Article
Common problems of
plausibility and probabilism
Maggie Wittlin
University of Nebraska College of Law, Nebraska, USA
Abstract
In this response to Allen and Pardo’s Relative Plausibility and Its Critics, I argue that while relative
plausibility presents certain advantages over probabilism, it also fails to avoid several problems
that the authors attribute to probabilism. I note that relative plausibility can be understood as
probabilism under certain constraints that characterise a typical trial. I then argue that two of
Allen and Pardo’s central problems with probabilism—the absence of an objective means for
measuring the strength of evidence and the conjunction problem—apply to both probabilism
and relative plausibility, although neither problem poses a serious threat to accuracy. I con-
clude that each theory, despite these problems, is useful for certain purposes—relative
plausibility better models how advocates present cases and how jurors process information;
probabilism serves as a valuable tool for modelling relevance and prejudice.
Keywords
evidence, probability, relative plausibility, Bayes, juries
In their characteristically rigorous and insightful article, Professors Allen and Pardo argue that relative
plausibility has supplanted probabilism as the best explanation of the proof process. The type of
probabilism they target with this argument is narrow: a strictly numeric, non-comparative, atomistic
version that is used to describe and explain our system of proof (see Allen and Pardo, 2019: 10–14).
Perhaps the earliest advocates of probabilism adhered to a rigid, numeric model or suggested that jurors
are skilled Bayesian updaters (see e.g. Finkelstein and Fairley, 1970; Kaplan, 1968: 1083–1091). But
more recent defences—from the past 40 years or so—have embraced broader versions of Bayesian
probability. In these incarnations, not only is the Bayesian model typically a normative aspiration or
heuristic device, rather than a description of how factfinding works in practice (see, e.g. Lempert, 1977:
1056–1057, 1986: 446–447), it also allows for flexibility in the updating process: Bayesian models may
be either atomistic or holistic (see Friedman, 1992: 86; Kaplan, 1968: 1085), they are typically com-
parative (see Friedman, 1992: 86; Lempert, 1977: 1023), they do not necessarily specify how likelihood
ratios are determined, and they may encompass non-numeric updating (see Friedman, 1992: 86).
Corresponding author:
Maggie Wittlin, Assistant Professor of Law, University of Nebraska College of Law, Lincoln, NE 68503, USA.
E-mail: maggie.wittlin@unl.edu
The International Journalof
Evidence & Proof
2019, Vol. 23(1-2) 184–190
ªThe Author(s) 2018
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/1365712718815349
journals.sagepub.com/home/epj
