Natural experiments: legal, moral, and possible

About

The preferred method for obtaining knowledge in the laboratory sciences is by the manipulative controlled experiment. For example, in order to establish the effect of a certain chemical, a scientist adds that chemical to one of two test tubes containing identical solutions, maintaining the other test tube as an unmanipulated control.

Unfortunately, for most important questions of the social sciences and field biology, decisive manipulations of interest would be illegal, immoral, or impossible. Instead, social scientists and field biologists often resort to so-called "natural experiments": i.e., comparing situations differing naturally with respect to the independent variable of interest. Examples include: comparing economic growth rates in countries and time periods when the political leader either did or did not die of a heart attack while in office; or comparing win/loss records of football teams following losing seasons after which the coach was or was not fired. Of course, natural experiments pose obvious difficulties, which have been addressed with increasing sophistication in the last two decades. I shall discuss recent natural experiments in the fields of history, business, sports, and religion.

Jared Diamond, Professor of Geography at the University of California (Los Angeles), is the Pulitzer-Prize-winning author of Guns Germs and Steel, Collapse, and other internationally best-selling books.