Statistical inference for causal effects

Research questions motivating many scientific studies are causal in nature. Causal questions arise in medicine (e.g., how effective is a drug treatment?), economics (e.g., what are the effects of job training programs?), sociology (e.g., is there discrimination in labor markets?), customer satisfaction (e.g., what are the effects of different ways of providing a service?), and many other fields. Causal inference is used to measure effects from experimental and observational data. Here, we provide an overview of the approach to the estimation of such causal effects based on the concept of potential outcomes, which stems from the work on randomized experiments by Fisher and Neyman in the 1920s and was then extended by Rubin in the 1970s to non-randomized studies and different modes of inference. Attractions of this approach include the allowance for general heterogeneous effects and the ability to clarify the assumptions underlying the estimation of causal effects in complex situations.