanction. Examples
include evidentiary privileges, class actions, and mandatory minimums.
It must then be the case that the consistency requirement mentioned above
can be fulfilled even if procedures and evidentiary rules vary among different
types of cases. This is enough for the purpose of the proposal in the Article.
One potential objection arises out of fairness considerations. To some,
it might seem that fairness considerations give rise to equality concerns.
Equality concerns suggest that “anyone who brings a civil claim should
134. See id. at 87–88.
135. Id. at 90.
136. See id. Both of these rights are rights in the strong Dworkinian sense, according
to which rights act as a trump over the balance of costs and benefits. See id.
137. Id. at 96 (“The basic procedural right in civil litigation is the right that the risk
of the moral harm of an unjust result be assessed consistently so that no less importance is
attached to that risk by a court’s procedural decision that is attached in the law as a whole.”).
193
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
have it judged by the same standard of proof that applies to everyone else[].”138
But this objection confuses equal treatment with being treated equally.
We ought to differentiate between these two ideas. To be treated equally
is not equivalent to receiving the same number of goods or the same level
of burdens as everyone else.139 Rather, it is better understood as being
treated with the same degree of concern and respect.140 Being treated equally
does not demand a single fixed standard of proof to govern vastly different
cases. In fact, being treated equally seems to require the exact opposite.
It requires that individuals have a right to procedures that correlate the
correct importance to the risk of moral harm they face. In that sense, a
system with varying standards of proof seems to be a dictate of equality.
It is unsurprising that Dworkin’s analysis is sharply distinct from Kaplow’s.
The point of this Article is not to offer one univocal defense of the proposal
for varying standards of proof. Rather, the point is to discuss different
types of arguments for this proposal.141
C. Distributive Considerations
Distributive analyses are almost always highly controversial. Perhaps
for that reason, distributive considerations are often disguised as well-known
considerations, such as formal legal equality or equality of bargaining power.
138. LAI, supra note 5, at 222 (citing C.M.A. McCauliff, Burdens of Proof: Degrees
of Belief, Quanta of Evidence, or Constitutional Guarantees?, 35 VAND. L. REV. 1293,
1334–35 (1982)); see also Walen, supra note 14, at 426–27 (arguing (1) against instrumental
considerations in setting standards of proof and (2) that retributive considerations should
lead us to uphold the same beyond a reasonable doubt standard across much of criminal
law). But see Jeffrey Reiman & Ernest van den Haag, On the Common Saying that It Is
Better that Ten Guilty Persons Escape than That One Innocent Suffer: Pro and Con, SOC.
PHIL. & POL’Y, Spring 1990, at 226, 227–28, 247–48.
139. See RONALD DWORKIN, TAKING RIGHTS SERIOUSLY 227 (1978).
140. Id.
141. For another interesting analysis that could be included under this section—and
that also shows some skepticism towards a system with fixed standards—see LAI, supra
note 5, at 215–23 (“For all these reasons, it is difficult to justify a categorical difference in
the standard of [proof] for civil and criminal cases; the standard in both contexts should
be determined on the same broad principle. There must be as many ‘standards of proof’
as material differences in the circumstances of cases—or, more accurately, there should
only be on standard, a variant one.” (footnote omitted) (citing G.H.L. Fridman, Standards
of Proof, 30 Canadian Bar Rev. 665, 670 (1955))). “Since criminal cases differ in both
the gravity of charges and the range of punishment, there is no basis for applying to all of
them a uniform standard of [proof].” Id. at 216; see also David Hamer, Review: A Philosophy
of Evidence Law—Justice in the Search for Truth, 13 INT’L J. EVIDENCE & PROOF 161, 161
(2009) (book review).
194
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
Rarely does someone explicitly advocate for a legal change because it transfers
resources away from one social group to another.142
The kind of distributive analysis I am interested in here focuses on the
role of background entitlements in the distribution of resources among
different social groups.143 The division of entitlements in society matters
greatly for the expected consequences of legal changes. Any argument for
the welfare virtues of any legal change necessarily depends on specific
assumptions about the background entitlements of those affected.144 This
is “[b]ecause entitlements are a component of wealth.”145 Entitlement
settings “influence the allocation of resources through ‘wealth effects’”—
that is, through composition of demand, price, and income elasticity of
demand.146 So, if we can manipulate background entitlements, we can bring
about changes in wealth so great that they should allow us to influence
greatly—as to dominate potentially—the cost-benefit analysis of any legal
change.147
Most dramatically, it might turn out that the setting of the entitlement
completely determines ex ante the outcome of the welfare analysis.148 For
example, in situations in which there can be no voluntary transactions
between the individuals affected by a proposed legal change—because of
prohibitive transaction costs, perhaps—the initial entitlement allocation
will dictate the analysis outcome. If firms are entitled to pollute, people
that own land nearby and are negatively affected by the firms’ pollution might
not be able to transact with the polluters to reduce the firms’ externalities
142. Kennedy, Distributive and Paternalist, supra note 7, at 588. Duncan Kennedy
makes this point eloquently:
Peace and happiness seem to require that most of the time we not think at all about
the justice of distributive shares. Otherwise, we risk falling into depression, or
into a rage, and, in either case, out of sympathy with one another. Or worse yet,
we may fall into the kind of sympathy—intense but selective—that leads to civil
war. There is therefore a taboo on the explicit consideration of distributive
consequences, let alone distributive goals.
Id.
143. For a seminal treatment of entitlements and disablements and their correlatives,
see generally Wesley Newcomb Hohfeld, Fundamental Legal Conceptions as Applied in
Judicial Reasoning, 26 YALE L.J. 710 (1917); Wesley Newcomb Hohfeld, Some Fundamental
Legal Conceptions as Applied in Judicial Reasoning, 23 YALE L.J. 16 (1913).
144. See Kennedy, Cost-Benefit Analysis, supra note 7, at 388.
145. Id. at 423.
146. Id.
147. Id. at 423–26.
148. Id. at 426.
195
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
because of high transaction costs. Consequently, a rule that allows firms
to pollute would be seen as effecting an efficient allocation of resources.149
Inversely, if landowners were entitled to clean air, firms would have to
transact with landowners to engage in pollution-causing activities. But if
transactions costs are prohibitively high, these transactions will not occur.
As a result, a legal rule that prohibits firms from polluting would seem
efficient.
Entitlements can have even more subtle effects. If landowners are entitled
to clean air, that makes their position more valuable—for instance, in terms
of higher real estate value. This, in turn, increases the amount landowners
are willing accept to allow pollution from nearby firms. Conversely, if firms
are entitled to pollute, they are better-off and, consequently might start to
value this entitlement at a higher selling price. If this price is greater than
landowners’ willingness to pay, there would be no gains from trade. Thus,
the initial allocation of entitlements will not only remain intact but will also
appear to be the efficient allocation under a welfare analysis. Theoretically,
there will always be a particular combination of background entitlements
that generates wealth effects so great as to render any change inefficient.
This introduces a strong bias in favor of the status quo in this type of analysis.
Moreover, because we often have no good reason to set the initial entitlement
allocation for one party over another, this might be seen as introducing
arbitrariness to the analysis.
Distributive analyses are at the fringes of legal scholarship. 150 Evidence
and procedural laws are often left untouched. I believe this is a mistake.
Just like private law, evidentiary and procedural mechanisms should also
constitute individuals’ background entitlements.151 The exact impact of
changing the applicable standard on buyers and sellers will partially depend
on the price and income elasticities of supply and demand curves as well as
the competitive structure of that specific market. There might also be
149. See generally R.H. Coase, The Problem of Social Cost, J.L. & ECON., Oct. 1960,
at 1 (1960).
150. Noteworthy exceptions include: RONALD DWORKIN, SOVEREIGN VIRTUE: THE
THEORY AND PRACTICE OF EQUALITY 307–19 (2000) (health-care policy); LOUIS KAPLOW
& STEVEN SHAVELL, FAIRNESS VERSUS WELFARE (2002) (providing an in-depth discussion
of the role of distributive considerations in the assessment of legal policies); Matthew D.
Adler & Eric A. Posner, Rethinking Cost-Benefit Analysis, 109 YALE L.J. 165 (1999)
(administrative practice); Carlos A. Ball, Autonomy, Justice, and Disability, 47 UCLA L.
REV. 599 (2000) (disabilities); Daniel Markovits, How Much Redistribution Should There
Be?, 112 YALE L.J. 2291 (2003) (developing a new account of egalitarianism based on a
new conception of nonsubordination); and Edward J. McCaffery, The Uneasy Case for
Wealth Transfer Taxation, 104 YALE L.J. 283 (1994) (considering distribution implications for
the choice of the tax base).
151. See generally Symposium, Presumptions and Burdens of Proof, 17 HARV. J.L.
& PUB. POL’Y 613 (1994).
196
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
distributive effects on third parties, who can be made better off as a
consequence of lower standards because the increased expected sanctions
incentivize firms to reduce the risks of their economic activity, which, in
turn, might involve negative externalities.
The conceptual point here is that, similarly to rules of private law, we
can manipulate evidence rules operating in the background to bring about
changes in wealth with noticeable effects for a welfare analysis. We can
also manipulate evidence rules that bring about changes in wealth so extreme
that the change will determine the outcome of the analysis from the outset.
