Cornell Law Library

Scholarship@Cornell Law: A Digital Repository

Cornell Law Faculty Publications Faculty Scholarship

4-1-2009

Standards of Proof Revisited

Kevin M. Clermont
Cornell Law School, kmc12@cornell.edu

Follow this and additional works at: http://scholarship.law.cornell.edu/facpub

Part of the Civil Law Commons, and the Evidence Commons

Recommended Citation
Clermont, Kevin M., "Standards of Proof Revisited" (2009). Cornell Law Faculty Publications. Paper 13.
http://scholarship.law.cornell.edu/facpub/13

This Article is brought to you for free and open access by the Faculty Scholarship at Scholarship@Cornell Law: A Digital Repository. It has been
accepted for inclusion in Cornell Law Faculty Publications by an authorized administrator of Scholarship@Cornell Law: A Digital Repository. For
more information, please contact jmp8@cornell.edu.
 STANDARDS OF PROOF REVISITED

Kevin M. Clermont*

ABSTRACT

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 fact-
finders 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 through their
standards of proof.

INTRODUCTION

Magisterially, but opaquely, the law tells its fact-finder to apply the
standard of proof. The law will have already chosen the appropriate
standard, or required degree of persuasion, from an array of candidates. In
the United States, that array includes preponderance of the evidence, clear
and convincing proof, and proof beyond a reasonable doubt. At one end of
the array is the preponderance standard, which prevails in civil cases and
translates into "more likely than not." This standard acts to minimize the
expected cost of error if an error against the plaintiff is just as costly as an
error against the defendant. At the other end, proof beyond a reasonable
doubt prevails in criminal cases and means a high degree of probability or
even almost certainty. This standard would also serve to minimize the
expected cost of error because the error of convicting the innocent is
especially costly.'
* Ziff Professor of Law, Cornell University.
t I want to thank Ron Allen, Adrienne Clermont, Christoph Engel, Valerie Hans, Reid
Hastie, David Kaye, John Leubsdorf, Roald Nashi, Jeff Rachlinski, Emily Sherwin, and Wendy
Williams for illuminating conversations and other communications on law and psychology.
1. See Kevin M. Clermont, Procedure's Magical Number Three: Psychological Bases for
Standards of Decision, 72 CORNELL L. REv. 1115, 1118-26 (1987) [hereinafter Clermont, Procedure's
Magical Number] (discussing three examples of original decision-making: standards of proof, different

HeinOnline -- 33 Vt. L. Rev. 469 2008-2009

 Vermont Law Review [Vol. 33:469

The law, most theorists, and I have all proceeded on the assumption
that the fact-finder-in a subsequent step, after having somehow processed
the evidence in order to test a hypothesis or to find tentatively a fact-
dutifully tries, within human limits, to compare the probability of the
burdened party's version of fact to the given standard of proof.2 This
contestable, and contested, leap of faith in assuming a description of
behavior is the subject of this Essay.
What this Essay is not about-and this caution is the critical starting
point-is how the fact-finder processes the evidence in the first place. Thus,
I need not enter the main battlefield of Bayesians versus non-Baysesians
over the model for evidence-processing. Although I myself find the
psychologists' theories on holistic and largely nonprobabilistic assessment
of evidence very well supported, and convincing based on my experience, I
realize that ascertaining the actual approach or even the prescriptive ideal
for processing evidence would be an exceedingly difficult endeavor.
In this Essay, to avoid such difficult problems, I simply assume that the
fact-finder has somehow evaluated a hypothesized fact. Then I focus on the
step when the fact-finder compares the processed proof to the standard of
proof-that is, not the evidence-processing step but the standard-of-proof
step-even if the two steps might not be entirely distinct in operation or
timing. Although the law tries forcibly to make the latter step probabilistic,
I shall consider its reality: How does the fact-finder actually decide whether
the burdened party has established its version of fact strongly enough?
The reader's first reaction could very well be to ask what turns on the
answer to my question. I think quite a lot. Most obviously, how the law sets
and formulates its standards of proof may depend on the answer. The setting
and formulating of these standards are certainly not beyond debate. Most
notably, the Civil Law system prevailing in much of the world applies a high
standard not only to its criminal cases but to all its cases, while our common
law does not apply its high standard to noncriminal cases. I begin there.

theories of the harmless-error doctrine, and standards of investigation used by police). For an idea of a
proof, assume that a dollar mistakenly paid by the defendant (a false positive) is just as costly to society
as a dollar mistakenly uncompensated to the plaintiff (a false negative). Let p be the apparent probability
that the defendant is liable (for D dollars). If p>1/, call it pl; and if p/ , call it P2. On the one hand,
under the preponderance standard, the expected sum of false positives and false negatives over the run
of cases is Y_[(I-pi)D + p2D]. On the other hand, under a very high standard that eliminates false
positives, the analogous sum is YX[pD + p2D]. Because (l-pl) must be less than pl, the preponderance
standard lowers the system's expected error costs. See D.H. Kaye, The Errorof Equal Error Rates, I
LAW PROBABILITY & RISK 3, 7 (2002) [hereinafter Kaye, EqualErrorRates] (arguing that the p> 2 rule
is appealing because "it minimizes expected losses").
2. See, e.g., Richard D. Friedman, "E" Is for Eclectic: Multiple Perspectives on Evidence,
87 VA. L. REv. 2029, 2045-46 (2001) (explaining Bayesian decision theory and giving an example of
its application).

HeinOnline -- 33 Vt. L. Rev. 470 2008-2009

2009] Standardsof ProofRevisited

I. THE CIVIL LAW'S PUZZLE

The Civil Law's position has provoked me to express wonderment in a

series of articles over the years. Instead of asking whether some fact X
(say, that the defendant executed the promissory note disputed in a
noncriminal, or civil, lawsuit) is more likely true than not, the Civil Law
asks whether the fact is so probable as to create an inner and deep-seated
conviction of its truth. That is, instead of a standard requiring merely a
preponderance of the evidence, they apply their intime conviction standard.
What exactly does intime conviction mean? Civilians are usually vague
on this point, only occasionally muttering something definitive, which
could even sound like proof beyond a shadow of a doubt. The essence,
though, is that they want to be very sure before upholding the burdened
party's proof. Alternatively stated, they are much more willing to accept a
false negative than a false positive.
The surprising aspect is that civilians say they apply the same or a very
similar standard in noncriminal cases as they do in criminal cases. And, in
fact, they really do apply the criminal law's high standard of intime
conviction in purely civil cases. The peculiarly French practice of joining
civil claims to parallel criminal cases demonstrates this fact. The high
criminal standard directly and unavoidably determines the outcome in these
French civil cases in which the plaintiff chooses to join with the criminal
proceedings by acting as a so-called partie civile. Thus, a loss on the
criminal case will dictate a loss on the joined civil case. Most interestingly,
French plaintiffs prefer to join their civil cases to parallel criminal
proceedings whenever possible in order to save time, effort, and money and
to gain leverage. In other words, potential civil plaintiffs seem to calculate
that no advantage will be lost by joining with the criminal proceedings,
because they realize "that even'
in a purely civil case there is a presumption
of the defendant's innocence.
I shall not repeat the ready explanation of how the approaches of
Civil and common law diverged on the noncriminal standard of proof
around the time of the French Revolution. My articles' difficulty lay
instead in explaining how these two systems could today share general
aims and yet preserve their historical difference on such a fundamental

