A variable more likely to be present in one group of subjects than another that is related to the outcome of interest and thus potentially confuses, or "confounds," the results.