ABSTRACT
Policymakers wish to have evidence on the effects of charter schools on their students' achievement. This chapter considers the three main forms of analysis used to evaluate the achievement of students in charter schools: comparison with controls based on observable variables, value-added analysis, and lottery-based comparisons. Comparison with controls based on observable variables comprises a large set of methods that can be very good at eliminating forms of selection based on observable variables. The methods under this heading include everything from fully nonparametric matching to linear multiple regressions. The chapter presents discussion on methods for estimating the average achievement effect of charter schools. It discusses most of the evaluation problems and solutions that are known, even if they have not been well described in the context of charter schools. The chapter draws upon more than 20 years of methodological work that developed around the question of whether training programs positively affected youths' outcomes—the "program evaluation problem."
