Duke Law Journal

The Innocence Effect

Abstract

Nearly all felony convictions—about 95 percent—follow guilty pleas, suggesting that plea offers are very attractive to defendants compared to trials. Some scholars argue that plea bargains are too attractive and should be curtailed because they facilitate the wrongful conviction of innocents. Others contend that plea bargains only benefit innocent defendants, providing an alternative to the risk of a harsher sentence at trial. Hence, even while heatedly disputing their desirability, both camps in the debate believe that plea bargains commonly lead innocents to plead guilty. This Article shows, however, that the belief that innocents routinely plead guilty is overstated. We provide varied empirical evidence for the hitherto neglected “innocence effect,” revealing that innocents are significantly less likely to accept plea offers that appear attractive to similarly situated guilty defendants.

The Article further explores the psychological causes of the innocence effect and examines its implications for plea bargaining. Positively, we identify the striking “cost of innocence,” wherein innocents suffer harsher average sanctions than similarly situated guilty defendants. Yet our findings also show that the innocence effect directly causes an overrepresentation of the guilty among plea bargainers and an overrepresentation of the innocent among those who choose trial. In this way, the innocence effect beneficially reduces the rate of wrongful convictions—including accepted plea bargains—even when compared to a system that does not allow plea bargaining. Normatively, our analysis finds that both detractors and supporters of plea bargaining should reevaluate, if not completely reverse, their long-held positions to account for the causes and consequences of the innocence effect. The Article concludes by outlining two proposals for minimizing false convictions, better protecting the innocent, and improving the plea bargaining process altogether by accounting for the innocence effect.