Exploration of Different School Value-Added Models in a Highly Heterogeneous Educational Context

• In Event: 75.014 - Making Sense of Accountability Policies: Value-Added Models and Teacher Evaluation

Abstract

Value-added (VA) modeling aims to quantify the effect of pedagogical actions on students’ achievement, independent of students’ backgrounds. To date, there is no consensus concerning the use of statistical models and covariates when calculating VA scores. The present study aims to systematically analyze various school VA models (including different combinations of covariates). Regarding the increasing diversity in student populations in many countries, the exploration of VA models is especially relevant in heterogeneous educational contexts. This is possible through the use of a longitudinal sample from standardized achievement tests of the Luxembourg school monitoring programme. Notably, we did not only analyze the correlations and explained variances of different VA models, but also accuracy and fairness at school level.

Authors

• Jessica Levy, University of Luxembourg

• Martin Brunner, University of Potsdam

• Ulrich Keller, University of Luxembourg

• Antoine Fischbach, University of Luxembourg