Validity of Value-added Estimation

Résumé du livre

Despite the growing use of "value-added" estimation to capture the contribution of a specific school or teacher to student achievement, there is still much debate about what exactly value-added estimates capture, and if these estimates are reliable enough to use for policy-purposes. As a whole, my dissertation examines the implications of choices we make as researchers about value-added modeling, especially for use in education accountability policies. Rarely do the details of statistical modeling play such a direct role in shaping policy. Producing valid value-added estimates is crucial to both the perceived fairness and ultimate success of any policy that relies on these measures. Here, a blunt use of statistics can undermine the efforts of struggling schools or teachers and hinder our ability as researchers to learn from places that do actually stimulate large growth in student learning. Thus, it is the crucial role of researchers in this field to both develop and be critical of the methods for obtaining valid value added estimates. In my dissertation, I explore three facets of value-added modeling, which together generate a deeper knowledge of how such measures should be created and used.