Accuracy and Preferences for Legal Error


We study the interactions between accuracy and standards used in the determination of legal liability. First, we show that accuracy and type-1 errors (wrongful findings of liability) must reduce each other's effectiveness in mitigating optimal type-2 errors (wrongful failures to assign liability) for previous results in the literature to hold. When this condition holds, for major crimes the median voter's tolerance for type-1 errors is reduced as the legal system's accuracy increases. However, this relationship need not hold for minor offenses. Our analysis also reveals that legal processes that emerge under electoral pressures convict more often than is optimal but less often than necessary to maximize deterrence. Moreover, when the median voter's preferences are implemented, an increase in accuracy can counter-intuitively reduce welfare.