A natural experiment is an empirical or observational study in which the control and experimental variables of interest are not artificially manipulated by researchers but instead are allowed to be influenced by nature or factors outside of the researchers' control. Unlike traditional randomized experiments, natural experiments are not controlled by researchers but rather observed and analyzed.
Natural Experiments Versus Observational Studies
So if natural experiments are not controlled but rather observed by researchers, what is there to distinguish them from purely observational studies? The answer is that natural experiments still follow the primary principles of experimental study. Natural experiments are most effective when they mimic as closely as possible the existence of test and control groups of controlled experiments, which is to say that there is a clearly defined exposure to some condition in a clearly defined population and the absence of that exposure in another similar population for comparison. When such groups are present, the processes behind natural experiments are said to resemble randomization even when researchers do not interfere.
Under these conditions, observed outcomes of natural experiments can feasibly be credited to the exposure meaning that there is some cause for belief in a causal relationship as opposed to simple correlation. It is this characteristic of natural experiments - the effective comparison that makes a case for the existence of a causal relationship - that distinguishes natural experiments from purely non-experimental observational studies. But that is not to say that natural experiments aren't without their critics and validation difficulties. In practice, the circumstances surrounding a natural experiment are often complex and their observations will never unequivocally prove causation. Instead, they provide an important inferential method through which researchers can gather information about a research question upon which data might otherwise not be available.
Natural Experiments in Economics
In the social sciences, particularly economics, the expensive nature and limitations of traditionally controlled experiments involving human subjects has long been recognized as a limitation for the development and progress of the field. As such, natural experiments provide a rare testing ground for economists and their colleagues. Natural experiments are used when such controlled experimentation would be too difficult, expensive, or unethical as is the case with many human experiments. Opportunities for natural experimentation are of the utmost importance to subjects like epidemiology or the study of health and disease conditions in defined populations in which experimental study would problematic, to say the least. But natural experiments are also used by researchers in the field of economics to study otherwise difficult to test subjects and are often possible when there is some change in law, policy, or practice in a defined space like a nation, jurisdiction, or even social group. Some examples of economics research questions that have been studied through natural experimentation include:
- The "return on investment" of higher education in American adults
- The effect of military service on lifetime earning
- The effect of public smoking bans on hospital admissions