3. Kevin M. Clermont & Emily Sherwin, A ComparativePuzle: Standardsof Proof in LAW

AND JUSTICE IN A MULTISTATE WORLD: ESSAYS IN HONOR OF ARTHUR T. VON MEHREN 629 (James
A.R. Nafziger & Symeon C. Symeonides eds., 2002); Kevin M. Clermont & Emily Sherwin, A
Comparative View of Standards of Proof 50 AM. J. COMP. L. 243 (2002) [hereinafter Clermont &
Sherwin, A Comparative View]; Kevin M. Clermont, Standardsof Proofin Japanand the United States,
37 CORNELL INT'L L.J. 263 (2004) [hereinafter Clermont, Japanand the UnitedStates].
4. Clermont & Sherwin, A ComparativeView, supra note 3, at 264 & n.92.

HeinOnline -- 33 Vt. L. Rev. 471 2008-2009

 Vermont Law Review [Vol. 33:469

aspect. After all, this difference on standard of proof-intime conviction

versus preponderance of the evidence-not only entails all sorts of practical
consequences but also implies some basic differences on attitudes toward
the nature of truth.
My suspicion as to the persistent difference between the two standards,
among all the possibilities developed at length in my previous articles, was
that the chosen standard of proof serves a very different function in each of
the legal systems. After deeming the expected-error-cost-minimizing
approach to be procedurally proper and recognizing that noncriminal cases
pose no substantive reason to weight the balance against one side of the
dispute, the common law chose the preponderance-of-the-evidence
standard. The Civil Law must have chosen to sacrifice any sort of error-
minimization in pursuit of a different goal, as to which my best surmise was
perceived legitimacy.
The Civil Law's quest for legitimacy is by no means silly. Public
perceptions of what courts do may be just as important as what courts do in
actuality. Even judicial myths that enhance legitimacy can sometimes have
great utility. Civil Law courts of course are not alone in seeking legitimacy
for their decisions. Common-law courts have had their own means of
maintaining an appearance of fairness and accuracy, such as delegating
fact-finding to an impartial jury of citizens who supposedly embody
common sense and community values.
What benefits of legitimacy does the intime conviction standard
bestow? Although the probabilistic nature of fact-finding is inevitable in
our uncertain world, probabilism does produce discomfort. It also requires
sophistication and imposes some destructive side-effects, at least if
acknowledged by the legal system. The Civil Law system could be
choosing to avoid all those ills by refusing openly to acknowledge
probabilism. If so, civilians would be relying on the widespread efficacy of
the people's unexamined intuition, which would yield the legitimating
misimpression that requiring virtual certainty comports with finding real
"truth." That is, the high standard would insinuate to the parties and the
public that judges will not treat facts as true on less-than-certain evidence.
Civil Law courts and their magisterial judges, sensitive to the special need
for legitimacy in a social structure historically wary of the judiciary, may
have been particularly eager to deal only in "truth."
Moreover, Civil Law judges would prefer not to appear to have free
rein to call close cases or to exercise discretion. That is to say, the high
standard helps them to appear to act only when more-or-less forced to act.
This rhetoric and form thus sound and appear good, and imply a seriousness
of purpose and a caution in action. When the court does render judgment,

HeinOnline -- 33 Vt. L. Rev. 472 2008-2009

20091 Standards of ProofRevisited

the high standard retrospectively implies that the evidence must have been
certain and the result necessary. All these implications are admittedly
appealing, even if false or misleading.
Relatedly, the high standard of proof gives comfort to judges. A high
standard gives the judge a rock to hide behind. The judge will not have to
call evenly balanced cases. The party with the burden of proof will win
only when clearly entitled to win; if that party loses, the standard of proof,
not the judge, is to blame. Likewise, the judge might duck a difficult
decision, or otherwise seek to self-protect, by relying on the
unrealistically high standard.5
Therefore, the current motivation for the difference on standards of
proof, in my view, lies in the subtle differences between the two systems'
objectives to which their standards of proof conform. The Civil Law seeks
the legitimating benefits of the myth that its courts act only on truly true
facts and not on mere probabilities. Courts of the common law seek
legitimacy elsewhere, and thus are free to adopt the standard of proof that
more efficiently, fairly, and openly captures the facts of the case.6
What is important to stress is that the explanatory difficulty in my
earlier articles did not regard which system is right or wrong, but instead
lay in figuring out why the two legal systems remain apart. Whether either
legal system is wrong would not only require a surer determination of the
goals underlying the system's standard of proof, but also turn on a subtle
balance of benefits and costs. The civilians have concluded that the cost-
benefit balance favors a high standard for their noncriminal cases. As I
previously wrote: "Now, I am not arguing that the civil law's valuing of
legitimacy is worse or better than the common law's view, but rather that
it reflects a different weighing of costs and benefits in a different legal
context.",7 I still believe that these very different standards of proof may
each be right for the respective legal system.

5. Similarly, common-law judges seek psychic comfort. The jury again bestows this benefit
by shouldering responsibility for decision. Judges like juries. See Clermont & Sherwin, A Comparative
View, supranote 3, at 273.
6. See id. at 274-75.
7. Clermont, Japanand the United States, supranote 3, at 283; see also Clermont & Sherwin,
A ComparativeView, supra note 3, at 274-75.
The civil law seeks the legitimating benefits of the myth that their courts act only
on true facts and not on mere probabilities. Common-law courts seek legitimacy
elsewhere... and thus are flee to adopt the standard of proof that more fairly and
efficiently captures the real truth of the case.