Consider what would happen if potential buyers are so impoverished that
they cannot pay a higher price for the product once the industry faces
higher expected accident costs due to a reduction in the standard of proof
for consumer product mass tort cases. Or consider the consequences for
private bargaining of a system with prohibitive judicial costs.
The point of this section is to argue that this distributive analysis of
evidence law also supports a system with varying standards of proof. If
the level of the applicable standard is seen—together with the rest of the
constellation of evidentiary and procedural regulation—as part of individuals’
background entitlements with relevant consequences for the distribution
of wealth and income in our society, then any legal reformer intending to
alter that distribution of wealth or income among different social groups
can also make use of standards of proof or evidentiary and procedural
regulation to reach his objectives. This is true regardless of whether distributive
considerations are instrumental to attaining other goals. It is possible, albeit
politically controversial, that one might attempt to promote a specific type
of legal reform with the sole objective of altering the distribution of wealth
or income between social groups. For instance, the example above involving
changing standards in mass consumer tort cases could be part of a greater
political strategy to reallocate funds from certain groups to others. Moreover,
even if such a proposed legal change is not publicly defended on such
grounds, the wealth effects following it constitute predictable consequences
which should be acknowledged and included in the public debate.
This might surprise some. Evidence law has traditionally been perceived
as a dry and highly technical field. Not surprisingly, it was thought to
belong almost exclusively to the professional interest of practitioners. But
nothing could be further from the truth—an especially problematic situation
for a field that has seeking truth as one of its main goals.
197
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
V. A POSITIVE ARGUMENT IN FAVOR
As I mentioned in Part II, we believe we are indifferent between errors
for plaintiffs or defendants in most civil suits and that we accept the same
error distribution for all criminal cases. The previous section presented
reasons why, in many circumstances, we might not want to do what we
say. This section argues that what we do is different from what we say.
Our practice is out of tune with our discourse. I attempt to show this
by referencing social science research on jury decision-making.
Although only a small percentage of cases are affected by jurors’
decisions,152 the jury has received more attention from behavioral scientists
and other researchers than any other comparable decision-making institution.153
Although there were early twentieth century anecdotal studies, “systematic
and [rigorous] empirical inquiry into . . . jury decision-making [can be] traced
to [the 1950s] University of Chicago Law School’s Jury Project.”154 That
project hosted a series of investigations into the role and functions of juries
in the American legal system, gathering information from over 3,500
criminal jury trials.155 Among other things, researchers compared the
verdicts rendered by juries with hypothetical verdicts that judges presiding the
juries would have rendered, revealing that juries and judges agreed roughly
75% of the time.156 Most disagreements stemmed from juries’ greater lenience
in comparison to judges.157
Studies on jury decision-making increased in the 1970s because of, in
part, critical decisions from the U.S. Supreme Court.158 In Witherspoon
152. In 2017, 327,557 cases were filed and 347,713 cases were terminated in all U.S.
District Courts. U.S. COURTS, JUDICIAL FACTS AND FIGURES, U.S. DISTRICT COURTS―
TOTAL CIVIL AND CRIMINAL CASES FILED, TERMINATED, AND PENDING Table 6.1 (Sept. 30,
2017), https://www.uscourts.gov/sites/default/files/data_tables/jff_6.1_0930.2017.pdf [https://
perma.cc/5AB6-2F2G]. However, only 11,134 jury trials were completed that same year.
U.S. COURTS, JUDICIAL FACTS AND FIGURES, U.S. DISTRICT COURTS—CIVIL AND CRIMINAL
TRIALS COMPLETED Table 6.4 (Sept. 30, 2017), https://www.uscourts.gov/sites/default/
files/data_tables/jff_6.4_0930.2017.pdf [https://perma.cc/V92S-N6CB].
153. See HASTIE ET AL., supra note 8, at 7.
154. David DeMatteo & Natalie Anumba, The Validity of Jury Decision-Making
Research, in 1 PSYCHOLOGY IN THE COURTROOM: JURY PSYCHOLOGY, supra note 12, at 1,
2 (citations omitted); see also DEVINE, supra note 8, at 14–15.
155. DeMatteo & Anumba, supra note 154, at 2–3. The project’s main findings were
summarized in two publications: HARRY KALVEN, JR. & HANS ZEISEL, THE AMERICAN
JURY (1966); HANS ZEISEL ET AL., DELAY IN THE COURT (1959).
156. DeMatteo & Anumba, supra note 154, at 2–3.
157. Id.
158. In the 1970s, psychology and law became an established independent field,
witnessing the first joint-degree program in psychology and law at the University of
Nebraska, the birth of the American Psychology-Law Society and the publication of Law
and Human Behavior, the first journal dedicated to research in psychology and law and to
this day one of the main journals in the field, as well as the publication of the first extensive
198
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
v. Illinois, the Court decided that excluding potential jurors during voir
dire based on their reservations regarding the death penalty violated
defendants’ Sixth and Fourteenth Amendment rights.159 Although the Court
expressed strong reservations about the jury decision-making studies
presented by the defendant—calling such studies “tentative and fragmentary,”160
that decision showed the Court’s willingness to consider such studies in
its reasoning.161 In Williams v. Florida, the court again referred to jury
decision-making studies when deciding whether the use of a six-member
jury in criminal cases—instead of a traditional twelve-member jury—was
constitutional.162 Citing six studies, the Court held that a jury of six was
equally as effective as a jury of twelve and was thus constitutional.163 Williams
sparked researchers’ interests, and several studies concerning jury size
followed.164 Many of these studies were later acknowledged in Colgrove
v. Battin when the Court upheld and extended Williams to civil trials.165
When the Court revisited the issue of jury size in Ballew v. Georgia—deciding
literature reviews of empirical research on juries. See James H. Davis, Robert M. Bray &
Robert W. Holt, The Empirical Study of Decision Processes in Juries: A Critical Review,
in LAW, JUSTICE, AND THE INDIVIDUAL IN SOCIETY: PSYCHOLOGICAL AND LEGAL ISSUES
326, 346 (June Louin Tapp & Felice J. Levine eds., 1977); DeMatteo & Anumba, supra
note 154, at 4–5; Kathleen Carrese Gerbasi et al., Justice Needs a New Blindfold: A Review
of Mock Jury Research, 84 PSYCHOL. BULL. 323, 340 (1977). Dennis Devine estimated
that over 1,500 jury studies were published by the end of 2011. DEVINE, supra note 8, at
8. This number, Devine notes, is most likely an underestimation, “with perhaps as many
as half of [the number of total studies conducted] never published.” Id.
159. 391 U.S. 510, 522–23, 528–29 (1968).
160. Id. at 517.
161. See Shari Seidman Diamond, Illuminations and Shadows from Jury Simulations, 21
LAW & HUM. BEHAV. 561, 568 (1997). The use of social science studies by the U.S.
Supreme Court goes back at least as early as Brown v. Board of Education. See 347 U.S.
483, 494, 494 n.11 (1954). However, the use of studies specific to jury decision-making
is more recent.
162. 399 U.S. 78, 135 (1970).
163. Id. at 86, 101–03; see also Robert J. MacCoun, Experimental Research on Jury
Decision-Making, 30 JURIMETRICS J. 223, 228 (1989); David L. Suggs, The Use of Psychological
Research by the Judiciary: Do the Courts Adequately Assess the Validity of the Research?,
3 LAW & HUM. BEHAV. 135, 144–45 (1979).
164. See DEVINE, supra note 8, at 17. Other Supreme Court decisions at around the
same time also sparked interest from psychologists. One such example includes the acceptability
of non-unanimous jury verdicts in restricted cases. See, e.g., Apodaca v. Oregon, 406 U.S.
404, 413–14 (1972); Johnson v. Louisiana, 406 U.S. 356, 380 (1972).
165. 413 U.S. 149, 158–60 (1973); see also Dennis J. Devine et al., Jury Decision
Making: 45 Years of Empirical Research on Deliberating Groups, 7 PSYCH., PUB. POL’Y
& L. 622, 668 (2001). After Colgrove, the scientific community heavily criticized four of
the studies cited by the Supreme Court. See Suggs, supra note 163, at 145.
199
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
juries smaller than six people violate defendants’ Sixth and Fourteenth
Amendments rights—it cited social science research indicating that deliberation
in groups smaller than six people increases the chance of biases.166
Many other topics besides jury selection and jury size became the subject
of empirical research about jury decision-making. For instance, following
the Supreme Court’s upholding of capital punishment in Gregg v. Georgia,167
many researchers turned their attention to death penalty cases, particularly
focusing on the possibilities of racial biases.168 Part of that research figured
prominently in McCleskey v. Kemp, when the Court decided that the available
data on existing racial biases in death penalty cases in Georgia was not
enough to overturn a death sentence because the defendant had not shown
that “purposeful discrimination ‘had a discriminatory effect’ on him.”169
Research on jury decision-making continued to accelerate into the first
decades of the twenty-first century.170
A key element of the newer research is that after federal and state
governments prohibited the recording or observation of jury deliberations,
studies since 1970 have focused mostly on simulated, or mock, trials.171
166. 435 U.S. 223, 234–38, 245 (1978); see also William C. Thompson, Research
on Jury Decision Making: The State of the Science, in INDIVIDUAL AND GROUP DECISION
MAKING: CURRENT ISSUES 203, 205 (N. John Castellan, Jr., ed., 1993); Suggs, supra note
163, at 144, 146–47.
