Lecture
This lecture discusses the conflation of statistical dependence with confounding, highlighting methods that fail to control for general forms of confounding. It covers topics such as nonsense associations, confounding due to dependence, and general confounding. The lecture also explores causal inference with multiple treatments and unmeasured confounding, emphasizing the distinction between existing methods and new, improved approaches. Additionally, it delves into confounding by dependence, network simulations, and the implications of confounding in various fields. The instructor presents recent methods to control for unmeasured confounding, the limitations of orthogonalization methods, and the development of novel semiparametric causal inference methods.