Rebooting the new evidence scholarship
| Published date | 01 October 2020 |
| DOI | 10.1177/1365712720943329 |
| Date | 01 October 2020 |
| Subject Matter | Articles |
Article
Rebooting the new evidence
scholarship
John R Welch
Saint Louis University-Madrid Campus
Abstract
The new evidence scholarship addresses three distinct approaches: legal probabilism, Bayesian
decision theory and relative plausibility theory. Each has major insights to offer, but none seems
satisfactory as it stands. This paper proposes that relative plausibility theory be modified in two
substantial ways. The first is by defining its key concept of plausibility, hitherto treated as pri-
mitive, by generalising the standard axioms of probability. The second is by complementing the
descriptive component of the theory with a normative decision theory adapted to legal process.
Because this version of decision theory is based on plausibilities rather than probabilities, it
generates plausibilistic expectations as outputs. Because these outputs are comparable, they
function as relative plausibilities. Hence the resulting framework is an extension of relative
plausibility theory, but it retains deep ties to legal probabilism, through the proposed definition of
plausibility, and to Bayesian decision theory, through the normative use of decision theory.
Keywords
Bayesian decision theory, legal probabilism, plausibility, probability, relative plausibility
1. Introduction
The new evidence scholarship is no longer new, but it may be due for renewal. Professor Richard Lempert
(1986: 440) characterised the movement as a ‘third wave’ of evidence scholars. The first wave included
systematisers like Wigmore, Maguire, McCormick and Morgan; the second wave began with the enact-
ment of the Federal Rules of Evidence by the American Congress in 1975; and the third wave seeks ‘to
build on or criticise mathematical models as modes of proof or as a means of understanding trial pro-
cesses’.
1
Hence the new evidence scholarship was developed by two disparate groups: a pro group,
disposed to build on mathematical models; and a con group, critical of the juridical use of these models.
Corresponding author:
John R Welch, Saint Louis University, Avenida del Valle, 34, 28003 Madrid, Spain.
E-mail: jwelch7@slu.edu
1. Lempert’s description of the new evidence scholarship was part of a presentation to the Symposium on Probability and
Inference in the Law of Evidence held at Boston University in April 1986, where he noted that evidence scholarship is changing
its focus from the articulation of rules to the process of proof. Symposium papers were published in Boston University Law
Review 66 (1986: 377–952) and in Tillers and Green (1988).
The International Journalof
Evidence & Proof
2020, Vol. 24(4) 351–373
ªThe Author(s) 2020
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1365712720943329
journals.sagepub.com/home/epj
The pro group includes both legal probabilists and Bayesian decision theorists. Legal probabilists
recommend reliance on probabilities in juridical decision making. They propose that probabilities be
employed to clarify standards of proof, for example. The standard of proof in most civil cases, ‘the balance
of probabilities’ in the United Kingdom or ‘the preponderance of the evidence’ in the United States, can be
defined as a probability > .5 (Kaye, 1982); the standard in exceptionally serious civil cases, ‘clear and
convincing evidence’, might be interpreted as a probability around .75 (Schauer and Zeckhauser, 1996:
34); and the standard in criminal cases, ‘beyond a reasonable doubt’, could be glossed as a probability of
.90 or .95 (Laudan, 2006: 56).
2
Probabilities could also be used to determine the probative value of
evidence (Finkelstein and Fairley, 1970; Friedman, 1986; Kaye, 1986). Like probabilists, Bayesian
decision theorists emphasise probabilities, which register beliefs, but they also insist on utilities, which
express desires. They recommend combining both kinds of input in order to ensure that juridical decisions
maximise expected utility. However, some decision theorists modify this basic idea by aiming at minimal
expected regret (Kaplan, 1968), maximal expected desirability (Cullison, 1969) or minimal expected loss
(Kaye, 1988, 1999).
Opposed to the pro attitudes of both legal probabilists and Bayesian decision theorists are scholars who
criticise mathematical models of evidence. Prominent among them are advocates of relative plausibility
theory (Allen, 1986, 1991, 1994).
3
These scholars insist that legal fact finders do not reason as legal
probabilists and Bayesian decision theorists say they should; rather, fact finders first consider plaintiff’s
and defendant’s explanations of disputed events and then compare these explanations in light of the
evidence, their background knowledge and the relevant standard of proof (Allen and Pardo, 2019: 17).
The comparison results in an inference to the best explanation (Pardo and Allen, 2008).
4
Relative plau-
sibility theory is consistent with empirical work in cognitive psychology, notably the story model of juror
decision making (Pennington and Hastie, 1986, 1991), which claims that jurors impose a narrative
structure on information received at trial and base their decisions on these narratives. However, relative
plausibility theorists stress that their theory is distinct from the story model, which ‘does not provide an
explanation of standards of proof and other features of the proof process’ (Allen and Pardo, 2019: 17, n.
86). In addition, some explanations may not count as stories at all; statistical explanations, for example, are
not stories in the usual sense of the term.
To sum up: a case of assault, for example, might be decided from three apparently divergent points of
view. A legal probabilist would find the posterior probability of the plaintiff’s account and determine
whether it meets the relevant standard of proof. In light of this standard, a Bayesian decision theorist would
weigh the expected utility of finding for the plaintiff against the expected utility of finding for the
defendant. A relative plausibilist would compare the plausibility of the plaintiff’s account and the plau-
sibility of the defendant’s account and find for the plaintiff only if her account meets the relevant standard
of proof.
Despite sharp divergences among the currents of new evidence scholarship, muted calls for reconcilia-
tion occasionally surface. Professor Friedman, who is a Bayesian decision theorist, claims that relative
plausibility theory and Bayesian decision theory are ‘perfectly consistent’ (Friedman, 2001: 2045); and
Professors Allen and Pardo, who are advocates of relative plausibility theory, note ‘there is no reason to see
explanatory [relative-plausibilistic] and Bayesian [decision-theoretic] approaches as necessarily incom-
patible’ (Pardo and Allen, 2008: 253; cf. Allen, 1994: 607). More recently, Professors Biedermann and
2. Even trained judges differ in their numerical interpretations of the criminal standard, though they generally vary between .75
and .90 (Posner, 1999: 1506).
3. Becauserelative plausibility theory has roots in the tradition of naturalised epistemology (Allen and Leiter, 2001), it would be
misleading to consider it simply as a critical response to the pro currents of new evidence scholarship. But since the focus of this
paper is new evidence scholarship, its treatment of relative plausibility theory will be circumscribed accordingly.
4. ProfessorAllen (2008: 327) admits that his sense of ‘best explanation’ may not be what philosophers mean by the term, but ‘the
central point is the explanation-based nature of juridical proof’. Ribeiro (2018) proposes a different approach to inference to the
best explanation.
352 The International Journal of Evidence & Proof 24(4)
To continue reading
Request your trial