167. 428 U.S. 153, 169 (1976).
168. See generally, e.g., David C. Baldus, et al., Racial Discrimination and the Death
Penalty in the Post-Furman Era: An Empirical and Legal Overview, with Recent Findings
from Philadelphia, 83 CORNELL L. REV. 1638 (1998); David C. Baldus, et al., Comparative
Review of Death Sentences: An Empirical Study of the Georgia Experience, 74 J. CRIM. L.
& CRIMINOLOGY 661 (1983).
169. 481 U.S. 279, 292–93, 313 (1987) (quoting Wayte v. United States, 470 U.S. 598,
608 (1985)).
170. That period saw the publication of influential works. See generally, e.g., JEFFREY
ABRAMSON, WE, THE JURY: THE JURY SYSTEM AND THE IDEAL OF DEMOCRACY (1994); STEPHEN
J. ADLER, THE JURY: TRIAL AND ERROR IN THE AMERICAN COURTROOM (1994); VALERIE P.
HANS & NEIL VIDMAR, JUDGING THE JURY (1986); HASTIE ET AL., supra note 8; RANDOLPH
N. JONAKAIT, THE AMERICAN JURY SYSTEM (2003); SAUL M. KASSIN & LAWRENCE S.
WRIGHTSMAN, THE AMERICAN JURY ON TRIAL: PSYCHOLOGICAL PERSPECTIVES (Barbara A.
Bodling ed., 1988); NEIL VIDMAR & VALERIE P. HANS, AMERICAN JURIES: THE VERDICT
(2007). For a good summary in the law review literature, see generally Lillquist, supra
note 52.
171. See Devine et al., supra note 165, at 625–28, 670, 677, 684. The expression
simulated trials actually hide different dimensions of research, such as sample employed—
undergraduates or former jurors; sample size; trial presentation medium—written description
of facts, written summaries, transcripts, audio or video-taped testimonies, live re-enactments;
trial settings—classroom, actual courtrooms, mock courtroom, laboratories; trials elements—
voir dire, opening and closing statements, witnesses, jury instructions, jury deliberations;
and measurement of dependent variable—dichotomous, guilty or not-guilty, or continuous,
estimation of probability of guilt. However, due to financial, practical, legal, and ethical
constraints, most simulated trials tend to be composed of small samples of undergraduate
200
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
One benefit of this method is that researchers can systematically and
predictably manipulate variables while excluding unimportant or confounding
variables.172 However, whatever is gained in “internal validity”—the degree
to which accurate conclusions can be drawn about what causes what or
the ability of a study to rule out alternative explanations—is lost in “external
validity”—the degree to which observed effects or relationships can be
expected to hold in other settings or the extent to which results generalize.173
This has led the legal profession and courts, with their high premium
on external validity, to be skeptical of jury decision-making studies as
well as on other types of psychological research on law-related topics.174
In other words, there is a concern that studies involving undergraduate college
students in simulated trials consisting almost entirely of short written
summaries do not yield relevant or predictive information for real world
application.
Although it is important to recognize the limitations of research, the
answer to the question of external validity is by no means straightforward.
In fact, supporters of jury decision-making research often argue that criticism
against such research is often accompanied by a lack of understanding
about methodology.175 They argue that the lack of realism of simulated studies
students who are asked to read a brief written summary of facts of a case in a laboratory
setting. See Joel D. Lieberman & Bruce D. Sales, Jury Instructions: Past, Present, and
Future, 6 PSYCHOL., PUB. POL’Y & L. 587, 589 (2000); Tara L. Mitchell, et al., Racial Bias
in Mock Juror Decision-Making: A Meta-Analytic Review of Defendant Treatment, 29
LAW & HUM. BEHAV. 621, 624 (2005). This makes findings of jury simulations less relevant
to actual cases, which are notoriously lengthier and more complex.
172. See, e.g., Thompson, supra note 166, at 204, 207.
173. DEVINE, supra note 8, at 10 (emphasis omitted); Devine et al., supra note 165,
at 698.
174. See Kerri F. Dunn, Assessing the External Validity of Jury Simulation Research: A
Meta-Analysis 2 (Aug. 2002) (unpublished Ph.D. dissertation, University of Nebraska) (on
file with author). For instance, in Lockhart v. McCree, the Supreme Court rejected fourteen out
of fifteen studies presented on the basis of supposed methodological problems. 476 U.S.
162, 169–73 (1986).
175. See, e.g., Norbert L. Kerr & Robert M. Bray, Simulation, Realism, and the Study
of the Jury, in PSYCHOLOGY AND LAW: AN EMPIRICAL PERSPECTIVE 322, 322 (Neil Brewer
& Kipling D. Williams eds., 2005) (noting it is common for critics to confuse verisimilitude for
ecological validity and ecological validity for external validity). There is also research showing
that there is little difference between samples constituted by students and nonstudents.
See, e.g., Brian H. Bornstein, The Ecological Validity of Jury Simulations: Is the Jury Still
Out?, 23 LAW & HUM. BEHAV. 75, 75, 78–81 (1999) (citations omitted); Joel D. Lieberman
& Bruce D. Sales, What Social Science Teaches Us About the Jury Instruction Process, 3
PSYCHOL., PUB. POL’Y & L. 589, 617, 623, 633 (1997) (citations omitted).
201
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
does not necessarily translate into a lack of external validity, nor does it
indicate that the studies’ findings are any less useful.176 Despite the legal
profession’s skepticism, research about jury decision-making might contribute
to a better understanding of how jurors make decisions.177 In fact, the
fields of psychology and law have already made substantial progress in many
interesting and relevant topics for the legal profession. Examples of topics
include eyewitnesses, 178 pretrial publicity, 179 jury selection, 180 false
convictions,181 and jury instructions.182
Here, I want to highlight a few key findings from jury decision-making
research that suggests that when deciding cases, jurors are not indifferent
to error distribution across different situations. A group of studies focuses
on the “severity-leniency hypothesis.”183 This hypothesis “asserts that
jurors . . . are less willing to risk convicting an innocent person as the negative
consequences”—the costs—associated with a guilty verdict increase.184 In
other words, there is a tradeoff between the perceived severity of the charge,
or of the punishment, and the likelihood of conviction: the more severe
the crime, the lower the probability of a guilty verdict. Although not explicitly
about variations in the standards of proof, these studies suggest that when
the charges or prescribed penalties are more severe, jurors reduce the
likelihood of mistakenly punishing an innocent man by requiring more
176. DeMatteo & Anumba, supra note 154, at 16; see also Kerr & Bray, supra note
175, at 322–23, 358; Thompson, supra note 166, at 204–05, 215.
177. See, e.g., HASTIE ET AL., supra note 8, at 238; Thompson, supra note 166, at
204, 207 (finding that use of simulated settings does not affect juror decision-making and
that the manner in which researchers present the trial to mock jurors rarely influences the
outcome); Bornstein, supra note 175, at 75, 78–81 (citations omitted).
178. See generally, e.g., Howard E. Egeth, What Do We Not Know About Eyewitness
Identification?, 48 AM. PSYCHOLOGIST 577 (1993).
179. See generally, e.g., Christina A. Studebaker et al., Studying Pretrial Publicity
Effects: New Methods for Improving Ecological Validity and Testing External Validity, 26
LAW & HUM. BEHAV. 19 (2002).
180. See generally, e.g., JOEL D. LIEBERMAN & BRUCE D. SALES, SCIENTIFIC JURY
SELECTION (2007).
181. See generally, e.g., Saul M. Kassin & Gisli H. Gudjonsson, The Psychology of
Confessions: A Review of the Literature and Issues, 5 PSYCHOL. SCI. PUB. INT. 33 (2004).
182. See generally, e.g., Lieberman, supra note 12.
183. DEVINE, supra note 8, at 79–80 (emphasis omitted). Norbert Kerr makes a
distinction between a hypothesis, according to which the more severe the penalty prescribed for
an offense, the more evidence of guilt necessary for conviction—which he calls the “severity–
criterion hypothesis”—and the more severe the penalty prescribed for an offense, the lower
the likelihood of conviction—which he calls the “severity–leniency hypothesis.” Norbert
L. Kerr, Severity of Prescribed Penalty and Mock Jurors’ Verdicts, 36 J. PERSONALITY &
SOC. PSYCHOL. 1431, 1431–32 (1978). In this Article, I will use the expression severity-leniency
hypothesis to refer to both hypotheses interchangeably.
184. DEVINE, supra note 8, at 79–80; Kerr, supra note 183.
202
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
conclusive evidence of guilt.185 But this is just another way of saying that
when the charge or prescribed penalty is more severe, jurors are more
willing to raise the de facto standard of proof.
