This paper, authored by George L. Priest and Benjamin Klein, explores the relationship between disputes that are litigated and those that are settled, both before and during litigation. The authors argue that the selection of disputes for litigation is influenced by economic factors, including the expected costs of favorable or adverse decisions, the information available to parties about the likelihood of success, and the direct costs of litigation and settlement. They propose a model where potential litigants form rational estimates of the likely trial outcomes based on all available information, including legal precedents and judicial or jury bias. The model suggests that the disputes selected for litigation do not form a random or representative sample of all disputes, but are biased towards disputes with outcomes close to the decision standard. Specifically, the model predicts that the rate of success for plaintiffs at trial or appellants at appeal will tend to be 50%, regardless of the substantive legal standard or the attitudes of judges or juries. The authors also discuss how the relative stakes of the dispute for each party can influence the rate of success, and present simulations to illustrate the model's implications. The paper concludes by refining the theory and discussing its implications for empirical studies of dispute resolution.This paper, authored by George L. Priest and Benjamin Klein, explores the relationship between disputes that are litigated and those that are settled, both before and during litigation. The authors argue that the selection of disputes for litigation is influenced by economic factors, including the expected costs of favorable or adverse decisions, the information available to parties about the likelihood of success, and the direct costs of litigation and settlement. They propose a model where potential litigants form rational estimates of the likely trial outcomes based on all available information, including legal precedents and judicial or jury bias. The model suggests that the disputes selected for litigation do not form a random or representative sample of all disputes, but are biased towards disputes with outcomes close to the decision standard. Specifically, the model predicts that the rate of success for plaintiffs at trial or appellants at appeal will tend to be 50%, regardless of the substantive legal standard or the attitudes of judges or juries. The authors also discuss how the relative stakes of the dispute for each party can influence the rate of success, and present simulations to illustrate the model's implications. The paper concludes by refining the theory and discussing its implications for empirical studies of dispute resolution.