Simple and Credible Value-Added Estimation Using Centralized Assignment
Education Reform, School Assignment, December 2020
Many large urban school districts match students to schools using algorithms that incorporate an element of random assignment. We introduce two simple empirical strategies to harness this randomization for measuring the causal effects of individual schools. In applications to data from Denver and New York City, we find that our models yield highly reliable school effectiveness estimates.
Joshua Angrist, Peter Hull, Parag Pathak, Christopher Walters