Many in the legal profession widely accept the validity of the severityleniency hypothesis, even if only anecdotally.186 In their seminal 1966
work on juries, Kalven and Zeisel asked a large group of judges to explain
why they believed jurors in their courts had reached certain verdicts.187
Often, judges speculated that a jury acquittal was the consequence of the
jurors’ feeling that the possible punishment was simply too severe.188
Empirical studies do not univocally confirm or rebut the severity-leniency
hypothesis,189 but available data suggests that the seriousness of the charge
the defendant faces affects verdicts. One study presented 227 mock jurors
with a homicide case and had them choose a verdict from different combinations
of the following charges: first-degree murder, second-degree murder, and
manslaughter.190 Some jurors had to choose between first-degree and seconddegree murder, others between first-degree murder and manslaughter, others
between second-degree murder and manslaughter, and so on for all seven
possible combinations of the three charges plus a control condition where
185. It is important to note, however, that this strategy for reducing the risk of a false
conviction would simultaneously increase the risk of a false acquittal. This means that if
jurors perceive the costs of a false acquittal as sufficiently large, they might well not use
this strategy. For example, if the defendant was charged as a psychopathic killer, the high
cost of setting a guilty person free might restrain one from raising one’s decision-making
threshold. See KALVEN & ZEISEL, supra note 155, at 306–07.
186. See, e.g., Johannes Andenaes, The General Preventive Effects of Punishment,
114 U. PA. L. REV. 949, 970 (1966); John F. Galliher et al., Nebraska’s Marijuana Law:
A Case of Unexpected Legislative Innovation, 8 LAW & SOC. REV. 441, 446–47, 453 (1974).
187. KALVEN & ZEISEL, supra note 155, at 287–88.
188. Id. at 306–07.
189. See, e.g., DEVINE, supra note 8, at 82 (citing Jonathan L. Freedman et al., Severity
of Penalty, Seriousness of the Charge, and Mock Jurors’ Verdicts, 18 LAW & HUM. BEHAV.
189, 202 (1994)).
190. Neil Vidmar, Effect of Decision Alternatives on the Verdicts and Social
Perceptions of Simulated Jurors, 22 J. PERSONALITY & SOC. PSYCHOL. 211, 215–17 (1972)
(finding a greater percentage (p < .001) of innocent verdicts when mock jurors were
presented with the alternatives “not guilty” or guilty of “first-degree murder” (54%) than
when allowed to consider the less severe charges of “manslaughter” or “second-degree
murder” (8%)). However, the author himself interpreted his results with caution because
the trials confounded the severity of charges with evidentiary requirements for conviction.
See id. at 216. For instance, charges of first-degree murder required the jury to find
intention to kill and premeditation, something not necessary for a manslaughter charge.
Id.
203
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
no verdict alternative was given.191 Jurors were less likely to convict when
the least-serious charge—manslaughter—was not listed among the
possibilities192—a result consistent with the severity-leniency hypothesis.
More specifically, “under conditions of restricted decision alternatives,
the more severe the degree of guilt associated with the least severe guilt
alternative, the greater were the chances of obtaining a not guilty verdict.”193
Subsequent studies involving mock trials similarly found an inverse trend
between jury verdicts and sanction severity.194
191. Id.
192. Id.
193. Id. at 215.
194. See generally, e.g., Phoebe C. Ellsworth & Lee Ross, Public Opinion and
Judicial Decision Making: An Example from Research on Capital Punishment, in CAPITAL
PUNISHMENT IN THE UNITED STATES 152 (Hugo Adam Bedau & Chester M. Pierce eds.,
1976) (finding that subjects answered positively to the question of whether they would
need more evidence to vote guilty if the penalty was death instead of life imprisonment);
V. Lee Hamilton, Obedience and Responsibility: A Jury Simulation, 36 J. PERSONALITY &
SOC. PSYCHOL. 126 (1978) (finding that for a subset of cases considered, restricting the
available verdict options lowered convictions); Martin F. Kaplan & Sharon Krupa, Severe
Penalties Under the Control of Others Can Reduce Guilt Verdicts, 10 LAW & PSYCHOL.
REV. 1 (1986) (finding penalty severity interacted with several other variables in a highly
complex fashion); Kalman J. Kaplan & Roger I. Simon, Latitude and Severity of Sentencing
Options, Race of the Victim and Decisions of Simulated Jurors: Some Issues Arising from
the “Algiers Motel” Trial, 7 LAW & SOC. REV. 87 (1972) (finding that the severity of
punishment associated with the guilty verdict is inversely related to the percentage of
guilty decisions); Kerr, supra note 183 (“Increasing the severity of the prescribed penalty
for an offence resulted in an adjustment of subjects’ conviction criteria such that more
proof of guilt was required for conviction and this resulted in a reduced probability of
conviction.”); Chantal Mees Koch & Dennis J. Devine, Effects of Reasonable Doubt
Definition and Inclusion of a Lesser Charge on Jury Verdicts, 23 LAW & HUM. BEHAV.
653 (1999) (“Juries with the option to convict on a lesser charge produced more overall
convictions than juries receiving only the primary charge, but only when ‘reasonable
doubt’ was undefined.”); William C. McComas & Mark E. Noll, Effects of Seriousness of
Charge and Punishment Severity on the Judgments of Simulated Jurors, 24 PSYCHOL. REC.
545 (1974) (manipulating the seriousness of the charge and the severity of the penalty and
finding that only the former, not the later, produced a significant effect on jury verdicts);
Rita James Simon & Linda Mahan, Quantifying Burdens of Proof: A View from the
Bench, the Jury, and the Classroom, 5 LAW & SOC. REV. 319 (1971) (finding that the lower
the penalty for an offense, the lower the probability that subjects thought defendant had
committed the crime he was being charged with). But see James H. Davis et al., Victim
Consequences, Sentence Severity, and Decision Processes in Mock Juries, 18 ORGANIZATIONAL
BEHAV. & HUM. PERFORMANCE 346 (1977) (finding that manipulating the severity of the
penalty associated with a rape charge using six-person mock juries has no effect on individual
judgments of guilt or on jury verdicts); Freedman et al., supra note 189 (finding an effect
for charge seriousness and no effect for penalty severity in an initial study, but the effect
of charge seriousness disappeared when the strength of evidence was held constant, but
still arguing that there are reasons to believe that severity of penalty might affect real—as
opposed to simulated—juries).
204
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
Field studies also offer some support for the severity-leniency hypothesis.
One study analyzed archival data from 201 jury trials in Indianapolis and
found a negative relationship between crime severity and the likelihood
of conviction when enough variables were taken into account.195 Another
study looked into 206 criminal trials in Utah and found a correlation—
albeit weak—between the seriousness of the charge and the jury’s verdict.196
These results suggest broader implications than the effects of severity of
charges or penalty. Any factor that can alter a juror’s perception of costs
associated with possible errors—for example, conditions of imprisonment,
attitudes towards the defendant, probability of recurrence—may influence
the juror’s decision-making threshold for conviction, that is the juror’s de
facto standard of proof.
The literature about juries in civil cases is less developed than its criminal
counterpart. Still, existing research has pointed to key variables that exert
verifiable influence over the frequency and magnitude of damage awards.197
One such variable is the reprehensibility of the parties’ conduct, particularly
the defendant. This tends to affect the amount of damages juries award.
Studies also show a bias against deep pocket defendants. In a study analyzing
over twenty years of verdicts in Cook County, Illinois, researchers found
juries were more likely to award more money to plaintiffs when corporations
appeared as defendants.198 Similar results were found in mock trial studies.199
195. Martha A. Myers, Rule Departures and Making Law: Juries and Their Verdicts,
13 LAW & SOC. REV. 781, 785, 795 (1979).
196. See Carol Werner et al., The Impact of Case Characteristics and Prior Jury
Experience on Jury Verdicts, 15 J. APPLIED SOC. PSYCHOL. 409, 409–10 (1985). But see
Dennis J. Devine et al., Strength of Evidence, Extraevidentiary Influence, and the Liberation
Hypothesis: Data from the Field, 33 LAW & HUM. BEHAV. 136, 145–46 (2009) (finding
real jurors were actually more willing to convict under lower levels of certainty when faced
with the prospect of errantly returning a violent perpetrator to their community).
197. See EDIE GREENE & BRIAN H. BORNSTEIN, DETERMINING DAMAGES: THE
PSYCHOLOGY OF JURY AWARDS 25 (2003). For a skeptical view of civil jury verdicts, see
CASS R. SUNSTEIN ET AL., PUNITIVE DAMAGES: HOW JURIES DECIDE 248–49 (2002) (describing
civil jury verdicts as essentially groundless).
198. AUDREY CHIN & MARK A. PETERSON, DEEP POCKETS, EMPTY POCKETS: WHO
WINS IN COOK COUNTY JURY TRIALS 43 (1985).
199. See, e.g., Valerie P. Hans & M. David Ermann, Responses to Corporate Versus
Individual Wrongdoing, 13 LAW & HUM. BEHAV. 151, 157, 163 (1989) (finding juries were
more likely to award plaintiffs higher damages when the defendant was referred to as
“Jones Corporation” than as “Mr. Jones”); Jennifer K. Robbennolt, Punitive Damage
Decision Making: The Decisions of Citizens and Trial Court Judges, 26 LAW & HUM. BEHAV.
