By Milo Schield (New College of Florida and U. New Mexico)
Information
Students see disparities used as evidence for discrimination. Confounder-based statistical literacy (critical thinking about everyday statistics) addresses these arguments involving causal inference. This course is designed for students in non-quantitative majors; it has less than a 30% overlap with the traditional introductory population-inference course. Just as correlation does not imply (is not sufficient for) causation, statistical literacy argues that disparity does not imply discrimination. In this class, students use weighted averages to work multivariate problems that control for the influence of a measured confounder.
They learn how standardizing (giving both groups the same mixture) can take into account the influence of a confounder. Students see how sex and race gaps in average incomes can be influenced what is taken into account. Students see how statistically significant differences can become statistically-insignificant and vice versa.
• Students see value in this course. See www.statlit.org/pdf/2022-Schield-ASA.pdf for UNM student evaluations. Of those in my fall classes (77 students), 51% agreed that Statistical Literacy should be required for ALL students for graduation (39% were neutral).
• Confounder-based statistical literacy has been taught at the U. of New Mexico (35 students/class; 5 classes/semester) for four years. Starting fall 2025, the department is requiring statistical literacy for all students majoring in statistics.
• Based on this student feedback, a review of the content by statistical educators, and an in-class review by the Provost on the teaching of a controversial topic, New College of Florida is requiring Statistical Literacy for all incoming students (25/class) starting fall 2025.