In statistics, econometrics, epidemiology and related disciplines, the method of instrumental variables is used to estimate causal relationships when controlled experiments are not feasible or when a treatment is not successfully delivered to every unit in a randomized experiment.