A Simple Example of Isolation in Building a Natural Experiment
What is isolation?
Large data systems increasingly record outcomes and exposures to treatments, but often the occurrence of an exposure/outcome event, the timing of the event, physical location of the event, and attending circumstances are biased in ways that cannot be controlled using data recorded by that system. In this example of isolation (an especially simple example), the data are from the U.S. Fatality Analysis Reporting System (FARS), which records information about motor vehicle accidents with at least one fatality. The system records interesting information about such accidents, including safety equipment, safety belt use, airbag deployment, and much more.
Some content is only viewable by ASA Members. Please login or become an ASA member to gain access.
