Gendered Teacher Feedback, Students' Math Performance and Enrollment Outcomes: A Text Mining Approach

This paper studies how student gender influences the feedback given by teachers, and how this affects the student's performance in school. Using the written feedback provided to the universe of French high school students by their math teachers over a five-year period, we show that teachers use different words to assess the performance of equally able male and female students. Teachers highlight the positive behavior and encourage the efforts of their female students while, for similarly-performing males, they criticize the students for unruly behavior and praise them for their intellectual skills. To understand how this relates to the student's subsequent educational outcomes, we then match these data to records from French national examinations, as well as these students' higher education application behavior and ultimate institution of enrollment. Using the quasi-random allocation of teachers to classes, we estimate that being assigned to a teacher with feedback that is one standard deviation more gendered improves student math performance by 1.6 percent of a standard deviation on average, but does not affect students' enrollment in higher education in the following year.