Paper Summary

Direct link:

Exploring the Development of Teaching Skills

Sun, April 7, 3:40 to 5:10pm, Metro Toronto Convention Centre, 200 Level, Room 202C


The effect of individual teachers on their students’ achievement varies substantially (Kane & Staiger, 2012; Rivkin et al., 2005; Rockoff, 2004). While few identifiable characteristics explain this variation, numerous studies using value-added scores demonstrate considerable returns to teachers’ experience during their first years of teaching (Boyd, Lankford, Loeb, Rockoff, & Wyckoff, 2008; Harris & Sass, 2011; Papay & Kraft, 2015; Rockoff, 2004; Wiswall, 2013). Meanwhile, the least-experienced teachers disproportionately teach in the least-advantaged classrooms (Clotfelter, Ladd, & Vigdor, 2005 & 2006; Jackson, 2009; Kalogrides, Loeb, & Bèteille, 2012; Sass et al., 2012).

Despite abundant evidence that teachers improve their ability to drive student learning over their early careers (Aaronson, Barrow, & Sander, 2007; Chetty, Friedman, & Rockoff, 2014; Rockoff, 2004), we know little about specific skills teachers develop over this time. We seek to understand how early-career teachers improve their practice by asking: 1) what is the pattern of early-career teachers’ overall skill development; 2) how much of this return to experience can be explained by general versus context-specific skill development (e.g., in a particular school); and 3) which practices best explain this?

To answer these questions, we use detailed administrative data from the Washington, DC Public Schools, where we can observe early-career improvement in specific teaching skills for several cohorts of individual teachers, beginning with those who entered teaching in 2009-10. These data include teachers’ performance on the district’s observation rubric, the TLF, which consists of nine practices on which teachers are evaluated multiple times per year (see Table 1). While many observation systems fail to distinguish teacher quality (Kraft & Gilmour, 2017), the TLF demonstrates considerable score variation (Figure 1).

We explore within-teacher returns to experience graphically and non-parametrically, following Papay and Kraft (2015) and Wiswall (2013), primarily using scores from external evaluators, who may be less subject to evaluation bias and ceiling effects. Next, we estimate regressions with separate functions for overall and school-specific experience to understand the importance of contextual factors for teachers’ development. Finally, we decompose variance in early-career gains to understand which practices account for the greatest share of teachers’ improvement.

Preliminary results demonstrate that the average educator improves 0.91 standard deviations by her fifth year in the classroom (see Figure 2). Approximately 20% of this gain (.17 SD) can be explained by school-specific returns to experience. The average teacher improves most on checking for student understanding (Teach 5), and improves least at teaching to all the students in the classroom (Teach 3 and Teach 4).

These results provide initial evidence of teachers’ patterns and trajectories of skill attainment. These data may allow us to better understand how to produce more-highly skilled teachers from the outset, as well as how to help both pre-service and in-service teachers develop the skills necessary for their students to have the best opportunity to succeed. This is particularly of concern for novice educators, who on average perform well below their potential yet are disproportionately assigned to teach the least advantaged children.


©2019 All Academic, Inc.   |   Privacy Policy