Abstract

This entry provides an introduction to causal inference. Causal inference refers to the estimation of the effect of a treatment, policy, or intervention on an outcome of interest. Causal inference is, therefore, at the centre of science and social sciences. This entry emphasises Rubin’s potential outcomes framework. Within this framework randomised experiments, where consistent estimates obtainable under the assumption of unconfounded assignment are introduced. The latter assumption is rarely met in observational studies. Therefore, this entry discusses estimation of local average treatment effects with confounded treatment in observational studies. Finally, the entry provides an overview of partial identification as a technique for the estimation of causal effects. It discusses how this novel technique enables researchers to bind causal effects in a variety of settings (under confounded or unconfounded treatment) without relying on strong assumptions that might not met in practice.