Confounding, a central issue in epidemiologic research, embodies some principles which are difficult to teach. By comparing crude and stratum-specific estimates of the risk ratio parameter on an appropriate diagram, confounding can be assessed visually, and its nature more easily understood. The pictorial approach presented here is not intended as a substitute for quantification of confounding in epidemiologic studies, but rather as a tool to facilitate understanding of the principles of confounding and its control. The relevant diagrams are presented and their use described for both the cohort and case-control study design.
