Association for Public Policy Analysis & Management

We consider the differences in two and four-year college attainment between students who used a voucher to attend private high school and students who attended traditional public schools. Our analysis is based on the population of voucher students in Milwaukee who began high school in September 2006 and remained in private school for the next four years, matched to a sample of students who were in public high schools during the same time. The match occurred in two stages: the first linking students within Census tracts on the basis of 9th grade test scores and student demographics, and second matching the groups by other characteristics--most notably parental education, which we argue proxies for a “pre-treatment” outcome. We adjust voucher-public differences using three related estimators, and discuss the extent to which unobserved sources of selection bias must be present to invalidate our estimates. We find that students who attended private high schools enrolled in four-year college at higher rates comparable to several recent (quasi-experimental and experimental) similar studies and eventually attained as much as 7/10 of an additional year of four-year college. Few differences are observed for two-year enrollment, and voucher students were no more likely to graduate college on time. Importantly, we find no evidence that the most advantaged students were those who realized the strongest voucher-related gains: students whose parents were high school drop outs and those whose 9th grade test scores were in the lowest quartile realized comparable returns to voucher use.