Causal Inference Using Observational Data
Aboozar Hadavand · 2025
Abstract Chapter 5 addresses the complexities and methodologies of establishing causal relationships using observational data, where researchers cannot control treatment assignments. It emphasizes the challenge of confounding bias, which arises when the relationship between treatment and outcome is due to noncausal channels. The chapter highlights the importance of identifying and controlling for confounding variables to ensure unbiased causal inference. Additionally, the chapter discusses the limitations and potential biases inherent in observational studies, including unobserved confounding and collider bias.