IntroductionA large literature in the economic, educational and sociological fields shows that students attending private schools outperform students attending traditional public schools in a wide range of results. Since a common feature of private schools is their autonomy from school district offices, these findings strengthen initiatives to improve public school student outcomes through the creation of public charter schools in the United States, free schools in the United Kingdom, independent public schools in Australia and community-run schools in many developing countries. It is often unclear, however, whether private, independent public schools causally improve student outcomes and, even if they do, what characteristics drive those improvements. Studies that attempt to identify the effects of private or charter education and disentangle the various causal mechanisms face several problems. challenges. First, it is difficult to identify the causal effects of independent public or private education on student outcomes based on most available observational data because unobserved selection biases are pervasive and difficult to address (Altonji et al. 2005a) . Although recent natural experimental evidence based on the random assignment of private school vouchers or oversubscribed school places to low-income applicants shows significant positive effects of these schools on student outcomes, it is still difficult to learn precisely which aspects of these schools explain the differences in results. When these studies compare outcomes between randomly selected recipients (treatment group) and nonrecipients (control group) of private school vouchers or charter school slots, the estimated effects of private… middle of the paper. ... .. holds an advanced teaching certificate and employs 8.5% of its teachers on a short-term contract. Over 96% of schools track their students' abilities in English and math classes, but less than 15% of schools track their students' abilities in Korean, science, and social studies classes. On average, 4% of high school students came from families receiving government social support. 12% of students receive free lunch support from the government. Ethnic minority students are rare; 0.1% of the student population. Finally, Table 1 summarizes student-level data for those who took the NAEA test in 2010. Of the 88,406 students scheduled to take the NAEA tests in 2010, approximately 3 percent missed some tests. We normalize NAEA test scores, which range from 100 to 300, so that they have zero mean and standard deviation 1.