Priest and Klein (1984) argued that, because of selection effects, the percentage of litigated cases won by plaintiffs will not vary with the legal standard. Many researchers thereafter concluded that one could not make valid inferences about the character of the law from the percentage of cases plaintiffs won, nor could one measure legal change by observing changes in that percentage. This article argues that, even taking selection effects into account, one may be able to make valid inferences from the percentage of plaintiff trial victories. First, it analyzes selection effects under asymmetric information models. It shows that, under a wide array of plausible assumptions, the standard screening and settlement models predict that a sufficiently more pro-plaintiff legal standard will lead to a larger percentage of plaintiff trial victories. Second, it reexamines Priest and Klein’s own model. The prediction that the plaintiff trial win rate will not vary with the legal standard is a limiting result which holds only as the variance of the parties’ prediction errors goes to zero. Nevertheless, this limiting result is not necessarily relevant to empirical work, because, as the variance of the parties’ prediction errors goes to zero, the number of litigated cases also goes to zero. Thus, whenever one is doing empirical work on litigated cases, one is necessarily dealing with a situation in which prediction errors are positive. Under a wide range of plausible parameters and distributions of disputes, plaintiff trial win rates increase as the law becomes sufficiently more pro-plaintiff, and this effect will be large enough to be detected in samples of reasonable size. Third, it examines the effect of an increase or decrease in damages. It concludes that unlike a shift in the standard of liability, changes relating to damages produce ambiguous, yet testable, predictions regarding the plaintiff trial win rate
