A specially-designed statistical literacy course is needed for college students in majors that don't require statistics or mathematics. This paper suggests that key topics in conditional probability, multivariate regression and the vulnerability of statistical significance to confounding should be included and presents some new ways to teach these ideas.