315, 315 (2002) (finding judge and jury punitive damage awards “were influenced by
wealth of the defendant”). But see VALERIE P. HANS, BUSINESS ON TRIAL: THE CIVIL JURY
205
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
Although not explicitly related to variations in the standard of proof, these
studies suggest jurors are not indifferent between errors in favor of plaintiffs
or defendants depending on the conduct involved and who is being sued.
Reactions to results like these have been mixed. At times, they take the
form of criticizing the findings or methodologies of the studies.200
Common critiques include the fact that most subjects are students201 and
that researchers provide a short supply of information for subjects to make
estimates, presumably less than real trials.202 Other reactions to research
on jury decision-making take the form of proposals to reform the legal
system. Some have suggested instruments to increase the precision and
rigor of standards such as beyond reasonable doubt.203 Others have advocated
for changes to jury instructions, ranging from more precise instruction to
doing away with the court’s definition of reasonable doubt altogether.204
Albeit different, these proposals share a common feature: they all assume
something is broken and needs to be fixed. More specifically, these suggestions
assume there can only be one level of the standard of proof in criminal cases
AND CORPORATE RESPONSIBILITY
216–17 (2000) (describing pro-plaintiff juries and antibusiness juries as myths); Robert J. MacCoun, Differential Treatment of Corporate Defendants
by Juries: An Examination of the “Deep-Pockets” Hypothesis, 30 LAW & SOC’Y REV. 121,
121 (1996) (“While juries do appear to treat corporations differently, the explanation may
have more to do with citizens’ views about the special risks and responsibilities of
commercial activity.”).
200. See, e.g., Lillquist, supra note 52, at 112.
201. The claim that students are representative of groups of citizens called for jury duty
is controversial. Compare Jonathan D. Casper & Kennette M. Benedict, The Influence of
Outcome Information and Attitudes on Juror Decision Making in Search and Seizure
Cases, in INSIDE THE JUROR: THE PSYCHOLOGY OF JUROR DECISION MAKING 65, 78 (Reid
Hastie ed., 2004) (finding “no substantial differences in the patterns of awards” between
citizens called for jury duty and students), with Robert J. MacCoun & Norbert L. Kerr,
Asymmetric Influence in Mock Jury Deliberation: Jurors’ Bias for Leniency, 54 J.
PERSONALITY & SOC. PSYCHOL. 21, 21 (1988) (finding differences), and Simon & Mahan,
supra note 194, at 322.
202. Lillquist, supra note 52, at 115.
203. See, e.g., Jon O. Newman, Beyond “Reasonable Doubt,” 68 N.Y.U. L. REV.
979, 979 (1993); Robert C. Power, Reasonable and Other Doubts: The Problem of Jury
Instructions, 67 TENN. L. REV. 45, 122–23 (1999) (proposing simplifying and clarifying
jury instructions); Lawrence M. Solan, Refocusing the Burden of Proof in Criminal Cases:
Some Doubt About Reasonable Doubt, 78 TEX. L. REV. 105, 105, 147 (1999).
204. See AMIRAM ELWORK ET AL., MAKING JURY INSTRUCTIONS UNDERSTANDABLE
25–26 (1982) (describing several approaches to improve comprehensability of jury
instructions); Jessica N. Cohen, The Reasonable Doubt Jury Instruction: Giving Meaning to
a Critical Concept, 22 AM. J. CRIM. L. 677, 678 (1995) (demonstrating why the reasonable
doubt standard should be more thouroughly “defined for the jury”); Shelagh Kenney, Fifth
Amendment—Upholding the Constitutional Merit of Misleading Reasonable Doubt Jury
Instructions, 85 J. CRIM. L. & CRIMINOLOGY 989, 1025–27 (1995) (arguing for improvement in
comprehensibility of reasonable doubt instructions). But see Note, Reasonable Doubt: An
Argument Against Definition, 108 HARV. L. REV. 1955, 1955, 1972 (1995) (“[C]ourts should
not attempt to define the . . . reasonable doubt [standard] to juries.”).
206
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
and most civil cases. Any variation is an anomaly that should be fought
against. This Article is intended to combat that assumption. The perceived
anomaly should instead be the norm. More importantly, data about how
juries decide cases strongly suggests our reality is already one of varying
standards of proof.
VI. OBJECTIONS AGAINST AND REPLIES
This Part addresses potential objections against this Article’s proposal
and offers initial responses. It begins by discussing the claim that a system
with varying standards of proof would lead to prohibitively high administrative
costs and necessarily have unpredictable dynamic effects. Next, it examines
the objection that individuals’ cognitive limitations make the current system
with few constant standards an optimal scheme. Lastly, it explores the
objection that the current proposal is moot because the legal system already
adjusts the distribution of error across different types of cases through
legal instruments other than standards of proof.
A. High Administrative Costs and Unpredictable Effects
Ronald Allen and Alex Stein raised a series of objections to Kaplow’s
analysis that are equally relevant to the proposal in this Article.205 They
started by pointing out the potentially high administrative costs associated
with a system that attempts to define the optimal standard of proof based
on a welfare analysis.206 This is a significant concern about any policy
proposal and ought to be taken seriously. I argue, however, that objectors
tend to overestimate this point.
A related objection concerns the difficulty of predicting the dynamic effects
of a system with varying standards of proof. Let us pretend we were able
to meet all the required administrative costs by focusing our efforts on
changing the applicable standard for one category of cases, such as unpaid
overtime claims. We have also successfully listed all expected social and
205. See Allen & Stein, supra note 9, at 582 (“To operationalize Kaplow’s proposal
requires policymakers to articulate all these categories, along with all other forms of
human activity, and gather dependable information about the mix of harms and benefits
associated with each category. This is outlandish.”); see also CLERMONT, supra note 9.
But see Kaplow, Burden of Proof, supra note 4, at 772 n.59 (“[T]he information requirements
for determination of the optimal evidence threshold do not differ greatly from those for
the determinants for the preponderance rule or other rules based on ex post likelihoods.”).
206. See Allen & Stein, supra note 9, at 580–81.
207
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
private harms and benefits associated with the acts as well as additional
fairness and distributive considerations, and we have settled on a level of
standard of proof we expect to be socially acceptable. According to objectors,
the problem is that once a standard is in place, people will modify their
behavior to conform to the new standard. We should hope so, because
deterrence operates through behavioral changes. However, the objection
persists; it is hard to predict the new equilibrium after people have adjusted
to the new standard.207 We would have to be able to predict avoidance and
exploitation efforts and continuously adjust standards. But, again, “[t]he
amount of information that these [constant] adjustments will require [would
be] unrealistic.” 208 For objectors, the twin problems of high costs and
unpredictable dynamic effects will most likely offset any potential gain
following from a system with varying standards of proof.
These two related objections touch on a more general point about the
function of standards as mechanisms for distributing errors. According to
some evidence scholars, standards simply cannot serve this function.209
This is because, to guarantee a particular error distribution occurs, we need
information concerning, among other things, the distribution of truly guilty
and truly innocent people who go to trial, as discussed above. For example,
imagine we have fifty defendants facing criminal charges; suppose forty
are innocent and ten are guilty.210 With a standard of proof of 0.9, we can
predict that the jurors will wrongly convict about 10% of the innocent,
which means a total of four.211 If we design our standard to distribute
errors according to the idea that we should let ten guilty people go free for
every innocent we condemn, we would expect to let forty guilty people
go free. However, even if we let all of the guilty free, we would still fall
short of our targeted error ratio of false acquittals to false convictions.212
This suggests that just defining the standard one way or the other offers
no guarantee that the desired error ratio will be verified. We need information
207. Ronald J. Allen, Rationality and the Taming of Complexity, 62 ALA. L. REV. 1047,
1063–66 (2011) [hereinafter Allen, Rationality].
208. Allen & Stein, supra note 9, at 583.
209. See, e.g., LAUDAN, supra note 11, at 73. See generally Allen, Juridical Proof, supra
note 11; Allen, Rationality, supra note 207; Pardo, supra note 11; Zoë Johnson King, The
Trouble with Standards of Proof (Apr. 15, 2016) (unpublished manuscript) (on file with
author).
210. See LAUDAN, supra note 11, at 73–74.
211. The standard of proof in this example is represented probabilistically for heuristic
purposes only. For a critique of that approach to standards, see supra Part II.
212. See LAUDAN, supra note 11, at 74 (arguing in favor of a ratio of true acquittals
to false convictions instead of false acquittals to false convictions).
208
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
about the world. The problem is that this information is difficult, if not
impossible, to obtain.213
There are at least two possible reactions to this difficulty. We can surrender
the idea that standards can serve as mechanisms to distribute errors.
Alternatively, we can change our understanding of how standards fulfill
this function. Here, I explore this second option. Only after exhausting our
efforts can we concede this very plausible and important role we assign to
standards.