HeinOnline -- 33 Vt. L. Rev. 473 2008-2009

 Vermont Law Review [Vol. 33:469

II. OTHERS' REACTIONS

Such comparative theorizing naturally ruffled some feathers. Some

responses expressed mostly irritation.8 But an interesting responsive
theme is the emerging attempt to stretch psychology's insights on the fact-
finder's processing of proof to the different question of the fact-finder's
handling of the standard of proof.9 From Germany, Professor Christoph

8. In a remarkably aggressive article on the subject of Clermont & Sherwin, A Comparative

View, supra note 3, a European author surprisingly concedes that the common law's preponderance
standard of proof is superior, but nevertheless criticizes our article's development of arguments that
would work in support of that view. Michele Taruffo, Rethinking the Standardsof Proof 51 AM. J.
COMP. L. 659, 666 (2003). He makes only one relevant and possibly accurate point: some Civil Law
countries, such as Spain and Italy but not the more rigorous Germany, are adamantly nonexpress about
their higher civil standard of proof and thereby free up their judges to apply whatever standard the
judges think appropriate in the search for "truth" or "moral certainty." See id at 666-69 (characterizing
intime conviction as "a sort of black box"). He even seems to approve of this practice. Cf id at 670-71
(maintaining fashionably that truth is what judges say it is). We in fact had developed at length his
observation about judicial freedom to lower the standard in practice. But we did not feel comfortable
resting on it, if only because such judicial discretion is so obviously dangerous that we would never have
felt competent to attribute it to the Civil Law. Instead, we discounted his explanation on the grounds that
some real difference in standards between common law and Civil Law does survive, not only in
language but also in practice. Clermont & Sherwin, A ComparativeView, supra note 3, at 261--64.
9. An early step in this direction was the clever article by Ronald J. Allen & Sarah A. Jehl,
Burdens of Persuasionin Civil Cases: Algorithms v. Explanations,2003 MICH. ST. L. REv. 893. The article
builds on the story model of evidence-processing, see infra note 11 and accompanying text, to produce
another theoretical brand of holism, this one called the relative-plausibility theory. See Allen & Jehl, supra,
at 929-43 (summarizing Professor Allen's previous work on the theory). The relative-plausibility theory
posits that the fact-finder constructs the story (or stories) that the plaintiff is spinning and another story (or
stories) that the defendant is spinning. The fact-finder then compares the two stories (or collections of
stories) and gives victory to the plaintiff if the plaintiff's version is more plausible than the defendant's.
This choice between alternative competing narratives is largely an ordinal process rather than a cardinal
one. This theory has the advantage of nicely circumventing the conjunction paradox, which comprises the
oddity that applying the standard of proof element-by-element effectively causes the standard to lower as
the number of disputed independent elements increases.
But this theory-which couples the holistic evidence-processing of the story model with a
new notion of ordinal comparison as the standard of proof-has problems. First, all its work on the
conjunction paradox is performed not by relative plausibility's ordinal comparison, but by the
underlying story model's insistence that the fact-finder ultimately views the claim as a whole rather than
going element-by-element. The story model leaves the conjunction paradox standing as a legal
aberration, but one of small practical importance. See RICHARD H. FIELD, BENJAMIN KAPLAN & KEVIN
M. CLERMONT, MATERIALS FOR A BASIC COURSE IN CIVIL PROCEDURE 1342-45 (9th ed. 2007)
(observing also that the story model conflicts not only with the law's words but also with many of the
law's practices such as nongeneral verdicts and judgments as a matter of law). Second, the most obvious
difficulties of ordinal comparison are that ordinality and cardinality are not really so different as
explanations for a more-likely-than-not standard and that ordinality cannot easily explain standards of
proof higher or lower than preponderance of the evidence. See Friedman, supra note 2, at 2046-47; cf
Clermont, Procedure'sMagical Number, supra note i, at 1119-20, 1122-26 (discussing not only the
standard of clear and convincing evidence, but also procedure's lower standards including slightest,
reasonable, and substantial possibilities). But cf Allen & Jehl, supra, at 930-37 (stressing the difference
between what ordinal and cardinal approaches compare); Ronald J. Allen, The Nature of Juridical

HeinOnline -- 33 Vt. L. Rev. 474 2008-2009

2009] Standards of ProofRevisited

Engel's new article represents a major step forward in the development of

that theme.Y°
Let me outline the argument presented in Engel's article. First, he
accepts the existence of the difference between Civil Law and common law
as to standards of proof. Second, he acknowledges the usual normative
position that different societal goals could call for different standards of
proof. Third, he shifts to a behavioral account of how the fact-finder
actually implements a given standard, which is where things eventually get
new and interesting.
In that behavioral account, Engel initially sets forth the broadly
accepted "story model" of evidence-processing." It holds that the fact-
finder, over the process's course, constructs from the evidence the story
that makes maximal sense. The fact-finder then chooses, among the
available decisions, the one that fits best with the story it has constructed.
Along the way, and by a similarly automatic or unconscious process, the
fact-finder generates a level of confidence in the decision by considering
the degree of coverage (which means the story accounts for all the
evidence), coherence (which means it is internally consistent, plausible
with respect to the fact-finder's world knowledge, and complete without
striking gaps in expected components), and uniqueness (which means the
absence of plausible alternative stories). The clearer the view of the case

Proof 13 CARDOZO L. REV. 373, 413 (1991) (attempting to explain the beyond-a-reasonable-doubt
standard as not being satisfied if the fact-finder "concludes that there is a plausible scenario consistent
with innocence," while admitting that the clear-and-convincing standard is "troublesome" under his
theory because it is unavoidably cardinal); Ronald J. Allen & Brian Leiter, Naturalized Epistemology
and the Law of Evidence, 87 VA. L. REV. 1491, 1528 (2001) (saying that the prosecution must "show
that there is no plausible account of innocence"). Third, the theory of ordinal comparison is only a
reasonable supposition, as the scholarship is nonempirical (and noncomparative). That supposition also
admittedly retains a bit of fogginess. See Allen & Jehl, supra, at 942-43 (acknowledging criticism that
the relative-plausibility theory is not well defined and is subject to multiple interpretations).
Nevertheless, the relative-plausibility theory merits further development and then testing.
10. Christoph Engel, Preponderanceof the Evidence versus Intime Conviction: A Behavioral
Perspective on a Conflict Between American and ContinentalEuropean Law, 33 VT. L. REV. 435 (2009).
11. See generally JEFFREY T. FREDERICK, THE PSYCHOLOGY OF THE AMERICAN JURY 296-99
(1987) (providing a brief overview of the story model of evidence-processing); REID HASTIE, STEVEN
PENROD & NANCY PENNINGTON, INSIDE THE JURY 22-23 (1983) (providing a brief summary of
empirical studies supporting the story model of evidence-processing); Paula L. Hannaford, Valerie P.
Hans, Nicole L. Mott & G. Thomas Munsterman, The Timing of Opinion Formation by Jurors in Civil
Cases: An Empirical Examination, 67 TENN. L. REV. 627, 629-32 (2000) (discussing "three
predominant models of jury decision making"); Jill E. Huntley & Mark Costanzo, Sexual Harassment
Stories: Testing a Story-MediatedModel of JurorDecision-Makingin Civil Litigation,27 LAw & HUM.
BEHAv. 29, 29 (2003) (presenting research that "extends the story model to civil litigation and tests a
story-mediated model against an unmediated model of jury decision-making"); Nancy Pennington &
Reid Hastie, The Story Model for Juror Decision Making, in INSIDE THE JUROR: THE PSYCHOLOGY OF
JUROR DECISION MAKING 192 (Reid Hastie ed., 1993) (detailing the story model and summarizing
empirical studies testing it).

