Conceptual and empirical differences among various value-added models for accountability

Accountability systems in education generally include indicators of student performance. However, these indicators often differ considerably among the various systems. More and more countries try to include value-added measures, mainly because they do not want to hold schools accountable for differences in their initial intake of students. This study presents a conceptual framework of these value-added measures, resulting in an overview of 5 different types. Using data from Dutch secondary schools, we empirically provide estimates of these different measures. Our analyses show that the correlation between the different types of school effects estimated is rather high, but that the different models implicate different results for individual schools. Based on theoretical considerations, arguments are given to use the following indicators in the value-added accountability models: prior achievement, student-level background characteristics, and compositional characteristics of the student population.