Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
This lecture by the instructor delves into deriving bounds for causal effects using sensitivity parameters on the risk difference scale. Starting with a motivating example on smoking and mortality, the lecture covers confounding bias, assumption-free bounds, and sensitivity parameters on the risk ratio scale. It explores the work of Ding and VanderWeele, the need for additional assumptions, and the proposed bounds by Sjölander and Hössjer. The lecture discusses the limitations of existing approaches, problems with sensitivity parameters, and the computation of bounds for the causal risk difference. It concludes with a summary highlighting the importance of estimating causal effects and the role of assumptions in narrowing the bounds.