HeinOnline -- 33 Vt. L. Rev. 475 2008-2009

476 Vermont Law Review [Vol. 33:469

that the chosen story delivers to the fact-finder, the more confident the
fact-finder will be.
Professor Engel next turns to the literature on "coherence-based
reasoning."' 12 He does so primarily to stress the dangers in this unconscious
process whereby the fact-finder strives to make sense of the evidence. A
major danger is the so-called coherence shift: the fact-finder, in
constructing its narrative, will tend to overvalue consistent evidence and
devalue conflicting evidence. This cognitive failing will increase the level
of confidence as well as the incidence of errors.
He finally reaches the law on standards of proof. Here is his real
contribution, because earlier psychological models, including the story
model, failed altogether to give attention to the standard-of-proof step in
decision-making. He posits that standards can act to set a "somatic marker,"
which gets the fact-finder's emotional juices flowing in a direction that will
offset the bias of the coherence shift. In particular, if attached to one of the
potential outcomes, the standards can serve as a warning of the
undesirability of a false finding in that direction. A high standard of proof
makes the fact-finder feel more accountable, inclining it to avoid a falsely
positive result. "When instructed to convict only if guilt is beyond a
reasonable doubt, subjects apply a stricter standard of coverage, coherence,
and uniqueness.' 13 By contrast, a low standard of proof such as the
preponderance of the evidence leaves the fact-finder 14
dispassionate,
indifferent as between false negatives and false positives.

12. See generally Dan Simon, A Third View of the Black Box: Cognitive Coherence in Legal
Decision Making, 71 U. CHI. L. REv. 511, 511 (2004) (explaining coherence-based reasoning, whereby
the fact-finder looks for an account with maximal coherence, utilizing a connectionist cognitive
architecture in which "the decision-making process progresses bidirectionally," that is, facts determine
conclusions while conclusions affect facts in return).
13. Engel, supra note 10, at461.
14. Actually, he writes: "By contrast, the preponderance-of-the-evidence instructions can be
interpreted as a tool for exonerating jury members from personal responsibility. Society is happy with
quite a number of materially wrong judgments. Accountability is reduced to avoiding gross errors." Id
at 464. He indeed believes that the preponderance standard "tolerates a substantially higher error rate"
than the criminal standard. Id. at 444 (citing as sole support James Brook, Inevitable Errors: The
Preponderanceof the Evidence Standardin Civil Litigation, 18 TULSA L.J. 79, 85 (1982), a fine early
article whose point was to defend the preponderance standard on the ground that it was error-
minimizing). Because he believes that the standard of proof asks only how confident the fact-finder is,
he naturally but wrongly concludes that a lower standard, and hence lower confidence, means more
errors. Instead, requiring high confidence will greatly increase the number of false negatives, even if that
strategy limits false positives; actually, low confidence, as long as the found fact is more likely than not,
will minimize the expected number of errors. See supra note I. In brief, Professor Engel's view reflects
a basic misunderstanding of the nature of the preponderance standard. See D.H. Kaye, Clarifying the
Burden of Persuasion: What Bayesian Decision Rules Do and Do Not Do, 3 INT'L J. EVIDENCE &
PROOF 1, 27-28 (1999) (establishing that preponderance minimizes expected error costs); Kaye, Equal
ErrorRates, supranote I (refining his own analysis).

HeinOnline -- 33 Vt. L. Rev. 476 2008-2009

2009] Standards of ProofRevisited

To summarize, one new idea is 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.
The other new idea is that the only role of the law on standards of proof is
to intrude beneficially on the unconscious process by using a high standard
to marshal human emotions against cognitive bias. I find this brace of ideas
intriguing. If accurate, I would see either or both as friendly addenda to the
viewpoint expressed in my articles. But I hesitate to accept either the
breadth of Engel's psychological explanation or the narrowness of the role
he leaves to law.

III. PSYCHOLOGY'S MECHANISM

Psychologically speaking, how does a fact-finder apply a standard of

proof to fact X, be it a simple fact or a more complexly formulated
narrative? The short answer is that no one yet knows.
Professor Engel says little about exactly how the fact-finder applies
the standard. Instead, he talks mainly of evidence-processing. He
mentions the well-recognized concept of confidence. He then appears to
equate the standard of proof to confidence: "The psychological correlate
of the standard of proof is confidence."' 5 But he fails to explain how

Therefore, my text here on the signal sent by the preponderance standard tries to restate his
position in sensible terms, if one were properly to view the standard as setting a low probability threshold.
15. Engel, supra note 10, at 458. He further links the standard of proof to confidence by his
discussion of statistical significance. Id. at 443-46. He claims that under the criminal standard there
must be enough evidence to make the fact-finder more than ninety-five percent confident that the null
hypothesis (not guilty) is not true. This view presents a couple of difficulties. First, it is difficult to
express communicatively the preponderance standard in like terms, because it would have to involve
bidirectionally equalized significance levels; he does not undertake this task, but instead jumps to
conclude wrongly that preponderance is very tolerant of error (in fact, it is relatively tolerant only of
the falsely positive type of error). See supra note 14. Second, statistical significance does not
necessarily imply a high probability, nor does the absence of statistical significance necessarily imply
a low probability; thus, unacceptable results will follow unless the criminal standard requires a high
estimate of probability. See SIMON BLACKBURN, THINK: A COMPELLING INTRODUCTION TO
PHILOSOPHY 224-25 (1999) (criticizing scientific research for inferring causation and correlation
from results that statistically could have been a product of chance); cf infra text accompanying note
23 (treating the psychological version of confidence). In brief, Professor Engel's view reflects a basic
misunderstanding of the nature of the criminal standard. See David H. Kaye, Statistical Significance
and the Burden of Persuasion, LAW & CONTEMP. PROBS., Autumn 1983, at 13, 17, 23 & n.47
(appreciating "why one cannot identify a unique level of 'statistical significance' that would
correspond to proof satisfying the burden of persuasion appropriate to a given type of case"); D.H.
Kaye, Do We Need a Calculus of Weight to UnderstandProofBeyond a Reasonable Doubt?, 66 B.U.
L. REV. 657 (1986) (applying his insight specifically). But see JAMES Q. WHITMAN, THE ORIGINS OF
REASONABLE DOUBT: THEOLOGICAL ROOTS OF THE CRIMINAL TRIAL 4 (2008) (arguing that the
reasonable-doubt standard originated "to make conviction easier, by assuring jurors that their souls
were safe if they voted to condemn the accused").

