|
Abstract:
|
Accuracy of jury verdicts is a societal concern, always paramount to the defendant, that becomes acute during high-profile litigation. But aside from exceptional cases, little is known about how often juries as a whole get it wrong. Yet, average accuracy of jury verdicts can be studied empirically and systematically when a replication consisting of a second rating of the verdict is available. Data from a recent study of U.S. criminal cases by the National Center for State Courts (NCSC) showed only fair to poor judge-jury agreement. Simple estimates developed from the judge-jury agreement rate provide (under plausible conditions) optimistic estimates of accuracy of jury verdicts. Under stronger conditions, log-linear models provide surprising estimates of conditional and unconditional probabilities of incorrect convictions and acquittals. Further data collection is recommended.
|
