2025/03/28 by Sung Tae Jang, Nicola A. Alexander · 1 voice
Decision Sciences · Social Sciences · #Educational Assessment and Improvement #Teacher Education and Leadership Studies #School Choice and Performance
paper · pdf · doi:10.1177/0013161x251322727
openalex publication_date 2025/03/28 · openalex created_date 2025/10/10 · openalex updated_date 2026/06/15
Purpose: This study investigated the links between states’ principal evaluation policies and student achievement. We focused on three levers that policymakers have used to hold principals accountable for improving student achievement: (1) including student academic growth data, (2) mandating observations of principal behavior (site visits), and (3) requiring or explicitly allowing stakeholder feedback via surveys. Data and Research Methods: We used repeated cross-sectional data for the principal evaluation policy levers collected at nine points in every odd numbered year from 2001 to 2017 by assembling a range of datasets collected by the National Center for Teacher Quality and previous studies; we also used the nationally representative National Assessment of Educational Progress (NAEP) reading and math scores of eighth-grade students. We adopted a novel framework of counterfactual estimation of matrix completion methods to examine the causal link between the state principal evaluation policies and student achievement. Findings : We found that requiring the inclusion of student academic growth data in principal evaluations had no significant impact on state eighth-grade NAEP achievement scores. However, we found that policies mandating observations or site visits and requiring or explicitly allowing stakeholder surveys were associated with small increases in state eighth-grade NAEP reading scores. Implications : Principal evaluation models can help policymakers to shape the impact principals have on student achievement, but policy design matters. The importance of focusing on principal behavior rather than student achievement is discussed. We also discuss the benefit of multiple data sources for principal evaluation models and suggest directions for future research.