The major theory in dispute is the selection effect and the 50 percent hypotheses. These hypotheses are inter-related and hold that cases that can only be resolved through trial are a selection of cases where the prevailing law does not clearly favor either the plaintiff or the defendant on the particular facts of the case. The 50 percent hypothesis limits this proposition by holding that generally in these situations the defendant wins 50 percent of the time and the plaintiff wins 50 percent of the time, but deviations from the 50/50 split may be explained by factors such as differential stakes to the parties or the law that contains some elements of strict liability that would naturally limit defendant's chances of winning. Literature is cited that challenges this hypothesis and also questions the methodology used to empirically test the hypotheses. Theoretical refinement of tests of the 50 percent and selection hypotheses is conducted and a reanalysis of prior studies is conducted such that the 50 percent hypothesis is rejected, but the selection hypothesis can be retained. The author asserts that the 50 percent hypothesis may apply adequately to tort litigation, but cannot be inferred to apply to all civil litigation. The selection hypothesis is retained because it is clear that some subset of all disputes are selected for litigation based on some perception of success rate, but different success rates are associated with different types of cases. For example, medical malpractice cases have a predicted success rate lower than that of rear-end collision cases. Methods for predicting success rates are provided and applied to different types of cases and different levels of data. 51 footnotes and 1 appendix.