HeinOnline -- 33 Vt. L. Rev. 477 2008-2009

478 Vermont Law Review [Vol. 33:469

confidence could actually work as a standard of proof. He thereby evades

three tough explanatory steps.
First, how does the fact-finder generate a usable confidence level,
where confidence has the psychological meaning of how sure the fact-finder
feels about its output? The task, according to prior research, seems to
require considering the coverage, coherence, and uniqueness of the factual
account. Formulating a confidence level with sufficient clarity to apply the
standard of proof would be no easy mental task. Presumably, humans would
be performing this complex task on the unconscious level, and so it defies
precise explanation. His lack of precision here is therefore understandable.
Second, why should we rely on confidence, which is admittedly a
very shaky measure? Much empirical research shows that confidence has
a weak linkage to accuracy, for reasons including that the evidence-
providers can easily manipulate the level of confidence as by the order of
presentation, that the fact-finder's generation of confidence falls subject to
all sorts of biases such as the coherence shift's confirmation bias, and that16
confidence depends on many external factors including self-image.
Here, I suppose, his argument is that humans can do no better with any
other alternative measure.
Third, and most important, how does the felt level of confidence
translate into whether or not the evidence meets the standard of proof? His
suggestion is that the standard of proof demands a specific level of
confidence from the fact-finder in its view on fact X. Thus, a high criminal
standard demands great confidence, whereas a preponderance standard
would settle for a shaky conclusion. The criminal fact-finder and the
noncriminal fact-finder would both be looking at whether they somehow
"believe" the fact is true, but might have to be much more confident in that
belief if the case were a criminal one. This attempt at brief description
shows that confidence is no self-explanatory tool for handling ambiguous

Note that I am not contending that the criminal standard could not contain a confidence
measure in addition to, rather than in lieu of, a required probability. Maybe the presence of that extra
something would explain why Professor Engel intuits that the criminal standard has a subjective element
that is lacking in the preponderance standard. See Engel, supra note 10, at 436, 441-42, 461-63 (saying
that the high standard looks for a personal conviction). Further, the culturally fraught criminal standard
of proof and its wordy formulation could convey, on an emotional level, the message to the fact-finder's
unconscious that there is this need for personalconviction based on confidence in guilt, plus a very high
probability of guilt. See id. at 460-62. But see JAMES Q. WHITMAN, THE ORIGINS OF REASONABLE
DOUBT: THEOLOGICAL ROOTS OF THE CRIMINAL TRIAL 4 (2008) (arguing that the reasonable-doubt
standard originated "to make conviction easier,by assuring jurors that their souls were safe if they voted
to condemn the accused").
16. See generally DAVID DUNNING, SELF-INSIGHT: ROADBLOCKS AND DETOURS ON THE PATH
TO KNOWING THYSELF 6-9, 37-61 (2005) (describing the tendency for people to be overconfident and
positing a theory of how we develop self-judgment).

HeinOnline -- 33 Vt. L. Rev. 478 2008-2009

2009] Standards ofProofRevisited

evidence and also that confidence would be an especially clumsy tool for
administering a preponderance standard in a legal system choosing to
deploy it in pursuit of error-minimization. 7

A. Confidence's PrescriptiveProblems

Yet, this confidence test is not how the standard of proof should work.
A focus on probability is the mathematical route to minimizing the expected
cost of erroneous decisions. The standard of proof therefore asks how likely
the fact is, but the confidence measure answers with how sure the fact-
finder is in its belief. In brief, confidence is a kind of second-order
probability that is irrelevant to the first-order
8
probability with which
standards of proof should be concerned.'
Confidence can at most serve as an imperfect surrogate for probability,
or likelihood. 19 It may be that as probability goes up, confidence tends to go
up too. 20 But correlation does not mean that the two measures are

17. See supra note 15 (explaining the difficulty of expressing preponderance in terms of
confidence). This difficulty likely explains why Professor Engel switches, at one point, to a different
explanation of how confidence would translate into various standards of proof:
The easiest way to translate these techniques of calibrating judgment into
mental mechanisms is by manipulating the threshold. ... [D]epending on the
standard of proof, the mind asks for a larger minimum difference in the activation
of the preferred alternative, compared to all competing alternatives. This
explanation fits best for the minimum gap in plausibility between competing
stories. ... For the remaining elements of the story model, one needs a richer
decision criterion.
Engel, supra note 10, at 461-63; cf supra note 9 (considering Professor Allen's similar explanation,
and observing that comparison of competing stories does not work well for standards higher or lower
than preponderance).
18. Various previous efforts to bridge the gap between first- and second-order probabilities for
this purpose have failed. Compare, e.g., Neil B. Cohen, Conceptualizing Proof and Calculating
Probabilities:A Response to ProfessorKaye, 73 CORNELL L. REv. 78, 91-93 (1987) (conceding that
confidence relates solely to the probability of avoiding false positives), and Neil B. Cohen, Confidence
in Probability:Burdens of Persuasionin a World of Imperfect Knowledge, 60 N.Y.U. L. REv. 385, 386
(1985) [hereinafter Cohen, Confidence in Probability] (advancing a new theory of standards of proof
based on the statistical concept of confidence intervals), and Neil B. Cohen, Commentary, The Costs of
Acceptability: Blue Buses, Agent Orange, and Aversion to Statistical Evidence, 66 B.U. L. REV. 563,
569 (1986) (qualifying his own argument), with D.H. Kaye, Apples and Oranges: Confidence
Coefficients and the Burden of Persuasion,73 CORNELL L. REv. 54, 54, 56-58 (1987) (responding to
Cohen, Confidence in Probability,supra).
19. In statistical terminology, "likelihood" (the chance that the data would be observed, given a
hypothesis as true) is not wholly equivalent to "probability" (the chance that a hypothesis is true, given
the observed data). E.g., RICHARD M. ROYALL, STATISTICAL EVIDENCE: A LIKELIHOOD PARADIGM 5-6,
28 (1997). But for most people, likelihood means probability. I use it here in that way, with perhaps the
connotation of an intuitive measure of probability and with the benefit of conforming to the common
legal, and probabilistic, usage of "more likely than not."
20. "Subjects are most confident in their decision if they believe the case to be clear, i.e., if

