Multilevel Multivariate Response Models for Studying the Consistency and Stability of Value-Added School Effects

• In Event: Poster Session 14
  In Poster Session: Advanced Issues in Multilevel Factor Analysis

Abstract

The traditional approach to estimating the consistency and stability of school effects across subject areas and time is to fit separate value-added multilevel models to each subject or cohort and to correlate the resulting empirical Bayes predictions. We show that this gives biased correlations and these biases cannot be avoided by simply correlating ‘unshruken’ versions of these predicted random effects. In contrast, we show that fitting a single multilevel multivariate response model simultaneously to all subjects or cohorts directly gives unbiased estimates of the correlations of interest. We illustrate our arguments with applications to schools in England, but our arguments apply more generally whenever multiple univariate response models are fitted in lieu of a single multivariate response model.

Author

• George Leckie, University of Bristol