Comparing Models for School Performance Across Four States

• In Event: Roundtable Session 110
  In Roundtable Session: Approaches in School Accountability

Abstract

This study further considers which model might be most appropriate to monitor school performance. The underlying question is whether there exists an optimal model that can consistently identify schools as performing well, or performing poorly. Specifically, there is increased interest using growth or value added models for school accountability. This study presents results from a comparison of several models using data from four states. Specifically, this study estimates school effects based on several different models, including gain, layered, student growth percentile, fixed effects, and growth models. Results indicate that models behave differently from one another and differently across states, and that it is unlikely that there is a single optimal model.

Author

• Pete G. Goldschmidt, California State University - Northridge