Data Watch: Tort-uring the Data
梳理了美国侵权法研究可用的数据来源，指出多数数据并非为研究目的收集，需拼凑多个不完整来源，并警示从审判数据推断总体时必须谨慎。
number of incidents that give rise to torts (auto accidents excepted). We know something about cases that are filed, but very little about the cases that are dropped and very little about cases that are settled. We have reasonable information on trials, but trials are not a random sample of cases and so inferences from trial data must be drawn with caution. Most of the data on torts has not been collected for research purposes and so analysis often requires cobbling together information from several incomplete sources. This article will discuss data available for researchers interested in the U.S. civil justice system and will illustrate the uses of the various datasets with some interesting findings. Our focus is on torts, defined as an injury to person or property that is not covered by contract and for which civil liability may be imposed. The most common tort is the result of an auto accident. We discuss data useful for analyzing trends, data that are useful for cross-sectional research and finally data covering only a specific type of civil litigation such as medical malpractice. Table We conclude by discussing the limitations of all civil litigation data.