Natural and quasi-experiments

Abstract

Natural experiments, defined as events or processes outwith the control of a researcher, which divide a population into exposed and unexposed groups, have long been used in epidemiology and public health. Evaluation methods that rely on observational rather than experimental data are classified as second best in conventional hierarchies of evidence. Natural experimental approaches have attracted renewed interest from public health researchers and decision-makers because they widen the range of interventions that can usefully be evaluated beyond those that are politically, ethically, or practically amenable to testing in randomized controlled trials. This chapter argues for seeing trials and natural experiments as part of a common toolkit for producers and users of evidence about the effectiveness of policies and programmes. It describes the most commonly used natural experimental approaches to evaluating population health interventions, and provides examples of their use in a wide range of countries and policy settings.