Association for Public Policy Analysis & Management

A growing body of literature has established that teachers make large gains in productivity during their initial years on the job, and that these gains can persist well into the mid-career if not beyond. These studies focus primarily on teacher productivity improvement as measured by value-added to students’ achievement on standardized tests. However, such estimates are available for only a small fraction of teachers – those in tested grades and subjects. Test scores also provide little information about the mechanisms behind these observed improvements and may not capture the full extent to which teachers become more productive over the course of their careers.

In this paper, we explore the multidimensional nature of teacher skill development using a nine-year panel dataset of principals’ subjective evaluations of teachers from Charlotte-Mecklenburg Schools. Principals rated teachers on eight performance areas providing a detailed window into teachers’ skill differences across a range of pedagogical and professional domains. Performance ratings are available for teachers across all grades and subjects providing both greater generalizability and the ability to test for previously unexplored differences across teacher characteristics and teaching assignments. They also capture teachers’ contributions to their school outside the classroom.

We find that these measures capture meaningful variation in teacher practice. Prior research suggests that principals can differentiate instructional practice only among the very lowest and highest performing teachers as measured by value-added to test scores. In contrast, we find that teachers’ average evaluation ratings are associated with value-added across the full performance range. We also find that performance rating gains predict teachers’ value-added gains. Finally, we find that novice teachers’ skills differ meaningfully across the eight domains of professional practice rated by principals. Novice teachers are most likely to struggle with managing student behavior and instructional delivery. Descriptive evidence suggests that, on average, novice teachers improve at similar rates across different domains of practices rather than becoming equally effective across all skill domains as they gain experience.

Using a within-teacher returns to experience modeling framework, we find that new teachers make large and rapid improvements in their instructional practices throughout their first ten years on the job. This positive marginally decreasing pattern of instructional improvement over time mirrors that of improvement based on test scores. We find that teachers’ performance improves by over 3/4^th of a standard deviation, on average, across their first decade on the job. These findings are consistent across a range of robustness tests examining sample selection bias and alternative constructions of evaluation scores. Finally, we show that these average estimates mask substantial variation in the improvement patterns of individual teachers as measured by evaluation ratings. These findings add further evidence that teacher effectiveness is dynamic across skills and over time, and that individual teachers differ substantially in their performance improvement profiles.