Causal Analysis of Observational Data

This lecture focuses on the causal analysis of observational data, highlighting the difference between experimental and observational studies. It covers the pitfalls of observational studies and provides tools to draw valid conclusions from 'found data'. The instructor motivates students to read key books in the field and discusses the challenges of conducting randomized controlled experiments. Various methods for addressing confounding variables in observational studies, such as matching and propensity scores, are explained. The lecture emphasizes the importance of achieving balance in study groups and introduces sensitivity analysis to assess the impact of unobserved covariates. Examples related to smoking and lung cancer illustrate the concepts discussed, concluding with a summary of key takeaways for project work.