HeinOnline -- 33 Vt. L. Rev. 479 2008-2009

 Vermont Law Review [Vol. 33:469

equivalent. 2 1 Confidence does not necessarily imply a high probability, nor

does the absence of confidence necessarily imply a low probability. To
focus on confidence is either22
to focus on the wrong thing or to conflate
confidence and probability.
A telling counterexample to the equivalence of confidence and
probability comes from a hypothetical that I posed on a recent
examination. The disputed fact was which of two dead people had been
driving a very negligently crashed car and which was the passenger; the
evidence was that the car's owner was driving on departure for the long
trip and normally did not let others drive, but the accident scene provided
no proof on the driver's identity, and no other proof was available. 23 In
that hypothetical, the fact-finder could be highly confident that on all
available evidence the owner was slightly more likely than not the driver.
If the standard of proof turns on estimated likelihood, the fact-finder could
find liability under a preponderance standard but could not find it beyond
a reasonable doubt. Given the fact-finder's sky-high confidence, however,
an approach that ties the standard of proof exclusively to confidence
would, albeit totally unacceptably, conclude the car owner was the driver
under the criminal standard!
My point is that a fact-finder could be very confident in its
determination that the fact was just slightly more likely than not, so that
confidence and probability are not equivalent. But a confidence theorist,
converting the vague definition of confidence into a virtue, might respond
that the fact-finder in my hypothetical should not be very confident because
of the paucity of evidence and the availability of a plausible alternative
story: the theorist would prefer to say that we want the fact-finder to be
confident in the "truth" rather than confident that the evidence shows a fact
to be more likely than not. I would rejoin with an altered hypothetical in
which there is a mountain of evidence all pointing to the car owner being
slightly more likely the driver, including scientific evidence establishing a
point estimate of sixty percent. More generally, I would rejoin that the

they acquit and subjective probability of guilt is low, or if they convict and subjective probability of
guilt is high." Engel, supra note 10, at 461. See also id.at 453.
21. Professor Engel equates his confidence-based standard of proof to statistical significance.
"The legal order is in the same position as a statistician." Id. at 444. But again, although statistical
significance correlates with probability, these two are separate measures that can diverge. See supra
note 15.
22. See Kaye, supra note 18, at 58 (exposing neatly, by formal and general proof, the persistent
misunderstanding "of the relationship between 'significance' and 'confidence,' on the one hand, and the
posterior probability and the burden of persuasion, on the other").
23. For the full hypothetical, see FREDRIC R. MERRILL & DOMINICK R. VETRI, PROBLEMS AND
MATERIALS ON FEDERAL COURTS AND CIVIL PROCEDURE 249-53 (1974).

HeinOnline -- 33 Vt. L. Rev. 480 2008-2009

2009] Standards of ProofRevisited

common-law fact-finder is not supposed to hold an unavoidable paucity of

evidence against the burdened party, but is instead in such a situation
supposed to decide likelihood based on the evidence.24
Epistemology provides another route to expose the difference between
confidence and probability. The epistemological aim of evidence law is that
the fact-finder should construct a belief that corresponds to the outside
world's truth.25 Probability is a measure of the chance of that
correspondence existing between finding and reality. An alternative, but
less desirable, approach would be for the fact-finder to seek an evidential
construct (or agree to a construct) that coheres with the rest of the world
image inside the fact-finder's head. Confidence is a measure of the degree
of that coherence between the new belief and all the old beliefs.
Psychologists might approve of this strategy of looking to confidence, so
celebrating the Civil Law's "subjective" approach over the common law's
"objectivism. ' '26 Such a strategy is understandable: people prefer looking to
an inside feeling rather than to any outside measure-and so might
operationalize an inquiry about probability of a fact by changing it into an
inquiry about confidence that the fact obtains-because of their assumption
that one has better cognitive access to the state of one's own mind. But

24. In other situations, fact-finders can properly use common experience of the world or
sometimes even specific knowledge of the case. See, e.g., J. Alexander Tanford, An Introductionto Trial
Law, 51 Mo. L. REV. 623, 700 (1986) ("Jurors are encouraged to use common experience, but are
prohibited from becoming a kind of 'expert witness'...."); see also THOMAS CATHCART & DANIEL
KLEIN, PLATO AND A PLATYPUS WALK INTO A BAR... UNDERSTANDING PHILOSOPHY THROUGH JOKES
53-54 (2007). Take, for example, the following story:
A defendant was on trial for murder. There was strong evidence indicating
his guilt, but there was no corpse. In his closing statement, the defense attorney
resorted to a trick. "Ladies and gentlemen of the jury," he said. "I have a surprise
for you all-within one minute, the person presumed dead will walk into this
courtroom."
He looked toward the courtroom door. The jurors, stunned, all looked
eagerly. A minute passed. Nothing happened. Finally the lawyer said, "Actually, I
made up the business about the dead man walking in. But you all looked at the
door with anticipation. I therefore put it to you that there is reasonable doubt in
this case as to whether anyone was killed, and I must insist that you return a
verdict of 'not guilty."'
The jury retired to deliberate. A few minutes later, they returned and
pronounced a verdict of "guilty."
"But how could you do that?" bellowed the lawyer. "You must have had
some doubt. I saw all of you stare at the door."
The jury foreman replied, "Oh, we looked, but your client didn't."
Id.
25. See Allen & Leiter, supra note 9, at 1497-503 (explaining naturalized epistemology in the
context of the law of evidence).
26. See Engel, supra note 10, at 436-37, 441-42, 461-63; cf supra note 15 (attempting to put
the distinction between subjective and objective to a more productive use).

HeinOnline -- 33 Vt. L. Rev. 481 2008-2009

 Vermont Law Review [Vol. 33:469

philosophers instruct us that the inside world does not provide steadier
ground than the outside world. 27 And more to my point, the law has no
reason to interest itself in any such internal feeling, which would be
irrelevant to societal needs. Instead, society wants the law to pursue an
optimal outcome in terms of reality. Accordingly, the law by its standard of
proof seeks to force the fact-finder, in the final decisional step, to link its
inside mental state to the outside real world.

B. Confidence's DescriptiveDifficulties

Regardless of the desirability of looking to probability, it could be

that many fact-finders in actuality look to confidence when called upon to
apply a standard of proof. But given the overall fogginess of the supposed
interaction between felt confidence and standards of proof, it was
foreordained that Professor Engel would provide no empirical showing
that fact-finders misinterpret standards of proof so as to impose only a
confidence test. His Erfurt Experiment measures subjects' confidence
levels and subjective probabilities, but fails to show which drives
decision.2 8 All he shows 29
is "that different standards of proof are
psychologically feasible.,
Other researchers have not advanced the ball much farther. Despite
considerable evidence that judges meet difficulty in conveying any standard
of proof to the jury, and that juries have difficulty in quantifying the
standards of proof,3 ° there is some evidence that is more hopeful. Finders of
a fact can focus separately on the question of how probable it is that the fact
is true.31 No need exists for them to perform a quantification step in that