One way to change our understanding of how standards of proof can
fulfill their function of distributing errors is by lowering our expectations
about our capacity to verify whether we have reached an exact errordistribution. An analogy might be helpful here. When discussing legal
rules, Frederick Schauer treats them as “prescriptive generalizations.”214
The desire to promote safe driving and the idea that most people when
driving over fifty-five miles per hour on a street are prone to unsafe driving,
allows us to generalize that all people driving over fifty-five miles per
hour on that street should pay a fine. Because this generalization is necessarily
over- and under-inclusive—there might be people who drive safely over
fifty-five—the results indicated by applying a rule will, in some cases, be
inferior to the results indicated by directly applying the rule’s background
justifications. This suggests that rule-based decision-making will be suboptimal,
failing to achieve the best result on every occasion—the result in which
the justification underlying each generalization is correctly applied to each
case.215 The fact that rules are suboptimal does not mean that the procedure
of deciding according to rules is suboptimal. In fact, the optimal decision
procedure may not be the one aimed at producing the optimal result in every
case. A procedure aimed at optimality in every individual case may produce
worse results in the aggregate. In this sense, rules are second-best solutions.
213. This challenge increases once we try to include in the analysis the effects of
other legal institutions on the error-distribution function of standards. Given their ubiquity
in contemporary adjudicatory systems, plea bargains and settlements stand out as important
examples to be analyzed. One way in which pleas may impact error-distributions is by
changing the distribution of truly guilty and truly innocent who go to trial. But how can we
measure those effects? Are they the same for all types of cases? Or should we expect the
effect of pleas for the distribution of truly guilty and truly innocent that go to trial to differ
depending on the type of crime and type of defendant involved?
214. FREDERICK SCHAUER, PLAYING BY THE RULES: A PHILOSOPHICAL EXAMINATION
OF RULE-BASED DECISION-MAKING IN LAW AND IN LIFE 25–27 (1991).
215. Id. at 100–02. See generally Frederick Schauer, In Defense of Rule-Based Evidence
Law—and Epistemology Too, 5 EPISTEME 295 (2008).
209
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
I suggest that something similar is true of standards of proof. We can see
them as generalizations with a proper underlying justification. In the case
of standards, this justification is the goal of achieving a given error distribution.
Similar to legal rules, standards can be seen as suboptimal generalizations.
But their suboptimality does not come from over- and under-inclusiveness.
Rather, it comes from constraints on whether we can verify that the targeted
error distribution has been achieved. These constraints are both epistemic—
we have imperfect knowledge about the distribution of truly guilty and
truly innocent that go to trial—and formal—we cannot wait for relevant
data to decide cases, nor can we revisit them if future data suggests something
different from what was thought when the decision was made.
It is important to note that the generalizations that standards represent
are not arbitrary; they are not spurious generalizations. First, we have
plausible intuitions about the expected distribution of truly guilty and truly
innocent people as well as the potential effects that establishing the standards
at different levels can cause. To the extent that these intuitions are plausible
and coherent with our background beliefs, they are prima facie epistemically
justified.216 Second, we can periodically stop to re-evaluate the standards.
If we notice any great deviation between our reality and our ambitions, we
can make the appropriate adjustments.217
More generally, this type of academic debunking can be aimed at any
policy analysis. The problem is that discarding a policy proposal as mere
speculation until further empirical information becomes available imposes
unreasonable constraints on policy-oriented scholarship. It might be premature
to dismiss recommendations simply for lack of sufficient data. Surely,
any model becomes increasingly dangerous as it ignores existing data.
216. This is based on the idea that the fact that a person has an intuition provides that
person with prima facie epistemic justification in believing the content of the intuition. In
other words, the fact that a particular proposition is intuitive provides us with a prima facie
justification for believing in it. For a more formal and extensive discussion about the role
of rational intuition in the epistemic justification for some of our belief about the empirical
world see, for example, LAURENCE BONJOUR, IN DEFENSE OF PURE REASON: A RATIONALIST
ACCOUNT OF A PRIORI JUSTIFICATION 100–06 (1998). But see generally R. M. Hare, Critical
Study: Rawls’ Theory of Justice—II, 23 PHIL. Q. 241 (1973) (arguing that a set of coherent
beliefs that have no independent initial credibility cannot produce epistemic or moral
justification because coherent fictions are still only fictions).
217. One might hold onto the seemingly plausible assumption that the justice system
as a whole works to create more evidence against truly guilty defendants than against truly
innocent defendants. See, e.g., Bell, supra note 19, at 560–62. The hope is that we would
then be able to estimate the prior allocation of truly guilty and truly innocent defendants
at trial and, consequentially, calibrate the error-distribution based on the level of the
standard of proof. This strategy is doomed to fail, however. It is not enough for a complete
analysis of error distribution that the justice system as a whole works to create more
evidence against guilty defendants than against innocent defendants. It must do so in a
consistent and predictable manner. Otherwise, no theoretical analysis can be successful.
210
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
However, absent good substitutes, models that include currently unverified,
but verifiable, assumptions about our empirical world can provide helpful
starting points. Most importantly, these starting points need not be spurious
assumptions. We can build our starting points on top of plausible intuitions
about a range of empirical phenomena, including the expected distribution
of truly guilty and truly innocent people and the number of individuals
from each group that goes to trial with a given set of facts. These intuitions
can give us prima facie epistemic justification for believing their content.
We should not remain uncritical, however. We can—and should—pause
from time to time to re-evaluate our beliefs and practices. If we note any
great deviation between reality and our ambitions, we should make the
appropriate adjustments and move forward. It makes little sense to substitute
some other model because it was easier to employ when that substitute
essentially has no relationship to our objectives.218 In this sense, we are
like Neurath’s sailors, who have to rebuild their ship on the open sea.219
One last point about these objections concerns the underlying institutional
details of this Article’s proposal. One configuration that might exacerbate
these difficulties is gathering all that information to design the optimal
standard and making relevant moral and political decisions on a case-bycase basis.220 Although the optimal level of generality of standards of proof
is beyond the scope of this Article, it is reasonable to suppose that—given
the amount of information needed to perform this analysis combined with
the fact that this information is likely to vary between different cases—
requiring courts to define the optimal standard based on a mix of welfare,
fairness, and distributive considerations for every case would be prohibitively
costly. This would be asking courts to perform tasks they might not be
accustomed or equipped to perform properly.221
218. The alternative with the lowest informational requirements would be to simply
toss a coin, but the ease of this method gives little reason in its favor.
219. See OTTO NEURATH, PHILOSOPHICAL PAPERS 1913–1946, at 92 (Robert S. Cohen &
Marie Neurath eds. & trans., 1983).
220. For arguments in defense of a case-by-case determination of standards of proof,
see, for example, Lillquist, supra note 52, at 147–62; Dale A. Nance, Evidential Completeness
and the Burden of Proof, 49 HASTINGS L.J. 621, 626–32 (1998). A case-by-case proposal
also exacerbates the difficulty behind individuating and counting standards. Thank you to
William Twining for elucidation on this point.
221. See Allen & Stein, supra note 9, at 601; see also Tribe, supra note 30, at 1384–
85 (“[O]ne will expect the lawmaker rather than the factfinder to use a model such as the
one Kaplan and Cullison propose, and one will define the decision problem to be solved
not as the one-shot problem of fixing a standard of proof for a particular trial with four
211
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
Moreover, requiring judges and jurors to define the standard based on
welfare, fairness, or distributive considerations might increase the possibility
of similarly situated individuals being subjected to different standards of
proof. This, in turn, might raise equal protection obstacles. The recommendation
that we view standards as prescriptive generalizations suggests a system
not very much unlike ours, which assigns standards ex ante by elected
representatives, instead of on a case-by-case basis. The main difference
is that we could presumably track societal choices about preferable errordistribution in different circumstances more closely. This alternative would
impose lower informational requirements on decision-makers in comparison
to a case-by-case alternative because decisions are made in the aggregate
by decision-makers institutionally equipped with the necessary tools to perform
those tasks. Most importantly, such a system would be preferable, not
because these institutions are more likely to get decisions right but because
this is a democratically fair way to decide politically and morally sensitive
issues that reasonable people disagree about.222
B. Cognitive Limitations
Kevin Clermont has argued that findings from cognitive psychology
literature give us good reasons to limit the number of standards of proof
into our well-known trinity.223 Clermont recognizes that a system of varying
standards is conceivable, and perhaps even theoretically, desirable.224 But
he quickly warns about practical dangers. Specifically, he refers to research
about individuals’ cognitive limitations to judge probabilities.225 Studies
suggest that judgments about probabilities are carried out in terms of a
very limited set of broad and fuzzy categories such as more likely than
possible outcomes, but as the much larger problem of establishing such standards for the
trial system as a whole.”). The Supreme Court argued similarly:
[T]his court never has approved case-by-case determination of the proper standard
of proof for a given proceeding. Standards of proof, like other “procedural due
process rules, are shaped by the risk of error inherent in the truth-finding process
as applied to the generality of cases, not the rare exceptions.” [Because] the litigants
and the factfinder must know at the outset of a given proceeding how the risk of
error will be allocated, the standard of proof necessarily must be calibrated in
advance.
Santosky v. Kramer, 455 U.S. 745, 757 (1982) (emphasis omitted) (quoting Mathews v. Eldridge,
424 U.S. 319, 344 (1976)).
222. See DWORKIN, supra note 5, at 24, 27.
223. See CLERMONT, supra note 9, at 113; see also Kevin M. Clermont, Procedure’s
Magical Number Three: Psychological Bases for Standards of Decision, 72 CORNELL L.
