A Primer on Value-Added Models: Towards a Better Understanding of the Quantitative Analysis of Student Achievement

Abstract

Value-added models (VAMs) have received considerable attention as a tool to transform our public education system. However, as VAMs are studied by researchers from a broad range of academic disciplines who remain divided over the best methods in analyzing the models and stakeholders without the extensive statistical background have been excluded from the technical debate, the implementation of the models have been undermined. This study conducts a comprehensive investigation of VAMs to find consensus and transparency in the understanding of the models. Unlike the majority of existing value-added research which conducts highly advanced analyses, this study takes a unique pedagogical approach. It targets a wider range of audience particularly those without an extensive background in statistics - policy makers, district officials, school principals, teachers, and parents. The study first clarifies the technical dimensions underlying the design of the models. It then applies this conceptual understanding to conduct a thorough and hands-on value-added analysis using the Washington State longitudinal data. A comprehensive diagnosis, revision, and validation of the estimated models will be conducted to ensure the most accurate, reliable, and robust estimates. The study concludes by illustrating how the conventional policy analysis which focuses only on the value-added estimates can be extended to incorporate student background equity related factors. This extended analysis enables us to identify teachers (or schools) who achieve quality and equity outcomes simultaneously by raising the performance particularly of the students who have historically underachieved. By providing a common ground for all stakeholders irrespective of their background, this study provides an essential platform for further development of VAMs and its application as a tool to improve our education system.