27. See TIMOTHY WILLIAMSON, KNOWLEDGE AND ITS LIMITS 94-98, 106-09, 190-93 (2000)
(developing an anti-luminosity argument).
28. See Engel, supra note 10, at 460-61. The experiment receives fuller description in Andreas
GlOckner & Christoph Engel, Can We Trust Intuitive Jurors?An Experimental Analysis (Max Planck
Inst. for Research on Collective Goods, Preprint No. 2008/36, 2008), available at
http://www.coll.mpg.de/pdf_dat/2008 36online.pdf.
29. Engel, supra note 10, at 463.
30. See Clermont, Procedure'sMagical Number, supra note 1, at 1141 & n. 110, 1146-47 &
nn.126-27, 1149 & n.135; see also Engel, supra note 10, at 458-61 (highlighting that the beyond-a-
reasonable-doubt standard translates into widely varying probabilities for the lay juror); Dorothy K.
Kagehiro, Defining the Standardof Proofin Jury Instructions, I PSYCHOL. SC. 194 (1990) (reviewing
research on juror comprehension that found if the standard of proof was stated in terms of probability,
then it had its anticipated effect; but when the standard was not so stated, it did not have its anticipated
effect); Elisabeth Stoffelmayr & Shari Seidman Diamond, The Conflict Between Precision and
Flexibility in Explaining "Beyond a Reasonable Doubt," 6 PSYCHOL. PUB. POL'Y & L. 769, 773-77
(2000) (discussing experiments in which mock jurors failed to distinguish between standards ofproof).
31. See Clermont, Procedure's Magical Number, supra note 1, at 1143-48 (arguing that
decision-makers should continue to use the "seven categories of uncertainty ... : (1) slightest

HeinOnline -- 33 Vt. L. Rev. 482 2008-2009

2009] Standards of ProofRevisited

process.32 They are quite capable of making gross distinctions in

probability, such as by distinguishing in application between the
preponderance standard and the criminal standard upon proper
instructions. Indeed, the main lesson from the Erfurt Experiment is to
exactly this effect, by showing a statistically significant difference between
the two standards of proof: the subjects rendered conviction under a
beyond-a-reasonable-doubt instruction only on a very high subjective
probability, but under a preponderance standard they very neatly decided
for the plaintiff if their subjective probability
34
exceeded fifty percent and for
the defendant at or below fifty percent.
Given that a probability-based standard of proof could conceivably
exist, how might it actually operate? The processing of evidence itself could
have yielded a rough measure of probability, if it were in part Bayesian in
nature, or perhaps yielded merely a level of confidence, if not at all
Bayesian. I readily admit that it is rather implausible that fact-finders apply
any rigorous probabilistic technique in complex processing of the evidence,
and that they may indeed be wholly nonprobabilistic. However, there is
nothing inconsistent or infeasible about even the nonprobabilistic fact-
finder subsequently fulfilling, at least in some intuitive way, the
straightforward probabilistic task of (1) estimating subjective probability,
by looking at the evidence to measure roughly the probability of fact X's
existence as against the probability of not-X,35 and (2) comparing it to a set

possibility, (2) reasonable possibility, (3) substantial possibility, (4) equipoise, (5) probability, (6) high
probability, and (7) almost certainty"); cf Engel, supra note 10, at 437-38, 460-61 (describing an
experiment showing that subjects could measure subjective probability).
32. See Clermont, Procedure's Magical Number, supra note 1,at 1146 & n.126, 1149 &
n.136 (arguing that "quantification ordinarily is not a necessary or desirable step in applying the
standard of proof").
33. See, e.g., SAUL M. KASSIN & LAWRENCE S. WRIGHTSMAN, THE AMERICAN JURY ON
TRIAL: PSYCHOLOGICAL PERSPECTIVES 156 (1988) (reporting research that "suggests that, whether by
instruction, intuition, or simply an appreciation for the differential consequences of criminal and civil
decisions, juries are already sensitive to variations in the standard of proof"); Martin F. Kaplan,
Cognitive Processes in the Individual Juror, in THE PSYCHOLOGY OF THE COURTROOM 197, 216
(Norbert L. Kerr & Robert M. Bray eds., 1982) ("It is also possible to observe the effects of differing
judicial instructions on the [decisional] criteria."); Norbert L. Kerr et al., Guilt Beyond a Reasonable
Doubt: Effects of Concept Definition and Assigned Decision Rule on the Judgments of Mock Jurors,
34 J. PERSONALITY & SOC. PSYCHOL. 282, 291-92 (1976) (describing the results of a study
examining reasonable-doubt criteria using mock jurors); Daniel B. Wright & Melanie Hall, How a
"Reasonable Doubt" InstructionAffects Decisions of Guilt, 29 BASIC & APPLIED SOC. PSYCHOL. 91
(2007) (describing two experiments investigating the effects of different reasonable-doubt
instructions on mock jurors).
34. See supra note 28.
35. See Friedman, supra note 2, at 2044-46 (discussing a theory "[c]losely associated with
Bayesian probability theory," called Bayesian decision theory, which posits that decision-making is
"based on the expected value of any course of action"). But see Allen & Jehl, supranote 9, at 930-36,

HeinOnline -- 33 Vt. L. Rev. 483 2008-2009

 Vermont Law Review [Vol. 33:469

scale of coarse gradations, such as asking whether the fact is more likely
than not or whether it is almost certain. 36 That is to say, a nonprobabilistic
model of the evidence-processing step, such as the story model, could be
consistent with this more probabilistic approach to the standard-of-proof
step. And this more probabilistic approach is entirely feasible. After all,
we perform such a task constantly in real life. We all competently apply
standards of proof similar to the law's. I might decide to go see a
particular movie, rather than stay home, only if going to the movie is more
likely than not to deliver more pleasure. Countless decisions rest on cost
minimization, where one decides on a particular route over all available
alternatives because it is likely the best way to go, or the way one will be
happiest. Other decisions may entail a higher standard, because one
weighs errors in one direction more heavily than in the other direction. I
might decide to marry a particular person only if I am really, really sure
that this is the route to go.
I further admit that it is easier to imagine a law-trained judge
undertaking a deliberate and probabilistic approach to the standard of proof.
But I can also picture jurors performing this task. First, the judge delivers
careful instructions that require the jurors to approach the standard of proof
by a separate step and in probabilistic terms. Second, the task is really not
so demanding, as it requires only an intuitive comparison of rough
probability to a gradated scale. Third, jury deliberations could help
immensely in performing the task. Group discussion could reinforce the
duty to relate one's inner conviction to likelihood in the outside world and
help to explain how to perform the task. We indeed know that jurors often
do not reach a decision until the deliberations phase,37 and that jurors can
use the standard of proof as an argument during deliberations.38
Consequently, rather than assume that fact-finders in applying the
standard of proof invoke some unspecified and unproven confidence-based

