Statistical Evidence and Individual Litigants: A Reconsideration of Wasserman's Argument from Autonomy
| Author | Amit Pundik |
| Published date | 01 November 2008 |
| Date | 01 November 2008 |
| DOI | http://doi.org/10.1350/ijep.2008.12.4.304 |
| Subject Matter | Article |
STATISTICAL EVIDENCE AND INDIVIDUAL LITIGANTS
Statistical evidence and
individual litigants: a
reconsideration of
Wasserman’s argument
from autonomy
By Amit Pundik*
Hughes Hall, Cambridge
Abstract The use of statistical evidence in court has attracted long-running
controversy. Some uses of statistical evidence seem intuitively wrong, both in
real cases (for example, Sally Clark) and in hypothetical examples (for example,
the Gatecrasher Paradox). Yet, explaining why has proven to be difficult. One
promising approach is that of Wasserman, who claims that using statistical
evidence demeans the litigant’s individuality and autonomy. This article
presents Wasserman’s argument, explores its merits, and defends it from some
objections. However, the article also identifies six significant weaknesses,
which have to be overcome before Wassermann’s account can successfully
identify the circumstances in which statistical evidence should be used or
restricted.
Keywords Statistical evidence; Individuality; Generalisations; Inferences; Fact-
finding; Gatecrasher Paradox; Predictions; Legal persons; Previous convictions
‘Some have said they have discovered the truth,
some have asserted that it cannot be apprehended,
and others are still investigating.’
Sextus Empiricus, Outlines of Scepticism, 2nd edn
(Cambridge University Press: New York, 2000) I 1.
doi:1350/ijep.2008.12.4.304
THE INTERNATIONAL JOURNAL OF EVIDENCE & PROOF (2008) 12 E&P 303–324 303
* Email: ap550@cam.ac.uk. I am grateful to Adrian Zuckerman, John Gardner, Victor Tadros, Rhonda
Powell, Tal Ofek and Guy Sela, and the anonymous referees for their helpful comments. I am
particularly in debt to the editor, Paul Roberts, for his guidance.
The probative value of statistical evidence holds seemingly endless fasci-
nation for Evidence scholars. One aspect of this topic, known as the
reference class problem, was the subject of a recent special issue of this
journal.1This article is intended to make a further contribution to the ongoing
debates surrounding statistical evidence.
Consider the following variation on the classic ‘gate-crasher paradox’.2Humpty,
an enthusiastic football fan, is sued for gatecrashing a local football match. He was
seen by Alice sneaking through the fence. Alice’s visual identification ability is
tested and shown to be accurate 9 times out of 10. This case seems straightforward
and finding Humpty liable seems intuitively right.3Hatter, another enthusiastic
football fan, is also sued for gatecrashing a different match. This time, the only
available evidence is that 1,000 spectators attended the match but only 100 tickets
were purchased. Nine out of 10 spectators, in other words, must have gatecrashed.
Surprisingly, this case is relatively straightforward too: finding Hatter liable based
on this statistical evidence seems intuitively wrong. Yet, at least prima facie, the risk
of error is similar to Humpty’s case where imposing liability was intuitively right.
In both cases, there is a 1 in 10 chance that finding liability would be mistaken.
A common intuitive response to these scenarios is that statistical evidence does
not prove anything about a particular individual. Whilst this explanation may have
some intuitive appeal, it begs the question as to why statistical evidence fails to
constitute proof of liability. In an article published in 1992, David Wasserman
suggested that statistical evidence fails to respect the individuality and autonomy
of the litigant. According to Wasserman, using statistical evidence regards the
individual as a predetermined mechanism rather than an autonomous individual
capable of determining his own conduct.4This article critically reconsiders
Wasserman’s argument from autonomy—hereafter designated ‘the individuality
account’—as a justification for restricting the use of statistical evidence in some
304 THE INTERNATIONAL JOURNAL OF EVIDENCE & PROOF
STATISTICAL EVIDENCE AND INDIVIDUAL LITIGANTS
T
1 R. Allen and P. Roberts (eds), Special Issue on the Reference Class Problem (2007) 11 E & P 243–317.
2 See L. J. Cohen, The Probable and the Provable (Clarendon Press: Oxford 1977) 74. The type of legal
proceedings (criminal or civil) and the precise heading of the claim are intentionally left vague in
my variation.
3 It could be argued that, if the case is in the criminal context, then the fact that Alice is mistaken in
one-tenth of her identifications is sufficient to establish a reasonable doubt. Even if this is true,
this objection may be bypassed by adjusting the example to accommodate one’s desired threshold
of reliability (e.g. Alice is mistaken only once in 100 or 1,000 times).
4 D. T. Wasserman, ‘The Morality of Statistical Proof and the Risk of Mistaken Liability’ (1992) 13
Cardozo Law Review 935. An earlier account which also puts individuality in the centre is
Zuckerman’s account of individualised justice, see A. A. S. Zuckerman, ‘Law, Fact or Justice?’ (1986)
66 Boston University Law Review 487. Although Zuckerman’s account includes important procedural
dimensions which are missing from Wasserman’s account, Wasserman’s treatment of individ-
uality is more developed and thus this article focuses only on Wasserman’s account.
To continue reading
Request your trial