Skip Navigation
Title:  Stabilizing subgroup proficiency results to improve identification of low-performing schools
Description: The Every Student Succeeds Act (ESSA) requires states to identify schools with low-performing student subgroups for Targeted Support and Improvement (TSI) or Additional Targeted Support and Improvement (ATSI). Random differences between students’ true abilities and their test scores, also called measurement error, reduce the statistical reliability of the performance measures used to identify schools for these categorizations. Measurement error introduces a risk that the identified schools are unlucky rather than truly low performing. Using data provided by the Pennsylvania Department of Education (PDE), the study team used Bayesian hierarchical modeling to improve the reliability of subgroup proficiency measures, allowing PDE to target the schools and students that most need additional support. PDE plans to incorporate stabilization as a “safe harbor” alternative in its 2022 accountability calculations. The study also shows that Bayesian stabilization produces reliable results for subgroups as small as 10 students—suggesting that states could choose to reduce minimum counts used in subgroup calculations (typically now around 20 students), promoting accountability for all subgroups without increasing random error. Findings could be relevant to states across the country, all of which face the same need to identify schools for TSI and ATSI, and the same tension between accountability and reliability, which Bayesian stabilization could help to resolve.
Online Availability:
Cover Date: February 2023
Web Release: February 27, 2023
Print Release:
Publication #: REL 2023001
General Ordering Information
Center/Program: REL
Associated Centers: NCEE
Type of Product: Descriptive Study
Questions: For questions about the content of this Descriptive Study, please contact:
Liz Eisner.