REV. 1115, 1146 (1987).
224. See CLERMONT, supra note 9, at 112.
225. See id. at 107.
212
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
not, high probability, and almost certainty.226 For Clermont, the existing
coarsely gradated standards of proof in law track these fuzzy categories.227
This allows legal decision-makers to use their built-in cognitive apparatus
while judging factual propositions in legal cases according to the three
familiar standards of proof. Any deviation from those hard-wired categories
would impose significant cognitive costs on individuals and increase the
chance of errors. This means that any proposal attempting to change
standards would have to compensate for those additional costs. Consider
Clermont’s words:
[L]awmakers can adequately serve the often imprecise policies underlying the
standard of decision by choosing from the set of seven categories, thus moving
up or down by fairly small quantum leaps rather than by unrealistically finer degrees.
....
[Any novel standard of proof] would be difficult to articulate and communicate
and comprehend and apply, would risk confusion and possibly abuse by the
decisionmaker, and would inevitably drift toward one of the customary standards
—and that these costs more than offset any benefit of utilizing an unusual standard [of
proof].228
It should not be surprising that the laypeople are not particularly good
at reasoning with probabilities. Decades of psychological studies have helped
to formulate, explore, and catalog the distinctive, persistent, and predictable
errors in people’s probabilistic reasoning, such as base-rate fallacy,229
gambler’s fallacy,230 and representativeness heuristic.231
This argument, however, has a special grip against probabilistic approaches
to standards of proof. According to these approaches, standards of proof
should be understood as probability thresholds ranging from zero to one,
226. Id. at 115.
227. See generally, e.g., David L. Schwartz & Christopher B. Seaman, Standards of
Proof in Civil Litigation: An Experiment from Patent Law, 26 HARV. J.L. & TECH. 429 (2013)
(conducting an experiment which showed that jury instructions tended to move away from
clear and convincing towards preponderance of the evidence).
228. CLERMONT, supra note 9, at 79–81.
229. See, e.g., Amos Tversky & Daniel Kahneman, Evidential Impact of Base Rates,
in JUDGMENT UNDER UNCERTAINTY: HEURISTICS AND BIASES 153, 153–60 (Daniel Kahneman
et al. eds., 1982); Maya Bar-Hillel, The Base-Rate Fallacy in Probability Judgments, 44
ACTA PSYCHOLOGICA 21, 213 (1980).
230. See, e.g., Amos Tversky & Daniel Kahneman, Judgment Under Uncertainty:
Heuristics and Biases, 185 SCIENCE 1124, 1125–30 (1974).
231. See, e.g., Daniel Kahneman & Amos Tversky, Subjective Probability: A Judgment
of Representativeness, 3 COGNITIVE PSYCHOL. 430, 430–31 (1972).
213
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
involving either objective or subjective probabilities.232 People who subscribe
to this interpretation commonly assume that preponderance of the evidence
means that it is proven by a probability higher than 0.5, while beyond a
reasonable doubt would require a much higher probability of conviction;
in fact, it is not uncommon to hear unofficial estimates of 0.9, 0.95, or
0.99.233
Some evidence law scholars push for a different approach to standards,
away from traditional probabilistic approaches. Under an explanation-based
approach, legal fact-finding involves a determination of the best explanation
of the admitted evidence rather than a determination of whether each
element is proven to a specific probability.234 In a nutshell, from the fact
that a given hypothesis—out of the set of hypotheses offered by the parties
or constructed by the fact-finders—best explains the admitted evidence,
fact-finders should infer the truth of that hypothesis and then decide the
case based on that inference. Because this approach removes probabilistic
reasoning as the only important consideration when dealing with standards
of proof, the fact that most people are bad at reasoning with probabilities
loses much of its argumentative strength.
Proponents of the explanation-based approach highlight how the approach
provides a better description of jurors’ reasoning.235 To support this, they
offer evidence that what the approach asks of jurors is compatible with
research on jury decision-making.236 The main finding in the studies is that,
contrary to what the probabilistic approach to standards says of jurors,
individuals do not usually reason with isolated pieces of evidence.237 That
is, we usually do not update our beliefs based on each piece of new information
232. See, e.g., Michael O. Finkelstein & William B. Fairley, A Bayesian Approach
to Identification Evidence, 83 HARV. L. REV. 489, 504 (1970); David Hamer, Probabilistic
Standards of Proof, Their Complements and the Errors that Are Expected to Flow from
Them, 1 U. NEW ENG. L.J. 71, 73–74 (2004); Kaplan, supra note 81, at 1066; D.H. Kaye,
Clarifying the Burden of Persuasion: What Bayesian Decision Rules Do and Do Not Do,
3 INT’L J. EVIDENCE & PROOF 1, 4–5 (1999); Jonathan J. Koehler & Daniel N. Shaviro,
Veridical Verdicts: Increasing Verdict Accuracy Through the Use of Overtly Probabilistic
Evidence and Methods, 75 CORNELL L. REV. 247, 253 (1990); Richard O. Lempert, Modeling
Relevance, 75 MICH. L. REV. 1021, 1124–25 (1977); Mike Redmayne, Exploring the Proof
Paradoxes, 14 LEGAL THEORY 281, 281, 285 (2008).
233. See, e.g., Ronald J. Allen, Explanationism All the Way Down, 5 EPISTEME 320,
320–21 (2008).
234. See Nancy Pennington & Reid Hastie, A Cognitive Theory of Juror Decision
Making: The Story Model, 13 CARDOZO L. REV. 519, 523–25 (1991).
235. See Allen, supra note 22, at 216.
236. See id.
237. See generally Nancy Pennington & Reid Hastie, The Story Model for Juror
Decision Making, in INSIDE THE JUROR: THE PSYCHOLOGY OF JUROR DECISION MAKING,
supra note 201, at 192; Pennington & Hastie, supra note 234.
214
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
we receive. Instead, we tend to construct narratives that will fit the evidence
and make decisions based on the explanatory virtues of those stories.
In a recent piece, Clermont addressed the explanation-based approach
to standards.238 He notes that a difficulty with this approach is that “it
does not track well what the law tells its factfinders about how to proceed,
and it diverges from the law by compelling the nonburdened party to
choose and formulate a competing version of the truth.”239 This is not entirely
accurate. The explanation approach is partly revisionary in that it pushes for
an interpretation of standards that is distinct from the way many people
think about those mechanisms. But, the approach is not completely revisionary.
First, many judicial opinions explicitly reference the importance of explanatory
value in legal reasoning.240 Second, many evidentiary mechanisms are better
understood and explained if we assume they are also meant to promote
explanatory value.241 Moreover, nothing in the explanation approach requires
the non-burdened party to present an alternative explanation to that presented
by the party bearing the burden of persuasion. In a typical civil case, the
defendant can remain silent while the plaintiff tries to convince triers that
he has met his burden—even though this is almost never the best defense
strategy.242
238.
239.
240.
case:
See generally CLERMONT, supra note 9.
Id. at 140.
The Supreme Court employed an explanatory approach in a summary-judgment
Neither the Court of Appeals, nor the respondents, nor the dissent provides
any reason to question the city’s theory. In particular, they do not offer a competing
theory, let alone data, that explain why the elevated crime rates in neighborhoods
with a concentration of adult establishments can be attributed entirely to the
presence of walls between, and separate entrances to, each individual adult operation.
City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 437 (2002).
241. See, e.g., FED. R. EVID. 106 (establishing as admissible surrounding material
relevant to specific testimony); FED. R. EVID. 612 (explaining if a witness relies on a piece
of writing while or before testifying, that writing is admissible regardless of other
exclusionary rules).
242. Even the party that does not carry the burden of persuasion almost always presents
evidence in her favor. The risks of losing and the effects of preclusion create enough
incentive. The fact that the burdens of production and persuasion are on the plaintiff does not
mean that the defendant should rest silent. Theoretically, he can. Strategically he never
should. See How Courts Work, AM. BAR ASSOC., https://www.americanbar.org/ groups/
public_education/resources/law_related_education_network/how_courts_work/defense.html
[https://perma.cc/DKA9-F9PY].
215
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
Most importantly for this Article, an explanation-based approach to
standards can offer a plausible reply to the cognitive limitation objection.
Because the approach does not reduce factual decision-making in light of
standards of proof to probabilistic reasoning even though probability can
still play some role, research showing how individuals are not particularly
good at reasoning with probabilities is not an insurmountable obstacle.
According to Clermont, to the extent that we lack a good reason to
do things differently from what cognitive research tells us we are used to
doing, we should refrain from adopting policies that have a high risk of
leading decision-makers to confusion and error.243 However, we do have
good reasons. Parts IV and V contain several reasons for moving to a system
of varying standards of proof. Those may not be conclusive reasons, but
it should be clear that, contrary to what Clermont claims, we do have
reasons—good reasons—to want to do things differently. And better, we
have reasons to think that the result will not be so pessimistic as the objection
from cognitive limitations suggests.
