D. SHAVIRO, STATISTICAL-PROBABILITY EVIDENCE AND THE APPEARANCE OF JUSTICE

D. SHAVIRO, STATISTICAL-PROBABILITY EVIDENCE AND THE APPEARANCE OF JUSTICE

103 Harv. L. Rev. 530 (1989)

Perhaps the most obvious purpose of the trial system is to reach accurate verdictsfor example, to convict criminal defendants when and only when they have committed the unlawful acts of which they are accused. Apart from any practical benefits of deciding cases accurately (such as improving deterrence), the accuracy of verdicts has moral implications. Consider a criminal trial in which the defendant, accused of rape, defends on the basis of consent, and the presence or absence of consent would be unambiguous if all the facts were known. A false conviction will punish an innocent man, perhaps severely, for a heinous crime. A false acquittal may cause the rape victim unjust humiliation, in addition to other harms.

The stakes in most trials may be lower than in a rape case, yet justice is at issue whenever one of the parties has a legal and moral right that would prevail if the true facts were known. One may favor verdict accuracy to minimize injustice even if it has no effect on future conduct. The goal of verdict accuracy implies that courts should hold in favor of whichever party appears more likely to be correct. Of course, a preference for verdict error in favor of one party may lead to tilted odds. For example, a judgment that an erroneous criminal conviction is exactly nineteen times more regrettable than an erroneous acquittal may lead to a requirement of ninety-five percent certainty of guilt before convicting.

Assuming no preference for either party in civil trials, concern about injustice would seem to dictate holding for the plaintiff in the Blue Bus hypotheticalindeed, by directed verdict if eighty percent is the only reasonable estimate of the likelihood that the defendant is liable. A verdict for the plaintiff would appear to minimize the unavoidable risk of an erroneous and thus unjust decision.

The Smith rule rejects this overtly probabilistic methodology and denies recovery despite the eighty percent certainty of responsibility. The moral intuition that underlies the rule appears to have wide appeal, and may be thought particularly compelling in criminal cases. Consider, for example, a hypothetical by Professor Charles Nesson in which twenty-four or twenty-five prisoners in an enclosed yard collaborate in a murder of a prison guard. The twenty-fifth prisoner hides in a nearby shed during the crime and is innocent. In the absence of relevant information other than the identities of the twenty-five prisoners, there is a ninety-six percent chance that a randomly selected prisoner is guilty. Reluctance to allow a jury to convict any one prisoner on the basis of this statistical evidence might arise partly from the starkly numerical nature of the evidence. Eyewitness testimony, even if the likelihood of guilt remained ninety-six percent, might not give rise to as great a reluctance.

Why should the two cases be treated any differently? Perhaps the difference lies simply in the clarity of the risk of error. In Nesson's prisoner case, the evidence on its face proclaims the existence of a four percent risk of error. By contrast, a conviction based on eyewitness testimony does not lend itself to so precise a calculation of the risk of error, and thus may make the risk less prominent and obvious. A difference merely in overtness does not seem to support a strong moral distinction between the two cases. However, the underlying intuition that they are different is stubborn enough to require more careful scrutiny....

Claim 1: "The legal system should not proclaim its willingness to convict a person or find a person civilly liable in the face of acknowledged uncertainty.''...

Claim 2: "Relying on statistical-probability evidence makes the unjust punishment of innocent persons certain rather than merely probable, and unacceptably singles out innocent victims to be sacrificed for general social benefit.''...

Claim 3: "Holding defendants guilty or liable based solely on statistical-probability evidence violates the principle that defendants should be treated as unique individuals.''...

Claim 4: "Holding a defendant guilty or liable where the evidence establishes only a bare probability (favorable 'betting odds') that the verdict ought to go against him wrongfully 'gambles' with his liberty or property. A bare probability should lead the legal system to decline to act by holding for the defendant.''...

Claim 5: "Defendants should not be held guilty or liable absent 'actual belief' by the jury, and such belief cannot be derived from statistical-probability evidence.''...

[Shaviro dismisses each of these claims and concludes as follows:] These moral claims in favor of the Smith rule all reduce to a goal of making the risk of verdict error less overt. Attempts to portray this difference as morally significant repeatedly encounter difficulty. Accordingly, the Smith rule seems to reflect squeamishness rather than coherent moralityan aversion not to dirty work itself, but to knowing too much about it....