A recent post on the Stata list involving law-related data (here, patent litigation) struggles with selection-effect questions. Those familiar with law- and litigation-related data understand why selection issues are almost always present in litigation data. (And those unfamiliar with law too often do not.)
Comments