Association for Public Policy Analysis & Management

Teachers have profound and long-lasting effects on students both through developing cognitive skills measured by test scores, and by developing non-cognitive skills measured through behaviors like attendance and discipline. Policy makers and school administrators have long sought the best ways to maximize student achievement through assigning teachers to the classrooms and students where they can be the most effective.

We examine an increasingly common, but poorly understood, aspect of classroom assignment policy: having a teacher for the second time in a different grade. We use precisely matched student-teacher data for Tennessee to document the extent and type of repeat student-teacher interactions and to estimate the causal effect of repeated teachers on students’ cognitive outcomes: test scores in reading and math, and non-cognitive behavioral outcomes: attendance, truancy, and discipline. Repeat student-teacher matches are both rare and common: 5% of student-teacher interactions are repeats but at least 44% of students will have a repeat teacher between 4^th and 12^th grade.

To estimate the causal effect of a repeat teacher we follow the value added literature and use both a flexible lagged outcome framework and student fixed effect framework. Because student assignment into classrooms is far from random, we also control extensively for sorting across districts, schools, classrooms, tracks, and courses. Our cognitive outcomes are TCAP test scores for elementary and middle school students and the End of Course exams for high school students. We use attendance, truancy, and suspensions as behavioral proxies for non-cognitive outcomes.

We find that having a repeat teacher improves test scores across all grades and subjects. High school scores improve by .02 to .03 standard deviations, roughly equivalent to an extra month of school. Elementary and middle school scores improve by .008 to .015 standard deviations or 4-11 days. We find significant effects on non-cognitive behaviors in high school including a decrease in full day absences of .12, driven entirely by a decrease in truancy, and a reduction in suspensions across all grades. These results are robust to including a large set of fixed effects to control for student sorting among teachers and schools.

We also find substantial heterogeneity in the effect of a repeat teacher depending on her measured value added, experience, grade level, and experience within the grade and subject. Repeating with an inexperienced teacher decreases test scores in high school by .02 standard deviations. However, students of lower performing teachers and teachers inexperienced in a test have the most to gain from a repeat match. Repeating with one of these teachers improves performance by relatively more than repeating with a higher performing teacher or one experienced in the specific grade and subject. We speculate that struggling teachers who already know their students have more time to improve other aspects of their teaching. Finally, we find that having a repeat teacher in grades 3-8 in one subject also improves scores in other subjects but there is no such spillover in high school. We believe that our results are further evidence of the importance of student-teacher relationships.