The Impact of Charter Schools on Student Achievement

Abstract

Charter schools are public schools that are funded by a per-person fee for each student they attract. They are freed from some regulations on school management but obey safety regulations, follow non-discrimination rules, participate in state testing, and choose students by lottery when they are oversubscribed. We investigate the impact of a group of schools that enroll most of Chicago’s and many of Illinois’charter school students. The schools have more applicants than places. We use the lotteriedout students as a control group for the lotteried-in students because randomization makes the groups similar on unobservable characteristics, such as motivation, as well as on observable traits, such as race and prior achievement. We estimate both the e¤ect of attending charter schools (the treatment-on-the-treated e¤ect) and the e¤ect of being o¤ered the chance to attend a charter school (the intent-to-treat e¤ect). We show that, compared to their lotteried-out fellow applicants, students who apply to and attend charter schools starting in the elementary grades score about six national percentile rank points higher in both math and reading. These e¤ects are for students who have spent an average of two years in charter school. In the Chicago schools we investigate, students who apply in later grades are too rare for the random lotteries to generate balanced groups of charter school and regular public school students. We therefore do not have meaningful estimates for them. This study includes discussions of methods for dealing with lotteries, pre-lottery information, re-application to charter schools, siblings, attrition, and grade-of-entry e¤ects. We also compare the results from the lottery method with results from value-added analysis and demonstrate the latter results are nonrepresentative and problematic.