The Effect of Grading in School Accountability Systems: An Investigation Using Propensity Scores In Second-order Growth Models

The development and implementation of school-level accountability systems has been mandated by recent federal law. However, there is a dearth of research into the construct validity of such measurement systems. This project adopts a latent factor perspective to assess the validity of a unidimensional definition of School Quality and estimate the impact of implementing a school accountability system using A-F grades for one application: The New York City Progress Report. A novel combination of propensity score matching and second-order latent growth modeling with adjusted error estimates is used. Results show receipt of a failing grade increases School Quality in the second year by 0.167 standard units compared with similar schools. The unidimensional definition of School Quality exhibits extremely poor model fit however, but model-based grades exhibit better consistency with other external measures of schools compared with the original formulation