1) Confounding occurs when a third variable is correlated with both the putative cause and outcome, making it difficult to isolate the impact of the cause on the outcome. 2) Controlling for confounding variables aims to evaluate the specific impact of a putative cause on an outcome, by keeping confounding variables constant. 3) From a structural modeling perspective, confounders can be seen as variables that influence the outcome through both direct and indirect paths, so accounting for all paths from confounders to outcomes is important.