Abstract:
|
School achievement scores are used to assess school effectiveness. However, using the mean achievement of a school as a proxy for its effectiveness is inherently unfair as it is a function of both the school's effectiveness and the demographic composition of its clientele. Empirical Bayes (EB) residual estimates of school-level random effects associated with different parameters in a multilevel model provide a method to identify schools that appear to exceed expectations in a variety of domains, including achievement, growth, and achievement gaps. We demonstrate the use of school level EB residual estimates as potential indicators of school effectiveness. We also discuss multivariate techniques for combining multiple sets of EB residuals. Such multivariate arrays of EB residuals could include multiple EB residuals within a given school year, the same EB residual, computed across multiple school years, or a combination of multiple EB residuals, computed across multiple school years. We discuss multivariate school selection strategies. Finally, we review strengths and limitations of the approach and suggest fertile avenues for future research.
|