C. Already Varying De Facto Standards of Proof
The last objection I consider questions my initial premise that our legal
system assigns the same standard of proof to drastically different cases
under the justification that we already accept the same error distribution
in those cases. According to this objection, our legal system already adjusts
the distribution of error across different types of cases, only it does so
through legal tools other than standards of proof. Insofar as these other
instruments impact how easy or difficult it is to impose liability on potential
defendants, they can cause the de facto standard of proof to vary, effectively
changing the error distribution.
Albeit a possibility, the strategy to pursue changes in the de facto
distribution of errors through mechanisms other than standards of proof is
highly questionable. In many circumstances, there will be important
reasons to prefer to adjust the error distribution through changes in the
applicable standard of proof. To see why, consider the following example.244
Suppose that an existing legal rule prohibiting certain acts thought to be
harmful causes a higher number of errors by chilling benign acts. Let us
refer to that set as set X. How do we remedy this problem? Should we
narrow the existing legal rule, perhaps by exempting from liability some
readily identifiable subset of set X (subset x) and leaving only the remaining
set X minus subset x (remainder subset will be subset X\x) subject to liability?
243. See CLERMONT, supra note 9, at 280.
244. The example that follows is adapted from Louis Kaplow, Multistage Adjudication,
126 HARV. L. REV. 1179, 1229–32 (2013).
216
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
Or is it better to increase the formal standard for set X, making it less likely
that those benign acts will be chilled because fewer defendants will be
convicted because of the higher standard, while simultaneously lowering
the standard for cases in the remainder subset X\x?
Let us start with the first strategy. Its benefits seem to depend crucially
on subset x having two features. First, a sizable portion of the benign acts
that may be subject to liability when the law targets the entire initial set X
must be easily mistaken for acts in subset x, those exempted from liability
under the proposed strategy. But the benign acts in set X cannot be
mistaken for those acts in the remaining set X\x, including those acts still
subject to liability. In other words, the benign acts in subset x, currently
exempted from liability, are indistinct from the benign acts initially subject to
liability in set X. Second, subset x must not contain a high number of
harmful acts. Most of the acts in subset x must be truly benign. If these
two conditions hold, then exempting acts in subset x from liability might
greatly reduce chilling effects without greatly undermining deterrence. This
is because most of the acts exempted from liability constitute benign acts.
This result makes the first strategy seem quite appealing.
The problem is that it is highly unlikely that we will ever be able to
identify readily a set of acts with these features. Most of the time, the best
we might do is identify a particular target subset (subset x’) with a slightly
lower ratio of chilling costs to deterrence benefits than the remaining
subset X\x’. That is, a subset x’ containing a higher number of benign acts
compared to harmful acts that might be a plausible candidate for exemption.
There are important reasons to believe that the second strategy, namely
one that increases the applicable standard of proof for cases within x’
while lowering the standard for cases in the remainder subset (X\x’), will
be superior. One way to see why is to note that exempting from liability
acts in new subset x’ is equivalent to setting an infinitely high standard of
proof for all cases involving acts in x’. Remember from Section IV.A that
from a welfare-based analyses, the optimal standard of proof for a given
situation is a function of marginal deterrence gains and chilling costs. If
we were to set standards of proof aimed toward welfare maximization, the
optimal standard in cases involving acts in subset x’, in which there are a
higher number of benign acts than harmful acts, would already tend to be
higher than the optimal standards in cases involving acts in remaining
subset X\x’, where there is a higher number of harmful acts compared to
benign acts. If implemented optimally, this is the result that the second
217
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
3/18/2019 11:17 AM
strategy suggests.245 Note, moreover, that when x’ is indistinguishable from
x, this result is indistinguishable from the result following the first strategy.
Let us consider briefly a situation in which there is no easily identifiable
subset x with a slightly lower ratio of chilling costs to deterrence benefits
compared to remaining subset X\x. Still, lawmakers consider exempting
acts in subset x from liability as a solution to the perception of excessive
chilling of benign acts. Here, it seems that a superior strategy would be
to increase the applicable standard instead of exempting those acts from
liability. Increasing the applicable standard removes those cases from the
legal system with the weakest level of evidential support in favor of the
party carrying the burden of proof. Instead, exempting an arbitrary subset
x from liability would remove cases that vary greatly in terms of evidential
support. We would be removing from not only weak cases, which is a good
thing for maximizing accuracy, but also strong cases, which is a bad thing
for accuracy. With a focus on maximizing accuracy, it seems like a better
strategy not to avoid cases with random levels of evidential support, but
instead to concentrate on those that have the lowest levels.
We end up with a scenario in which either the procedural strategy is
superior or equally preferable to the substantive strategy.246 This gives us
a good reason to prefer the procedural strategy.247 It will usually be best
to alter the error-distribution to a socially desirable ratio through the mechanism
that influences that distribution—standards of proof.
245. Remember that the subset x is one in which the following two favorable conditions
hold. First, a large portion of the benign acts that may be subject to liability when the law
targets the entire initial set X are readily mistaken for acts in x, but not for those acts in
the remaining set X\x. Second, x does not contain very many harmful acts.
246. This does not mean that a substantive strategy will never be preferable. Procedural
rules are not always optimally created and enforced, which can require substantial informational
costs, as discussed above in Section VI.A. Moreover, there may be other benefits associated
with a substantive strategy, such as constraining abuses of discretion and providing clearer
guidance for citizens.
247. Here, one might accuse that I have failed to acknowledge an alleged substantial
cost associated with the procedural strategy. The cost is related to a purported fact that
fixed formal standards would have an important legitimizing function. See, e.g., Charles
Nesson, The Evidence or the Event? On Judicial Proof and the Acceptability of Verdicts,
98 HARV. L. REV. 1357, 1362 (1985); Tribe, supra note 30, at 1393. Insofar as the procedural
strategy advocates for a system with varying standards, it deprives it of its legitimizing
function. Even if I concede that standards might have such legitimizing goal—a questionable
proposition—it is far from clear that a system with fixed standards would have that desired
effect. A system with fixed standards of proof might very well have a delegitimizing effect
instead. This may be because such a system incurs costs related to high chilling or loss of
deterrence—because the fixed standard is far from the optimal standard from a welfarebased analyses—or because such a system is considered unfair—because it fails to give
people procedures that correlate the correct importance to the risk of moral harm they face.
218
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
[VOL. 56: 161, 2019]
3/18/2019 11:17 AM
Varying Standards of Proof
SAN DIEGO LAW REVIEW
VII. CONCLUSION
We often hear the suggestion that, because standards of proof affect the
distribution of errors in adjudication, the choice of the applicable standard
in a case should reflect an assessment of the comparative weights of the
errors involved. At the same time, we tell ourselves that the same standard
should govern every criminal case and, virtually, every civil case. These
two ideas contradict each other. If we are serious about using the socially
desirable error distribution to determine the optimal standard, then we
cannot plausibly also maintain that one fixed standard should govern cases
in which errors for the defendant or plaintiff carry such vastly different
weights. We need to abandon one of these two initial ideas. This Article
argues that the idea of fixed standards should be rejected.
This rejection might seem particularly problematic because today there
appears to be only three viable choices in the range of standards. But this
was not always the case. Only relatively recently did courts recognize that
the legal profession had reduced all conceivable possibilities into three
options.248 Since then, we have become complacent in thinking that our
limited menu exhausts all theoretical and practical possibilities. This does
not have to be the case. Instead, we can—and should—adopt a system with
varying standards of proof.
248. See CLERMONT, supra note 9, at 278; J.P. McBaine, Burden of Proof: Degrees
of Belief, 32 CAL. L. REV. 242, 246–47 (1944); see also Addington v. Texas, 441 U.S. 418,
423 (1979) (“[T]he evolution of this area of the law has produced across a continuum three
standards or levels of proof for different types of cases.”).
219
POST RIBEIRO PAGES.DOCX (DO NOT DELETE)
220
3/18/2019 11:17 AM
RELATED PAPERS
2024 •
2013 •
E. Cancik-Kirschbaum and I. Schrakamp (eds), Transfer, Adaptions und Neukonfiguration von Schrift- und Sprachwissen im Alten Orient, Wiesbaden: Harrassowitz 2022, pp. 363-386
Non-Literal Translations and Re-interpretations: The Reflection of Sumerian in Akkadian Translations and Commentaries in the First Millennium BCE2022 •
Context: Journal of Interdisciplinary Studies
Prikaz vjerskih običaja Bošnjaka iz životnog ciklusa / A Review of Bosniak religious customs related to the lifecycle2022 •
Bulletin de Correspondance Hellénique Moderne et Contemporain
Moving Objects, Images, and Memories: Hamza Bey Mosque/Alcazar Cinema as an Affective Archive of Thessaloniki2021 •
2016 •
1994 •
arXiv (Cornell University)
Multi-Slice Fusion for Sparse-View and Limited-Angle 4D CT Reconstruction2020 •
2011 •
Canadian Journal of Anesthesia/Journal canadien d'anesthésie
Assessment of neuromuscular blockade at the vastus medialis2003 •
Choice Reviews Online
Genetically engineered organisms: benefits and risks1991 •
Sarhad Journal of Agriculture
Growth and Yield Response of Pea (Pisum sativum L.) Cultivars to Phosphorus Fertilization2021 •