Association for Public Policy Analysis & Management

Researchers are increasingly evaluating school leaders, and principals in particular, to understand how, through their myriad responsibilities, they affect student outcomes. While principals may not directly affect student achievement through instruction, they can affect students through the hiring, development, and retention of the teaching workforce at their schools. Because teachers directly interact with students, rendering them the most important within-school factor that affects students (Rivkin, Hanushek, & Kain, 2005), the shaping of teacher quality is an especially strong lever through which principals can improve students’ lives.

Evaluating principal effectiveness, however, has proven challenging (Clifford & Ross, 2012; Goldring et al., 2009). This is especially true when considering principal effects on teachers, because isolating the principal’s effect from other school factors is difficult. Specifically, teacher quality may also be influenced by factors outside the principal’s control, such as neighborhood crime rates, district policies, and local labor market conditions. Further, principals’ influence on teachers may take years to appear, and may materialize long after a principal has left a school—too late for effects to be attributed to the appropriate individual. Lastly, principal evaluation tools are subject to measurement error and may inadequately capture the scope of principals’ roles.

All of these issues have mitigated the scope of the literature around principal effectiveness, especially as it pertains to teacher quality. I expand our limited understanding of principal quality using administrative data provided by the District of Columbia Public Schools (DCPS) from 2013-14 to 2016-17. These include scores from DCPS’s principal and teacher evaluation tools, School-Leader IMPACT (SL-IMPACT) and IMPACT, respectively, as well as principal, teacher, and student demographic information, and student academic and non-academic outcomes. In addition to SL-IMPACT, I use teacher ratings of their school leaders as an alternative measure of principal quality. Using these data, I address the following research questions:

1. Does SL-IMPACT accurately capture principal effects on teacher quality—as assessed by the proportion of high-quality teachers hired, the proportion of high-quality teachers retained, and improvements in teacher quality, over time—in DCPS?
2. Does SL-IMPACT correlate with teachers’ perceptions of their principal’s effectiveness?

Preliminary results show that: after controlling for school-level covariates, changes in teacher quality vary substantially across schools in DCPS; these measures do not correlate with SL-IMPACT scores; and teachers’ perceptions of their principals do correlate with SL-IMPACT.

As more states move toward rigorous evaluation systems for school leaders, examining the ability of these systems to capture changes in school-leader effectiveness presents a relevant and important policy question. DCPS provides a unique opportunity to examine the mechanisms through which principals influence teacher quality, how high-stakes evaluation systems influence leaders, teachers, and students, the variation in principals’ ability to improve teacher quality, and consequently student outcomes in their schools, and in a field where measures are largely new and not empirically tested, an opportunity to validate an evaluation tool that is being used in practice.