How do you address confounding variables in observational studies where randomized controlled trials are not feasible?
