Abstract
Teaching is anecdotally considered a good way to improve one’s own knowledge, but field evidence on learning by teaching is sparse. In this study, I partner with 22 middle and high schools in the midwestern United States to test the impact of creating math tutoring videos on students’ math skills. Stratified by teacher, 130 math classrooms are randomly assigned to either 1) Control, 2) Watchers, or 3) Creators conditions. In the “Watchers” classrooms, teachers assign a weekly PSAT math problem and students are given a link to a “help” video to watch. Students in the “Creators” classrooms are assigned the same problem, but are asked to create a video explaining the solution. The primary outcomes are students’ math class grades, as well as their score on a 15-question test of PSAT/ACT questions that cover material related to the treatment tasks. I additionally track how often the “help” video link is clicked in each class period as a mediator variable, as well as the compliance and accuracy of the weekly tasks for both Creators and Watchers.