Koehler, J. J. & Shaviro, D. N. (1990). Veridical Verdicts: Increasing Verdict Accuracy Through the Use of Overtly Probabilistic Evidence and Methods. Cornell Law Review. 75, 247-279.

ABSTRACT

This article has examined the capacity of overtly probabilistic evidence and methodologies to improve verdict accuracy. Policy and other practical reasons for avoiding an overtly probabilistic trial process have been downplayed in order to explore the idea that base rates and other probabilistic evidence are rich sources of information that, at present, are not always understood in legal circles.

Base rate evidence, while neither a panacea nor devoid of problems, is relevant to the truth of asserted facts, and indeed is no less relevant in principle than case-specific evidence. An 80 percent probability of guilt based entirely on base rate information carries with it the same 20 percent chance of a false conviction as an 80 percent probability of guilt based on, say, the somewhat unreliable testimony of an eyewitness. All evidence contains a risk of error if relied upon to support a conclusion; overtly probabilistic evidence simply makes the risk more visible.

A wide range of base rates are potentially relevant to a question of fact. Some base rates are more informative or reliable than others because they are based on more appropriate reference classes. However, from a verdict accuracy standpoint, a base rate need not meet any minimum standard of specificity in order to have evidentiary weight. There usually is no such thing as thecorrect base rate for a particular issue. Even base rates that are relatively unspecific, or that are based on relatively small samples, may provide the best estimate for a given frequency in the absence of other information.

Overtly probabilistic methodologies that employ rules of probability theory (such as the product rule or Bayes' theorem) also can improve verdict accuracy. In combining multiple items of evidence, intuitive strategies are often inconsistent with and inferior to mathematical techniques. Even with regard to subjective probabilities, probability theory is likely to improve verdict accuracy when there is a positive correlation between jury probability estimates and actual states of affairs.

To be sure, there are situations where overtly probabilistic methodologies may bring little benefit. Sometimes fact-finders may have difficulty using them correctly due to complexity and translation difficulties. The use of base rates may also cause feedback problems.

However, there appears to be little evidence that base rate information will "dwarf the soft variables," leading jurors to pay too little attention to case-specific evidence. Psychological research suggests instead that base rate evidence is more likely be underweighted. Yet whether jurors pay too little attention to the "hard" evidence or the "soft," the difficulties that they experience in combining the two provide a strong reason for acquainting them with mathematical rules that treat the two as equivalent.

In closing, we wish to reiterate that we do not claim that overtly probabilistic evidence and methods should be incorporated into trials. Nor do we claim that policy concerns other than verdict accuracy are unimportant. However, we do claim that the use of an overtly probabilistic approach is neither logically nor theoretically inappropriate.

Commentators who oppose such an approach even when it is feasible are urging a sacrifice of verdict accuracy in order to advance other policy objectives. There is nothing wrong in principle with such a tradeoff, and it may be appropriate in many instances. But this tradeoff requires explicit acknowledgement and justification based on a thorough understanding of the benefits of overtly probabilistic evidence and methods.