(bringing out the logical complexities in Friedman's simple formulation ofX and not-X).
36. See generally Clermont, Procedure's Magical Number, supra note I (exploring the
psychological tendency to handle probability by coarse categories).
37. See Shari Seidman Diamond, Beyond Fantasyand Nightmare: A Portraitof the Jury, 54
BUFF. L. REv. 717, 758-61 (2006) (discussing the effect ofjury deliberations on juror decision-making);
Hannaford et al., supra note 11, at 640 fig.3 (a graph detailing the percentage of jurors that made up
their mind at different stages of a trial; compared to other stages, more jurors formed a decision in the
deliberations phase); David H. Kaye, Valerie P. Hans, B. Michael Dann, Erin Farley & Stephanie
Albertson, Statistics in the JuryBox: How Jurors Respond to MitochondrialDNA Match Probabilities,4
J. EMPIRICAL LEGAL STUD. 797, 815-18 (2007) (explaining the effect of jury deliberations in a mock
criminal trial for robbery in which expert testimony on mitochondrial DNA sequencing was presented).
38. See Robert J. MacCoun & Norbert L. Kerr, Asymmetric Influence in Mock Jury
Deliberation: Jurors' Bias for Leniency, 54 J. PERSONALITY & SOC. PSYCHOL. 21, 30 (1988)
(suggesting that "deliberation tends to amplify the effects of these instructions").

HeinOnline -- 33 Vt. L. Rev. 484 2008-2009

2009] Standards ofProofRevisited

approach, which would be both inaccurate and undesirable to apply, I shall

continue to assume that the fact-finder does, even if in some rough and
intuitive manner, what the law with good reason actually tells it to do: settle
on a sense of subjective probability and compare it to the standard of
proofs coarse gradation.

IV. LAW'S ROLE

Whether one accepts as the manner of implementing the standard of

proof either (1) a confidence-based process in the unconscious or (2) the
traditional view of a more deliberate (even if largely intuitive) comparison
of rough probability to a gradated scale, there remains the question of the
law's contribution to the mental process. On this question, a psychologist
sees the law's formulation of a high standard of proof as a way to set a
somatic marker that on an emotional level conveys to the fact-finder's
unconscious the message to protect the defendant; the high standard drives
home the duty to take fact-finding seriously and to demand a considerable
feeling of confidence before finding for the burdened party. I tend to see the
expression of a higher standard as a way to inform the fact-finder that the
burdened party must provide a stronger showing in terms of probability.
Importantly, we agree that no matter how the law's standards actually work,
they seem to succeed in their mission of tilting the playing field in the
normatively desired direction.
The common law is obvious in trying to accomplish more, however, by
adjusting the standard of proof to the circumstances of the case. This
observation brings me back to why both the common law and the Civil Law
apply a high standard to criminal cases, but only common-law jurisdictions
lower the standard for noncriminal cases. Oddly, although Professor Engel
set out to explain this very difference between the common law's and the
Civil Law's standards of proof, his psychological detour has no explanatory
power whatsoever on the divide. His bottom line reduces to this: "The heart
of the conflict between American and Continental law is therefore the
different assessments of cost and benefit [of unconscious decision-making].
Whereas Continental law is attracted by the benefit, American law is scared
of the cost."' 39 Quite simply, Engel's explanation has to be inadequate

39. Engel, supra note 10, at 438. His logic appears to be that humans do confidence levels, not
probability; that the only thing any standard of proof can do is impose a high-confidence requirement;
and that because low confidence means more errors, the law should always require high confidence and
never tolerate the low confidence permitted by a preponderance standard: therefore, the Civil Law is
generally right in its intime conviction approach, and the common law inexplicably ceases to care about
errors in its noncriminal cases. See id. at 464-66. Even assuming his psychology to be sound, his logic
leads astray because of its faulty premises regarding the incidence of error under legal standards: he does

HeinOnline -- 33 Vt. L. Rev. 485 2008-2009

 Vermont Law Review [Vol. 33:469

because it draws no distinction between noncriminal and criminal cases.

Whether both systems proceed by confidence or by probability, or whether
the Civil Law goes on the basis of confidence and the common law on
probability, the question remains why the common law alone sometimes
lowers its standard of proof. That is to say, for an adequate explanation, he
would have further to explain why the Civil Law's standard treats all cases
alike, but the common law has decided to treat some cases one way, other
cases another way, and still other cases a middle way. He does not and
could not explain this because his confidence-directed concerns are
orthogonal to the question at hand. Most likely, no psychology-driven
theorizing can explain why the common law chooses to treat different
cases differently because the explanation lies in the goals of the
procedural system.
The common law in its standards of proof consistently tries to
minimize expected error costs. Because costs of error obviously differ in
different settings, the common law is forced to apply multiple standards of
proof: preponderance of the evidence, clear and convincing proof, and proof
beyond a reasonable doubt. The elevated standards help to avoid the high
social costs of false positives, as in convicting the innocent, and the
preponderance standard minimizes expected social costs in any setting
where false positives are no more costly than false negatives.
The Civil Law, by contrast, applies basically the same standard of
proof in all settings. In all cases, in the face of differing costs, the Civil Law
sends to its fact-finder the strong signal to protect the nonburdened party. It
follows that the Civil Law is not trying solely to minimize expected error
costs. The Civil Law must have another purpose. My best surmise remains
that it is seeking the perceived legitimacy of the higher standard. But it
pursues that goal at the expense of the burdened party, whoever that party is
after the law's burdens and presumptions play out in a particular case.
In sum, the law does more by its standard of proof than signal the fact-
finders. It also expresses a choice among the many possible goals of
procedure. Given their different purposes, the common law and the Civil
Law have chosen different but appropriate standards of proof. And, as
already suggested, the law of each of the two systems has at least partially
succeeded in expressing that standard to the fact-finders. Thus, the law
appears to be relatively effective in this domain.

not realize that a lower standard, whether based on confidence or on probability, is in terms of error-
minimization a more efficient and fair standard for noncriminal cases. See supra notes 14 & 15. Also
perplexing is that his logic would seem to paint the Civil Law, in its avoidance of its fact-finders'
relying on best hunches in noncriminal cases, as being more rather than less wary of unconscious
decision-making than the common law.

HeinOnline -- 33 Vt. L. Rev. 486 2008-2009

2009] Standards of ProofRevisited 487

CONCLUSION

The psychological mechanism for implementing standards of proof

remains to be discovered. For the time being, science gives law no reason to
abandon its traditional hope that its intuitive fact-finder roughly estimates
the subjective probability of the burdened party's version of fact, and then
compares the probability to a set scale of coarse gradations of probability,
such as asking whether the version appears more likely than not or whether
the version appears almost certain.
Meanwhile, the Civil Law and the common law apply different
standards of proof in noncriminal cases, namely, the high intime conviction
and the low preponderance standard. Again, psychology provides no
magical explanation of this divide. Regardless of how the mind works in
applying the standard of proof, the explanation for the difference in the two
systems' laws likely lies in their pursuing different goals when they
formulated their standards of proof.

HeinOnline -- 33 Vt. L. Rev. 487 2008-2009