“…Over the last decade, a growing number of VAMs have been developed, and different model specifications have been explored, with the dual goal of understanding and improving the quality of teacher value-added scores (Ballou, 2005;Ballou et al, 2004;Briggs & Domingue, 2011;Buddin, 2010;Harris & Sass, 2010;Harris, Sass, & Semykina, 2012;Harris, 2011;Hill, et al, 2011;Kane & Staiger, 2002a;2002b;McCaffrey et al, 2004;McCaffrey, Sass, Lockwood, Milhaly, 2009;Papay, 2011;Rothstein, 2007;Schochet & Chiang, 2010). Yet, understanding which data and model specifications have a larger effect on teacher VAEs compared to others has been challenging because the number of model specifications and their possible combinations is too large to be fully investigated in any single study.…”